A note on what’s new at Artemis. This newsletter marks the first issue under our new Artemis platform: our full digital-finance research and investment platform. Artemis now ingests the same public-company fundamentals you’d get from CapIQ or FactSet (filings, transcripts, KPIs, IR presentations) and combines them with our blockchain, stablecoin, and protocol data in one workspace. If you’re covering COIN and Aave in the same model, or modeling Circle’s float against USDC supply growth, this is the stack we built for you.

Market Overview: The Weekly Recap

Market Sentiment: cautious risk-on. Prices have continued to rebuild meaningfully off the February lows this week. The U.S.–Iran two-week ceasefire pulled geopolitical risk premium out of commodity and crypto markets, and spot BTC ETF flows turned net positive for the second week in a row. Sentiment sits in greed territory: 62 for crypto, 68 for equities (Artemis).

Assets across digital finance returned an average of 10.66%, led by AAVE, DRIFT, HOOD, COIN and CIRCLE, each of which returned above 20%. Crypto majors BTC, ETC and SOL returned about 5% each.

Today We Highlight

• Aave “Will Win” vote passes. The DAO now captures 100% of Aave-branded app-layer revenue. Arguably the most consequential DeFi governance event of the year so far.

• Artemis Research Spotlight:

• Robinhood leads digital finance stock growth as earnings approach April 28th.

1. Aave Just Rewired Itself Around $AAVE

The Aave DAO has spent the better part of four months re-litigating a deceptively small question: who owns the front end? The dispute traces to late 2025, when delegates flagged that a CoWSwap integration on Aave’s interface had quietly routed swap-related fees away from the community treasury to an external recipient. That surfaced a deeper tension. Aave Labs had been building application-layer products (Aave Pro, Aave App, Horizon, Aave Kit) that sat atop the permissionless protocol but whose economics bypassed AAVE holders entirely. The Labs vs. DAO split, in other words.

On April 12, the “Aave Will Win“ (AWW) proposal passed with overwhelming support: 75% of voters voted in favor (aave.com). The proposal does three things: (i) redirects 100% of revenue from every Aave-branded product to the DAO; (ii) pays Aave Labs a one-time package of $25M in stablecoins plus 5,000 AAVE; and (iii) consolidates brand and economic rights under the token itself. An on-chain AIP ratification still has to clear. That’s the real catalyst.

Aave protocol revenue hit $140M in 2025 and is tracking to roughly match that in 2026. The incremental piece is app-layer revenue: swaps on Aave.com and Aave Pro are already running at $10–20M/year that previously accrued to Labs, with Aave App, Horizon, and Aave Kit layered on top. A conservative 10× revenue multiple on ~$20M of incremental accrual is roughly $200M of implied fundamental value, or ~13% of AAVE’s fully-diluted cap. AAVE traded up nearly 30% after the vote.

Now execution has become the biggest question for Aave. BGD Labs, the DAO’s primary technical contributor, wound down its partnership on April 1 (BGD governance post, Feb 2026). ACI, another leading service provider for Aave, is choosing to leave the ecosystem in July of this year. V4’s cross-chain hub-and-spoke rollout, which Stani has publicly targeted as a trillion-dollar credit layer, now needs to ship without its most senior engineering counterparty.

2. Artemis Research Spotlight: $PURR

Artemis Research Spotlight: Hyperliquid Strategies ($PURR): The Only DAT With a Productive Underlying Asset by Zheng Jie Lim.

The framing: most DAT vehicles wrap an asset that either earns nothing (BTC) or runs at a negative operational cash flow once staking emissions are netted against fees (ETH: $526M fees vs. $3,035M in emissions; SOL: $680M vs. $3,936M). Hyperliquid is the outlier. It generated $857M in 2025 fees (of which $797M from perps at a 2.72 bps take rate), with 99% of earnings flowing to the Assistance Fund for buyback-and-burn ($837M on $857M of fees). Staking emissions come from a pre-allocated reserve rather than fee revenue, which the piece argues is the right way to evaluate operational cash flow. On that basis, HYPE is the only major crypto asset with positive underlying unit economics. Per outstanding HYPE (477M), that works out to roughly $1.76 of earnings per token. Hyperliquid is also the highest revenue generating company in the world per employee:

The vehicle: PURR (NASDAQ) is a zero-debt, zero-preferred, zero-convertible wrapper around 18.8M HYPE plus $112.6M cash, trading at ~1.12x mNAV with a $30M buyback program (~$10.5M deployed, retiring ~3M shares) and a $1B equity line as optional dry powder. Cap table anchored by Paradigm, D1, Galaxy, and Pantera; board includes Bob Diamond (ex-Barclays), David Schamis (Atlas Merchant / ex-JC Flowers), Eric Rosengren (ex-Boston Fed), and Larry Leibowitz (ex-NYSE COO). The piece frames the structure as the inverse of the Strategy trade: Strategy issued equity at premium to fund BTC accumulation (a mechanism that breaks when the premium compresses, which is where the category sits today at ~1.15x mNAV); PURR’s buyback program mechanically increases HYPE-per-share only when the stock trades below NAV.

Growth vectors laid out in the note: HIP-3 (permissionless perps, 500k HYPE stake per market, expanding TAM from $3–5T crypto derivatives toward the >$100T global derivatives market), HIP-4 (options + prediction markets, mainnet Q4 2026), builder-routed distribution (~40% of DAUs already flow through third-party frontends), and USDH (native stablecoin, 50% of reserve yield to the Assistance Fund, currently ~$1M/yr to buybacks at $93M supply, scaling to $40–100M/yr at $2–5B supply).

Valuation summary:

• Base: HYPE at $76 by 2030 on $1.71B of earnings at 20x P/E, PURR held at 1.1x mNAV → $10.59/share, +63% over 5 years.

• Bull: HYPE at $127, mNAV 1.3x → $20.84, +220%.

• Bear: HYPE at $27 at 16x P/E, mNAV 0.95x → $3.32, -49%. Bear-case mitigant: zero liabilities mean no forced-selling dynamic, the protocol still generates $782M of fees at trough, and the token remains deflationary net of emissions.

Key risks: HYPE drawdown, potential CFTC/SEC enforcement on Hyperliquid’s no-KYC operations (where PURR itself actually benefits as the regulated wrapper), equity-line dilution if drawn at the wrong price, and general DAT-premium compression.

Full note on Artemis Research.

3. Robinhood leads Digital Finance theme growth

HOOD was +31.16% on the week, the best-performing name on the crypto-adjacent equities list (ahead of COIN +22.93% and CRCL +20.30%). The stock had seen a nearly 60% drawdown since October highs before the rebound. A major catalyst this week was the SEC’s reversal of the “pattern day trader” rule, which had been an inhibitor to retail trading volume. This is expected to bring more trading volume to Robinhood, arguably the biggest beneficiary of this rule reversal.

Fundamentals:

Robinhood has also seen exceptionally strong fundamental growth. KPIs have grown significantly over recent quarters. The growth in Gold Subscribers is a particularly important metric to look at since it is a key component of Robinhood’s non-trading revenue.

Robinhood saw record quarterly revenue in Q4 2025, growing at 26.5% YoY. Revenue growth was driven by strong net interest income growth. NII made up about 30% of Robinhood’s top line in Q4 2025, and grew 38.9% YoY.

The sell-side view: Bernstein reiterated Outperform with a $130 PT on April 14, ~60% above spot. Their thesis centers on Bitstamp, the $200M institutional crypto exchange Robinhood closed in June 2025, which now drives ~60% of HOOD’s total crypto transaction volume. Bernstein models 2026 crypto revenue at $1.1B (+23% YoY, 31% above consensus) (Bernstein via Parameter, Apr 17; Robinhood/Bitstamp press release, Jun 2, 2025).

HOOD is one of the clearest expressions of retail brokerage and crypto infrastructure converging into one product surface.

Charts of the Week

Chart 1: HIP-3 Open Interest has stalled but remains above $2 billion

Chart 2: We highlight the effect of incentives on prediction markets in our new Prediction Markets dashboard: Dashboard Link, Analysis Link

Other Notable News

• Morgan Stanley launches MSBT at 0.14%: First spot BTC ETF issued directly by a major U.S. bank. Over $100M in week-one flows.

• Schwab introduces spot crypto trading: Schwab is one of the world’s largest brokerages, and is planning on a phased roll out of spot crypto trading to its 38.5m accounts.

• SEC CLARITY Act roundtable held April 16. Market-structure bill has bipartisan noise but no markup date. Watch the Senate Banking Committee calendar next week.

Thanks for reading! Stay ahead this week by using the Artemis Terminal to pull the underlying data on any of the stories above (Aave revenue, Robinhood financial and comps, HIP-3 OI by market type) or pull live numbers straight into your models with =ART() in Excel.

Disclaimer: This newsletter is produced by Artemis for informational and educational purposes only. It does not constitute investment advice, a recommendation to buy or sell any security or digital asset, or an offer to provide advisory services. Artemis and its employees may hold positions in assets discussed. Figures are accurate to the best of our knowledge as of publication; markets move quickly.

The market is pricing PURR as another digital asset treasury copycat — an MSTR-style vehicle where a premium lets management issue equity, buy more tokens, and accumulate more per share. We think that framing misprices the business. Strategy spends $2.3 million a day to service $835 million of annual preferred and convertible obligations on top of 780,897 bitcoin that earn nothing, and its mNAV has collapsed from peaks above 6x to 1.15x today. PURR carries zero debt, zero preferreds, and zero recurring obligations, and owns 18.8 million HYPE tokens in the only major protocol that produced positive earnings in 2025: $857M of fees ($797M from perps at a 2.72 bps take rate), 99% of which flows to the Assistance Fund — $837M of Earnings bought back and burned at near-zero operating cost. The token is structurally deflationary: ~19M HYPE bought back annually against ~7M emitted from staking reserves. The stock trades at 1.12x mNAV. Our base case carries HYPE to $76 on $1.71B of 2030E Earnings at 20x P/E, holds PURR at 1.1x mNAV, and implies a $10.59 share of ~63% upside over five years. Bull case at $127 HYPE and 1.3x mNAV implies ~$20.84, +220%. Bear case compresses HYPE to $27 at 16x P/E and 0.95x mNAV at −49%.

Comps:

PURR is the only DAT that combines a productive underlying, positive earnings, and zero liabilities.

What Does $PURR Do?

Hyperliquid Strategies Inc (NASDAQ: PURR) is a digital asset treasury company whose sole mandate is to accumulate and hold HYPE, the native token of the Hyperliquid protocol. The company was formed in December 2025 via an $888M business combination involving Sonnet BioTherapeutics, Rorschach I LLC (a Paradigm-affiliated SPAC), and a new entity spun up by Atlas Merchant Capital.

The balance sheet is the cleanest in the DAT category: 18.8M HYPE, $112.6M of cash, zero debt, zero preferreds, zero convertibles. A $30M stock buyback program was authorized in January 2026. As of February 3, 2026, $10.5M was deployed, retiring ~3.0M shares and pulling the fully diluted count to 150.8M. A $1 billion equity line sits on top of the cash position as optional dry powder for HYPE drawdowns.

The underlying is the asset that makes this structure unusual. HYPE is an equity-like claim on Hyperliquid, a perps exchange that generated $857M in 2025A fees with 11 employees (~$78M rev/employee, the highest of any company in the world).

The Fee Machine

Hyperliquid is an exchange.

The model is a volume × take-rate business with near-zero marginal cost.

The fee waterfall breaks down into six line items:

Thesis #1: The Only Underlying Asset That Earns

Every DAT on a U.S. exchange today wraps an asset that either earns nothing (BTC) or runs at a loss once you account for the tokens minted to keep the chain alive. Ethereum generated $526M of fees but paid $3,035M in staking emissions; Solana generated $680M but paid $3,936M. Hyperliquid generated $857M in fees. Earnings definitions matter here. For Hyperliquid, earnings = fees after the 1% HLP vault cut (3% pre-Aug 30, 2025), with 99% flowing to the Assistance Fund for buyback-and-burn: $837M on $857M of fees. For Ethereum and Solana, the comparable figure is fees minus staking issuance, since their chains cannot function without token emissions to pay validators. This is a true operational cost, and both are negative at time of writing. Hyperliquid’s $312M of staking emissions are drawn from a pre-allocated reserve, not from exchange revenue, so they sit below the line. On the adjusted basis ($545M), Hyperliquid is the only protocol with positive underlying cash flow assets. The performance gap shows up on DAT balance sheets. BMNR’s average ETH entry was $2,826; Forward Industries bought SOL at $232. Both are underwater.

PURR is the only DAT sitting on meaningful unrealized gains, up roughly $600M on its HYPE position. The underlying asset isn’t just productive, it’s appreciating.

Five growth drivers are compounding this advantage:

1. HIP-3: Hyperliquid becomes a listing platform. Live since October 2025, HIP-3 lets any builder stake 500,000 HYPE to deploy permissionless perps on any asset — commodities, equities (China, Korea, Japan), FX, exotics. This transforms HL from a crypto trading venue into a general listing layer. TradeXYZ’s crude-oil HIP-3 market processed $305M of weekend notional during the Strait of Hormuz crisis; cross-asset weekend prices tracked TradFi reopen at R² = 0.785 (Shaunda from Blockworks). The listing TAM expands from $3-5T of crypto derivatives to the >$100T global derivatives market. Each HIP-3 deployment requires a permanent 500k HYPE stake. HIP-4 carries the same requirement. Twenty markets at scale retire ~10M HYPE (2.1% of Outstanding Supply) from the float permanently.

2. DAT as a builder partnership. At current prices, 500k HYPE is roughly $23M. Most builder teams cannot self-fund that stake. A large holder like PURR (18.8M HYPE) is a natural counterparty: financing or partnering on HIP-3/HIP-4 deployments in exchange for a share of the market’s fees. This creates a revenue stream and ecosystem influence that individual HYPE holders cannot replicate at scale.

3. HIP-4: options and outcome contracts. Testnet launched March 2026; mainnet targeted Q4 2026. Deribit, which got acquired by Coinbase, traded over $1,875B in options notional value in 2025, making up over 85% of the market (~$2.2T). Options are just ~3% of crypto derivatives today (as Coinglass sizes the total crypto derivatives volume at $85.7T). If onchain options reach 15% penetration and HL captures half, that’s $165B of volume at ~8bps (options carry wider spreads than perps, deribit charges 12bps) = $135M/yr in fees. Our Base model assumes $50M in 2026E scaling to $130M by 2030E. Prediction-market outcome contracts open a second HIP-4 fee line (Annualized fees for Polymarket are reaching $700M).

4. Builder codes: negative CAC distribution. ~40% of HL DAUs already route through third-party frontends (Phantom is the largest). Builders earn a rev-share for every trade routed — $40M+ paid out to date (Dwellir). Compare Coinbase’s $400-600 retail CAC per funded account: HL’s is negative. This is why the model grows perp volume from $2.9T (2025A) to $5.2T (Base 2030E). The distribution layer is outsourced.

5. USDH: native stablecoin with revenue sharing. Native Markets won the USDH bid (Sept 2025). Reserves sit in a BlackRock-managed fund; 50% of reserve yield routes to the Assistance Fund.

At current supply (~$93M) and 3.7% T-bill yields, USDH contributes ~$1M/yr to the AF buyback. At $2-5B supply, that scales to $40-100M/yr. With the addition of HIP-4 which could only be traded on USDH, this could be a massive tailwind for the Native Markets stablecoin.

None of this expansion requires Hyperliquid to hire. Every growth vector (HIP-3 listings, HIP-4 options, builder distribution, and USDH float) is outsourced to external teams who take listing and distribution risk in exchange for a rev-share. The 11-person core team writes the protocol and the fee plumbing. This is a platform-scaling model: throughput, listings, and frontends grow without headcount.

Thesis #2: Buy the Wrapper at Cost

Strategy issued equity at a premium, converted the cash to BTC, and called it financial engineering. The mechanism inverts the moment the premium collapses, which is where the entire DAT category sits today (Strategy at 1.15x, down from 6x). PURR runs the opposite trade: the stock sits at ~1.12x NAV, and the $30M buyback program deploys only when the market prices PURR below the tokens it already owns. Every dollar deployed sub-NAV mechanically increases HYPE-per-share. The mechanism works both ways. Below NAV, management buys back stock and each remaining share represents more HYPE. Above NAV, they can issue shares at a premium and use the proceeds to buy more HYPE, also increasing HYPE-per-share. Either direction accretes value to existing holders.

No other DAT operates this way. Strategy, BMNR, and Forward Industries issue shares to buy tokens but have no buyback mechanism when the premium compresses. Their shareholders absorb the dilution on the way up and get nothing back on the way down.

Where direct HYPE wins: no corporate overhead, no dilution risk from the equity line, no regulatory risk specific to PURR management, and 100% of the token upside. Staking/airdrop optionality accrues only to direct holders.

But the wrapper offers four things direct HYPE doesn’t:

1. Automatic accretion without additional risk: the buyback described above is funded by cash already on the balance sheet. No margin calls, no liquidation risk, no action required from the holder. A direct holder can lever up through perps or borrowing, but that adds counterparty and liquidation exposure.

2. The regulatory moat: PURR and HYPD are the only two NASDAQ-listed vehicles for HYPE exposure, and any enforcement action on Hyperliquid’s no-KYC operations pushes institutional demand INTO the wrapper

3. Zero obligations: no debt service, no forced selling, no preferred shares.

4. Tax efficiency: PURR is taxed as a standard equity. Holders who hold for over a year pay long-term capital gains (20% federal max), can hold shares in IRAs and 401(k)s for tax-deferred or tax-free growth, and use tax-loss harvesting against other equity positions. Direct HYPE holders pay ordinary income tax (up to 37%) on every staking reward the moment it’s received, have no access to tax-advantaged retirement accounts, and face unresolved IRS guidance on airdrop cost basis. For a top-bracket U.S. investor, the wrapper roughly halves the tax drag.

The underlying exchange produces $1.76 of earnings per Outstanding HYPE ($837M ÷ 477M Outstanding Supply - we introduced this metric with Pantera Capital back in August 2025).

Valuation & Scenarios

Valuation is a two-step chain: first we value HYPE on fundamentals (2030E earnings × P/E ÷ Outstanding Supply), then we value PURR as NAV per share at that HYPE price multiplied by a target mNAV.

HYPE Scenarios

We value HYPE on earnings (fees × 99% after HLP) multiplied by a terminal P/E:

PURR Scenarios

Translating HYPE scenarios to PURR uses Adjusted NAV per share = (18.8M HYPE × price + $112.6M cash − $95.8M DTL + $4.5M adj) ÷ 150.8M FD shares, then applies a target FD mNAV. DAT stocks with productive underlyings have historically re-rated to 1.1x–2.0x NAV in up cycles; category norm at par is 1.0x.

Base case: HYPE at $76 by 2030, Adjusted NAV per share of $9.63, held at 1.1x NAV, implying a $10.59 stock and +63% upside over five years.

Bull case: HYPE reaches $127 by 2030, Adjusted NAV of $16.03, re-rate to 1.3x NAV, implying $20.84 and +220% upside.

Bear case: HYPE compresses to $27, Adjusted NAV drops to $3.49, mNAV slips to 0.95x, implying $3.32 and –49% downside. Base case returns ~10% annualized, but the distribution is asymmetric: bear downside is floored by zero liabilities and a protocol that still generates $782M of fees at trough, while bull upside compounds from both HYPE appreciation and mNAV re-rating.

Management & Cap Table

PURR is a balance sheet vehicle. The only job is capital allocation: when to buy HYPE, when to buy back shares, when to draw the equity line, and when to sit still. The team running that decision has 80+ combined years in capital markets, bank balance sheet management, and exchange infrastructure. Paradigm, the largest crypto-native fund in the world ($12.7B AUM), anchored the SPAC. D1, Galaxy, and Pantera round out a cap table that bridges TradFi and crypto. Diamond’s institutional network doubles as a distribution channel for allocators who want HYPE exposure but can’t custody tokens or navigate crypto tax complexity.

• Bob Diamond (Chairman) — Former CEO of Barclays; co-founder of Atlas Merchant Capital.

• David Schamis (CEO) — Co-founder of Atlas Merchant Capital; ex-JC Flowers partner.

• Eric Rosengren (Board) — Former President, Federal Reserve Bank of Boston (2007-2021).

• Larry Leibowitz (Board) — Former COO of the New York Stock Exchange; Operating Partner at Atlas Merchant Capital

Anchor investors include Paradigm, D1, Galaxy, and Pantera.

Risks

1. HYPE Price Drawdown

PURR is a levered claim on HYPE’s price. A broad crypto bear market compresses volumes, fees, and the buyback pool. Our bear case models this: TAM plateaus at $85T, HL share holds at 3%, take rate drifts to 2.3 bps, P/E de-rates to 16x, implying $27 HYPE and $3.32 PURR (−49%). Outstanding Supply is ~477M; Bear assumes a 35% team claim rate in 2026–27.

Mitigant: Zero liabilities mean no forced-seller dynamic. Even in the bear case, Hyperliquid still generates $782M of fees and $774M of earnings — the buyback-and-burn continues to exceed staking emissions, and the token remains deflationary (−23M HYPE/yr net). The $1B equity line is optional, not mandatory.

2. Regulatory Action on Hyperliquid

Hyperliquid operates without KYC. The Futures Industry Association has filed formal complaints with U.S. regulators requesting enforcement on U.S.-person access to offshore perps. An adverse action would compress HYPE directly. The HIP-3 expansion into TradFi assets (silver, oil, equities) increases regulatory surface area — commodity perps may draw CFTC attention beyond the crypto-specific enforcement actions seen to date.

Mitigant: PURR itself is a fully compliant NASDAQ-listed entity. A regulatory action would most likely push U.S. institutional demand into the PURR wrapper, which is the regulated way to own HYPE economics.

3. Equity Line Dilution

The $1B equity line can be drawn if HYPE dips, but drawing at the wrong price dilutes existing holders. At the bear-case $27 HYPE, PURR Adj NAV is $3.49/share — drawing at that level would issue shares below $6.51 spot.

Mitigant: The line is optional and at management’s discretion. Current posture is buyback, not issuance — $10.5M of the $30M authorized program was deployed by early Feb 2026, retiring ~3M shares and pulling FD count to 150.8M. The Paradigm-led cap table has no incentive to dilute itself.

4. DAT Premium Compression

Any premium that opens above NAV in a rally can compress quickly when sentiment turns. Strategy’s 6x mNAV premium collapsed to 1.15x FD in under 12 months. In the bull case, PURR re-rates to 1.3x NAV ($20.84); if sentiment reverses and the multiple compresses back to 1.1x, the loss is capped at the mNAV delta.

Mitigant: PURR trades at 1.12x mNAV today. The premium is 12 cents on the dollar versus Strategy’s peak of $5+ on the dollar. The asymmetry runs upward: 1.3x in a bull cycle is modest relative to the 2.0x+ that DAT stocks with productive underlyings have historically reached. At 1.0-1.1x, the wrapper’s value is the underlying asset.

Disclosure: This material is provided for informational purposes only and does not constitute investment advice, financial advice, trading advice, or any other form of advice. The views expressed are those of the authors and should not be relied upon as a recommendation to buy, sell, or hold any asset. The authors or affiliated entities may hold positions in the assets discussed. You should conduct your own research and consult appropriate financial professionals before making any investment decisions.

Geopolitics dominated again — but this week it broke the other way. After five weeks of war-driven compression, Tuesday night’s (April 7, 2026) Trump-Iran ceasefire announcement triggered one of the sharpest single-day risk reversals in recent memory. The Dow surged 1,325 points (+2.85%) on Wednesday, its best day since April 2025. The S&P 500 jumped 2.51% to 6,782, breaching its 200-day moving average for the first time since the conflict began. WTI crude collapsed 16.4% to $94.41 — its biggest daily drop since April 2020 — as Iran confirmed it would allow passage through the Strait of Hormuz for the two-week truce period. The deal is fragile: Iran’s parliament speaker briefly claimed a US violation Wednesday afternoon, and JPMorgan cautioned that “a two-week pause is not a resolution.” The clock expires April 22.

Crypto finally caught a bid. BTC surged from ~$68K to ~$72.7K on the ceasefire news, finishing the week up ~9% and triggering nearly $600M in leveraged liquidations — over $400M from shorts that had piled in during weeks of range-bound drift. ETH gained ~9.3% on the week. The standout was ZEC, up a striking +59.7%, though on idiosyncratic rather than macro drivers. HYPE continued its strong run, up +16.8%, while SKY gained +4.97%. On the losing side, DRIFT extended its post-exploit collapse, down another -28.1% on the week following the $285M April 1 hack, and TAO gave back -16.59% after its recent surge. Equities proxies tracked closely with the relief rally: eq-COIN +3.6%, eq-QQQ +4.46%, while eq-HOOD (-2.11%) and eq-CRCL (-2.46%) lagged on continued CLARITY Act yield-ban uncertainty.

The week’s average across tracked assets came in at +1.25%, median +1.25% — a modest but real improvement from the prior four weeks.

Today We Highlight:

• Morgan Stanley MSBT — the first US bank spot Bitcoin ETF

• CLARITY Act: Senate returns April 13 — the next major catalyst

• Corporate Bitcoin buying: Strategy crosses 767K BTC as DAT total nears 900K

• Artemis Analyst Sector Deep-Dives: Agentic Payments, Lending, Trading, Neobanks, and Prediction Markets

1. Morgan Stanley MSBT: the first US bank spot Bitcoin ETF

On April 8, Morgan Stanley launched the Morgan Stanley Bitcoin Trust (MSBT) on NYSE Arca, becoming the first major US bank to issue its own spot Bitcoin ETF. The fund drew ~$34M in net inflows on day one. Bloomberg ETF analyst Eric Balchunas placed the debut in the top 1% of all ETF launches and projects $5B AUM in year one.

Why does the fee matter?

MSBT launches at 0.14% — 11 basis points below BlackRock's IBIT at 0.25%. On a $100K investment that's $110/year in savings. Narrow in absolute terms, but it sets a new floor. Fee compression in Bitcoin ETFs is accelerating, and Morgan Stanley just fired the loudest shot yet.

Why is Morgan Stanley's distribution so powerful?

Morgan Stanley's 16,000 wealth advisors oversee $9.3 trillion in client assets. Previously they could recommend third-party Bitcoin ETFs; now they can recommend one that keeps the fee in-house. That internal incentive shift is structural — and it means new long-duration capital gets directed to MSBT rather than existing funds. The first week of flows will be the real tell.

Is MSBT a standalone product or something bigger?

It's the wedge. Morgan Stanley filed for Ethereum and Solana trusts in January, applied to the OCC for a National Trust Bank Charter in February covering custody and staking, and plans to launch retail spot crypto trading on E*Trade in H1 2026. The pattern is clear: Morgan Stanley is building a full-stack digital asset business under its own brand. MSBT is how they get distribution.

Custody: Coinbase (cold storage).

Cash custody and administration: BNY Mellon.

Authorized participants: Jane Street, Virtu Americas, Macquarie Capital.

Benchmark: CoinDesk Bitcoin Benchmark 4PM NY Settlement Rate.

What this means for the Bitcoin ETF landscape?

IBIT remains structurally protected by its liquidity depth and options ecosystem, most active traders won't leave for 11bps. But the long-duration capital flowing through wirehouse advisors is a different market entirely. The landscape is bifurcating: IBIT for traders, MSBT for allocators.

2. CLARITY Act: Senate returns April 13 — what to watch

The Senate returns from Easter recess on April 13 with the CLARITY Act heading for Senate Banking Committee markup in the final weeks of April. This is the most consequential near-term regulatory catalyst for crypto — and the one most likely to move individual names.

What's actually been agreed?

The yield language is "99% resolved" per Senators Tillis and Alsobrooks: passive yield banned, activity-based rewards permitted. That same logic is now baked into the FDIC's GENIUS Act proposed rule — the two regulatory tracks are moving in deliberate lockstep.

What's still unresolved?

Two things.

First, DeFi scope — the “digital asset service providers” language leaves it ambiguous whether protocols like Aave or Morpho are in scope, and that ambiguity alone is a headwind for the sector.

Second, token classification provisions that determine the SEC vs. CFTC jurisdictional split. XRP has been the primary beneficiary of optimism on this front.

On the winners side: Coinbase benefits most clearly, as the yield ban improves USDC distributor margins while its OCC trust charter positions it as the institutional plumbing layer the new framework requires. Circle's $900M+/year revenue share from Coinbase is untouched, though the retail yield acquisition story weakens. Tether is largely unaffected — its model already aligns with the framework.

Why does the timing matter now?

Market participants have been treating CLARITY as a second-half 2026 story. With the Senate back April 13, final-week markup sessions in April are now realistic. That timeline repricing alone could move names.

3. Corporate Bitcoin buying: Strategy crosses 767K BTC as DAT total nears 900K

Strategy added another ~4,871 BTC early in the week, pushing its total to 767K BTC as of April 9 — extending its lead as the dominant corporate holder by a wide margin. The broader Digital Asset Treasury (DAT) cohort now collectively holds ~899.9K BTC across seven tracked entities, approaching the psychologically significant 900K mark. That's roughly 4.3% of total Bitcoin supply sitting on corporate balance sheets.

How does Strategy compare to everyone else?

The gap is extraordinary. Twenty One Capital sits at 43.5K, Metaplanet at 35.1K, Bitcoin Standard Treasury Company at 30K, Strive at 13.1K, Nakamoto Holdings at 5.8K, and American Bitcoin at 5.4K. The next six combined don't reach 20% of Strategy's stack.

Are these companies actually making money?

Not yet, at current prices. Strategy is sitting on approximately -$2.9B in unrealized losses against its blended cost basis. Most of the DAT cohort accumulated heavily during BTC’s $80K–$100K+ run in late 2025, meaning the majority are currently offside. The ceasefire rally this week provided partial relief, but a sustained break above $80K is what most of these balance sheets need to return to breakeven.

The accumulation continues regardless. Strategy’s behavior through the drawdown reflects the thesis it has always operated on: price is irrelevant at the margin, supply absorption is the objective.

Charts of the Week

Privacy Coins lead all sectors in performance

USDs Stablecoin Supply Surges +36% since December 2025

Hyperliquid leads 7D Top Net Flows Chart

Other notable news

• Polymarket exchange upgrade — Rebuilt trading engine, new smart contracts, and a native Polymarket USD collateral token replacing USDC.e, rolling out over 2-3 weeks. Comes as daily fees hit $1.9M.

• Polygon Labs — $100M stablecoin payments raise — Early talks to raise $100M for a regulated stablecoin payments business. Would put Polygon in direct competition with Stripe/Paradigm-backed Tempo, which launched mainnet last month.

• ZachXBT — North Korean IT worker network — Data from an internal DPRK payment server exposed 390 accounts running ~$1M/month through fake identities and crypto-to-fiat networks. The most detailed public view of North Korea’s crypto laundering infrastructure to date.

• Canary Capital PEPE ETF S-1 — Filed following prior MOG and Pengu filings. Whether the SEC engages substantively will signal how far the new regulatory posture actually extends.

• France crypto wallet declaration law — Taxpayers holding >€5K in self-custody must now declare to tax authorities. Significant expansion of crypto surveillance in the EU’s largest retail crypto market.

• Coinbase ASIC license — First crypto exchange to secure an Australian Financial Services Licence. Consistent with the broader Coinbase institutional buildout narrative this week.

• Roman Storm / Tornado Cash — Prosecutors rejected the bid to dismiss charges; retrial on money laundering and sanctions counts stays open. The most-watched smart contract developer liability case remains live.

Thanks for reading! Stay ahead this week by using the Artemis Terminal or pulling live data with =ART() in Excel.

Disclaimer: The authors of this content, as well as affiliates of Artemis Analytics, may have financial interests in the protocols or tokens mentioned. This does not constitute investment advice or a recommendation to buy, sell, or hold any asset. The information provided is for educational purposes only and should not be relied upon for financial, legal, or tax decisions. Readers should assess their own circumstances before making any financial choices. Views expressed may change without notice, and Artemis Analytics is not liable for any losses resulting from the use of this content.

Main Takeaway: Stablecoin capital flows between chains predict cross-sectional L1 returns, delivering a 1.67 Sharpe over 5 years that is almost entirely independent of broad crypto market direction.

Executive Summary

• We constructed a weekly-rebalanced long-short factor from tracking stablecoin flows.

• The raw strategy delivers a 1.67 Sharpe and +83.6% annualized return over five years, with a -43.9% max drawdown. Adding a volatility-targeting risk overlay compresses the Sharpe to 1.17 but cuts the max drawdown to -31.9%.

• The factor is effectively market-neutral and inversely correlated with down markets: market beta is -0.03 (not significant) with an R² of 0.1%. In BTC-negative months, the raw factor averaged +6.8% while BTC averaged -10.9%.

• Annualized alpha is +73.8% for the raw factor after controlling for market exposure (t-stat 3.31, p = 0.001), statistically significant at the 1% level.

• The factor is the least-correlated factor in our stack (max pairwise correlation 0.16) and retains significant alpha (t-stat 2.54) after controlling for all other factors simultaneously in a spanning regression.

• The long leg generates 84% of returns. The factor is fundamentally a positive-flow detector for mid-cap L1s and L2s rather than a structural short on mega-caps. Five chains (Polygon, Mantle, Optimism, BSC, Sei) account for 84% of total returns.

Goal

Stablecoins 1 was created to test the theory that stablecoin flows can predict L1 performance. The core idea is that stablecoins moving onto a blockchain is a leading indicator that the chain’s native token is about to appreciate. For example, when dollars in the form of stablecoins (eg. USDC, USDT, etc.) flow onto Solana, it signals capital is arriving. That same capital will buy SOL, pay SOL gas fees, provide liquidity in SOL pools, and generally drive demand for the SOL token. On the other hand, when stablecoin supply is shrinking, capital is leaving, and the chain’s token will underform.

This thesis, if true, should produce returns that are independent of the broader crypto market because the factor trades on relative capital allocation between chains rather than market direction. We test this independence property below.

To learn more about our methodology, request detailed testing results, or explore commercial opportunities, please reach out to the Artemis team.

Process

Stablecoins 1 is a long-short factor using on-chain data to look at stablecoin and L1 data sourced from Artemis.

Every week, the strategy looks at all 18 chains (Ethereum, Solana, BNB, Arbitrum, etc.) and ranks them by signals that all measure stablecoin flow direction. The signals are then converted to a rank score to see how each chain stands relative to its peers, averaged into a composite score, and sorted into quintiles. The strategy longs the top quintile (about 3-4 chains with strongest inflows) and shorts the bottom (about 3-4 chains with the worst outflows), holding for one week, and then rebalances.

We also explored adding risk overlays to limit downside. Positions are scaled each week to target 20% annualized realized volatility with a leverage cap of 1.5x, and exposure is further reduced during deep drawdowns measured against a rolling peak. Using a rolling peak ensures that we can resume the full-size trading strategy after extended flat periods, avoiding permanent deleveraging after a deep drawdown. Unlike Fundamentals 1, there’s no BTC regime filter because the factor naturally backs off during down crypto markets as capital leaves, stablecoin supply contracts, and the strategy’s exposure shrinks.

Performance & Results

Backtested Performance

We look at the performance of Stablecoins 1 since its inception on March 1, 2021. We can compare looking at the raw factor without the risk overlay and the factor with volatility scaling.

Returns are gross of transaction costs as realized performance would depend on execution costs and liquidity at time of trade. However, we have found this strategy to be highly cost resilient. The breakeven transaction costs are estimated to be 197 bps per trade. This is a result of moderate weekly turnover (57.9%) and high liquidity of L1 tokens.

The raw long-short strategy delivers the highest annualized return (+83.6%) and the best Sharpe (1.67), with cumulative returns of +2,164% since inception. Adding the risk overlay reduces the Sharpe to 1.17 but also cuts the max drawdown from -43.9% to -31.9% and the worst month from -25.3% to -20.8%. The risk overlay trades roughly half the raw return in exchange for a meaningfully smoother profile.

Looking at yearly performance, Stablecoins 1’s strongest years were 2021 (+262% raw) and 2025 (+315% raw), with more muted performance in 2022 (+47%) and 2023 (+20%). 2024 was the factor’s one losing year (-13% raw, -18% with overlay), driven by a single sustained drawdown (sets the -43.9% max drawdown) from September to November where chains the signal identified as capital-losing outperformed chains it identified as capital-attracting for roughly two months. The factor recovered fully in Q1 2025 and went on to deliver its best year on record. Solana in particular contributed disproportionately to this period. Short positions on Solana have been the single largest negative contributor across the entire backtest, suggesting the signal does not capture Solana’s idiosyncratic dynamics well, and the September-November 2024 window was one of the periods where this weakness was most acute.

The 2024 drawdown coincided with a period of stagnant aggregate stablecoin supply. Total stablecoins in circulation across the universe had been roughly flat since mid-2022 and didn’t resume meaningful growth until November 2024, as the chart above shows. Cross-sectional factors like this one work best when capital is actively moving at scale: when aggregate supply is expanding, the marginal dollar of new stablecoin issuance has to pick a chain, and those choices reflect real allocation decisions that the factor can profitably rank. When aggregate supply is stagnant, the cross-sectional signal is noisier because the flows that do exist are smaller and more likely to reflect idiosyncratic activity (exchange migrations, bridge events, yield rotations) rather than genuine capital migration. Our regime analysis found that the factor earned +93.9% annualized in weeks when aggregate stablecoin supply was expanding vs +30.2% in contracting weeks, still positive, but meaningfully weaker. The regulatory clarity provided by the GENIUS Act in 2025 and the subsequent institutional issuance wave are plausible drivers of the renewed aggregate growth visible from late 2024 onward, and the factor’s 2025 performance (+315% raw) is consistent with a signal that was always there but needed sufficient flow volume to produce clean cross-sectional dispersion.

Analysis Against the Market

We can compare the performance of Stablecoins 1 with and without the overlays to market (using BTC as proxy for broad crypto market exposure).

Stablecoins 1, both with and without the risk overlay, dramatically outperformed the market. In terms of monthly outperformance, the Raw L/S strategy beats BTC in 41 of 61 months (67%), and the risk-overlay version beats BTC in 36 of 61 months (59%). BTC delivered only +3.5% annualized over the backtest period with a Sharpe of 0.06 and a -74.5% max drawdown, so both versions of the stablecoin factor offer much better risk-adjusted returns than simply holding crypto.

The more interesting property is the factor’s behavior across market regimes. We split the backtest into months where BTC was positive (31 months) and months where BTC was negative (30 months) and measured the factor’s performance in each.

Stablecoins 1 underperforms BTC in rallies (+5.9% vs 13.6% average monthly) because the short leg acts as a drag when most tokens are rising, which is the expected cost of market neutrality. But in BTC negative months, the raw factor averaged 6.8% per month while BTC averaged -10.9%. Not only is the factor positive during drawdowns, it actually performed slightly better in down months than up months. The short leg of chains with contracting stablecoin supply tends to fall greater than the long leg during selloffs, generating positive spread returns exactly when broad crypto exposure hurts the most.

This asymmetry makes Stablecoins 1 valuable as a portfolio diversifier as its return stream actively profits during crypto bear markets.

Market Independence

To test Stablecoins 1’s independence from broad market dependence, we regressed the weekly returns against BTC.

The raw long-short factor has essentially zero market beta (-0.03) and an R² of just 0.1%, meaning less than one-tenth of one percent of the factor’s return variation is explained by whether crypto is going up or down. A t-test on the null hypothesis that beta equals zero fails to reject (p = 0.65), confirming that the long-short construction is statistically indistinguishable from market-neutral. After applying the risk overlay, market beta moves slightly to -0.04 with an R² of 0.4% and is still not statistically different from zero (p = 0.32).

Annualized alpha is +73.8% for the raw factor (t-stat = 3.31, p = 0.001) and +41.6% after the risk overlay (t-stat = 2.66, p = 0.008). Both are statistically significant at the 1% level, indicating the stablecoin factor generates significant returns that cannot be attributed to market exposure.

One question worth addressing is: is this factor a bet on one stablecoin issuer over another? We checked by computing the dollar-weighted USDC, USDT, and “other” share of every long and short pick across the full backtest. The long book averages 46% USDC, 37% USDT, and 17% other stablecoins (DAI, FDUSD, native chain stables, etc.), while the short book averages 37% USDC, 40% USDT, and 23% other. The 9% USDC tilt on the long side is real but modest. The more interesting gap is in the “other” bucket: the factor is meaningfully less likely to go long chains dominated by native or algorithmic stablecoins (cUSD on Celo, FDUSD on BSC). This is consistent with Stablecoins 1’s core thesis of rewarding chains attracting external stablecoin capital and penalizes chains dependent on internally-minted stables. Notably, the chains that drive most of the strategy’s returns span the full composition spectrum (Polygon at 52% USDC, Mantle at 54% USDT, BSC at 54% USDT, Sei at 95% USDC), which is the strongest possible evidence that the factor is a flow-direction bet rather than a stablecoin-issuer bet.

Where the Alpha Comes From

To find what is driving our 1.67 Sharpe, we attributed Stablecoins 1’s cumulative return to each chain individually, splitting contribution between long-leg and short-leg trades using the per-asset weights and forward returns from the backtest.

The long leg is doing nearly all the work. Of the strategy’s total contribution, 84% comes from long positions and only 16% comes from shorts. The factor’s primary alpha source is identifying which chains are gaining capital, not which are losing it.

The alpha is concentrated in mid-cap L1s and L2s. The top five contributors are Polygon (+19.5% of total returns), Mantle (+18.8%), Optimism (+17.9%), BSC (+14.8%), and Sei (+13.0%). Together these five chains account for 84% of total strategy returns. They share a profile: chains in the $1-15B stablecoin supply range, mature enough to have meaningful flows but not so dominant that their cross-sectional signal is washed out by a massive base. None of them are in the small-newcomer tier (Sonic, Kaia) or the mega-cap tier (Ethereum, Tron). The factor is fundamentally a positive flow detector for the mid-cap segment.

L2s worked particularly well on this signal. Arbitrum, Optimism, and Mantle together represent 3 of 18 chains (17%) but contribute roughly 40% of total returns, almost entirely from the long side. The cross-sectional signal works particularly well on L2s, likely because L2 stablecoin growth tends to reflect genuine adoption (bridging from L1 to access cheaper transactions) rather than the trader-driven flows that dominate large L1s.

Factor Correlation

We wanted to ensure that Stablecoins 1 is capturing return variation not already covered by factors we already have. To evaluate this, we computed the pairwise correlation between the Stablecoins 1 (S1) and every other factor in our research stack: Fundamentals 1 (F1), SMB (size), Value, Momentum (Momen), Growth, Market (BTC).

Stablecoins 1 is the least-correlated factor in our entire stack. Its maximum absolute correlation with any other factor is 0.16 (with Value), which shows low pairwise correlation between factors we’ve already built out.

We also ran a spanning regression to make sure that Stablecoins 1 isn’t well-explained by a combination of other factors as well. We regressed Stablecoins 1’s weekly returns against all six other factors simultaneously and tested whether the intercept (alpha) is statistically different from zero. We found an alpha t-statistic of 2.54 (statistically significant at the 1% level), with an R² of just 6.1%. In other words, 94% of the Stablecoin Factor’s variance cannot be explained by any linear combination of our other factors. Whatever this factor is capturing, it is genuinely orthogonal to every other alpha source we have built.

When we tested multi-factor portfolios across our existing factors, Stablecoins 1 consistently appeared at 10-20% weight in all top-performing combinations. This shows that the factor is genuinely orthogonal and profitable: too small a weight to dominate, but always present.

Stablecoins 1 is not a substitute for any factor we already have; it’s not a repackaging of momentum, value, or size when you regress them out. It measures something different (capital migration between chains as a leading indicator of token returns) and it gets paid for measuring it.

Key Takeaways

• Stablecoin capital flows predict cross-sectional L1 returns. Stablecoins 1 delivers a 1.67 Sharpe with +83.6% annualized return over five years, with a conservative out-of-sample Sharpe estimate of 0.96 after a degrees-of-freedom haircut.

• Stablecoins 1 is the most market-independent alpha source we have built. Market beta is -0.03 with an R² of just 0.1%, and Stablecoins 1 is the least-correlated factor in our entire stack (max pairwise correlation 0.16). Annualized alpha is +73.8% (t-stat 3.31) for the raw factor after controlling for market exposure.

• It performs best in down markets. Across 30 BTC-negative months in the backtest, the raw factor averaged +6.8% per month while BTC averaged -10.9%. Stablecoins 1 doesn’t just survive crypto bear markets, it actively profits from them.

• The alpha comes from being long mid-cap chains. The long leg generates 84% of returns, and the top five contributors (Polygon, Mantle, Optimism, BSC, Sei) account for 84% of total returns.

• Stablecoins 1 is genuinely orthogonal to every other factor we have built. After controlling for all other factors simultaneously in a spanning regression, the stablecoin factor retains a t-statistic of 2.54 and an R² of just 6.1% meaning 94% of its variance is unique. Adding it to existing multi-factor portfolios consistently improves Sharpe.

• Performance has been resilient across multiple market regimes, with strong years in 2021 (+262%), 2022 (+47%), and 2025 (+315%). Stablecoins 1’s one losing year (2024, -13%) coincided with stagnant aggregate stablecoin supply growth, and performance recovered immediately when the ecosystem resumed expanding.

If you have any questions, feel free to reach out to team@artemis.xyz or shoot us a DM on X.com @Artemis.

Thesis Summary

SoFi Technologies is one of the largest neobanks in the United States and one of a handful of fintechs to hold a national bank charter, providing structural advantages in cost of funding, product breadth, and regulatory positioning. The company delivered its first $1 billion revenue quarter in Q4 2025, grew members 35% to 13.7 million, and achieved nine consecutive quarters of GAAP profitability. Financial Services and Technology Platform now represent 57% of revenue and are growing 61% year-over-year, driving rapid diversification away from historical lending dependence. CEO Noto’s expansion into crypto (SoFi USD stablecoin, crypto trading), business banking, and international payments via SoFi Pay opens multiple new growth vectors. At $15.85—down 42% year-to-date and 52% from the 52-week high—the stock trades at ~10.7x FY2026E adjusted EBITDA despite guiding for 30%+ revenue growth in FY2026. The average analyst price target of $25.70 implies 62% upside. The selloff appears driven by broad market risk-off (tariff fears, macro uncertainty) rather than fundamental deterioration, creating an asymmetric risk/reward profile.

2. Company Overview

Business Description

SoFi Technologies was founded in 2011 as a student loan refinancing company and has since transformed into a comprehensive digital financial services platform. Under CEO Anthony Noto, who joined in 2018, SoFi obtained a national bank charter from the OCC in January 2022—a pivotal strategic move that enabled the company to take deposits, fund lending at lower cost, and offer FDIC-insured accounts. Revenue has compounded at nearly 50% annually from $240 million in 2018 to $3.6 billion in 2025, while the member base grew from 650,000 to 13.7 million.

The company operates three reportable segments: Lending (personal loans, student loan refinancing, home loans), Financial Services (SoFi Money, SoFi Invest, SoFi Relay, SoFi Credit Card, SoFi Smart Card), and the Technology Platform (Galileo and Technisys, providing banking-as-a-service infrastructure to other fintechs and financial institutions globally). SoFi’s “one-stop shop” model enables members to borrow, save, spend, invest, and protect within a single integrated app, driving a 40% cross-buy rate and deepening multiproduct relationships.

In 2025, SoFi expanded aggressively into crypto—becoming the first nationally chartered bank to offer crypto trading, launching the SoFi USD stablecoin on public blockchain, and deploying SoFi Pay for international payments across 30+ countries. Business banking and institutional crypto services are planned for 2026, representing the next wave of product expansion. Nine sell-side analysts cover the stock with a consensus Buy/Hold rating and an average price target of $25.70.

Products & Services

• Personal Loans: SoFi’s largest lending product, targeting prime borrowers refinancing high-rate credit card debt. Per management, SoFi originates roughly 15% of total U.S. prime personal loan volume.

• Student Loan Refinancing: The company’s founding product. Per company disclosures, SoFi saved members over $400 million in interest on 2025 refinances alone. Addressable market: ~$400 billion.

• Home Loans: Record $3.4 billion in FY2025 originations, with Q4 running at a $4.5 billion annualized pace. Per management, 90% of SoFi members with mortgages currently hold them at other institutions.

• SoFi Money: High-yield checking/savings account with up to 4.50% APY and FDIC insurance. 97% of deposits are direct-deposit customers, indicating high-quality primary relationships.

• SoFi Invest: Brokerage platform offering stocks, ETFs, crypto, IPO access, private market funds, and Level 1 options. Revenue grew 2.2x year-over-year in 2025.

• SoFi Relay: Free credit monitoring and financial tracking dashboard serving as the top-of-funnel member acquisition tool.

• SoFi Smart Card: All-in-one debit/credit card with up to 5% grocery cashback. Per CEO Noto, built and launched in 4.5 months using the in-house tech platform.

• SoFi Crypto & SoFi USD: First national bank to offer crypto trading and issue a stablecoin. SoFi USD is backed 1:1 by cash at the Fed master account with zero credit, liquidity, or duration risk.

• SoFi Pay: Blockchain-powered international payments across 30+ countries including Mexico, India, Brazil, Philippines, and Europe.

• Galileo (Technology Platform): Card issuing, payments, and digital banking API platform powering other fintechs and financial institutions. Known as the “AWS of Fintech.”

• Technisys (Technology Platform): Cloud-native core banking platform enabling enterprise clients to modernize their banking systems. Expanding into Latin America and the Middle East.

Market Positioning

SoFi is one of the largest neobanks in the United States and one of the most comprehensive fintech platforms globally. Its national bank charter differentiates it from most digital-only competitors (e.g., Robinhood, Cash App, Chime) who rely on sponsor banks. This charter enables SoFi to offer higher APYs while maintaining healthy margins—deposits sit in an FDIC-insured account earning interest, whereas funds at non-bank competitors earn nothing while awaiting deployment. The combination of lending, banking, investing, and technology infrastructure under one roof is unique among publicly traded fintechs and positions SoFi at the intersection of multiple secular growth trends: digital banking adoption, crypto/blockchain infrastructure, and the unbundling of traditional financial services.

3. Financial Analysis

Revenue growth reaccelerated in FY2025 to 36% from 28% in FY2024, driven by the rapid scaling of Financial Services (which grew 88% year-over-year in Q4) and the loan platform business. The company crossed $1 billion in quarterly revenue for the first time in Q4 2025. Revenue has compounded at a nearly 50% CAGR since 2018, yet the company remains in the early stages of penetrating its addressable markets. Management has guided for 30%+ revenue growth in FY2026, with a long-term aspiration for sustained growth at similar rates, supported by expanding crypto services, business banking, and international expansion.

Adjusted EBITDA grew 58% year-over-year in FY2025 to $1.05 billion, with margins expanding from 25% to 29%. Q4 margins hit 31%, surpassing the company’s original long-term target of 30% set at IPO. The incremental EBITDA margin of ~41% demonstrates strong operating leverage. FY2025 GAAP net income of $481 million appears to decline from FY2024’s $499 million, but this is misleading: Q4 2024 included a ~$271 million non-cash DTA valuation allowance release that inflated FY2024 results. Stripping that out, underlying profitability improved significantly, as evidenced by net income margins expanding sequentially from 9.3% in Q1 to 17.0% in Q4 2025.

Stock-based compensation peaked at $306 million in FY2022 and has since moderated to approximately $250–$270 million annually, even as headcount and scope have expanded. More importantly, SBC as a percentage of revenue has declined sharply—from 24.5% in FY2021 (the SPAC year) to 7.3% in FY2025—as revenue growth has far outpaced compensation expense. The cumulative dilution remains significant: diluted shares outstanding have grown from 527 million post-SPAC to 1.25 billion, a 137% increase. However, the rate of dilution is slowing, and if absolute SBC remains flat while revenue scales toward $5 billion+, the SBC-to-revenue ratio should converge toward the 3–5% range typical of mature fintechs.

SoFi’s balance sheet strengthened dramatically in FY2025. The company raised $3.2 billion in new capital, growing tangible book value by $4.3 billion to $8.9 billion ($6.98 per share). Total debt was cut by 40% from $3.2 billion to $1.9 billion, flipping the company from a net debt to a $3.0 billion net cash position. Deposits grew 44% to $37.5 billion, now representing 95.1% of total funding—up from 89.1% in FY2024—replacing higher-cost warehouse borrowings. CEO Noto described this as a “fortress balance sheet” that provides “a broad range of optionality.”

SoFi maintains an exceptionally high net interest margin of 5.85%—well above traditional bank peers—reflecting its focus on higher-yielding personal and student loans funded by lower-cost deposits. The deposit franchise has scaled at a remarkable pace, growing from $7.3 billion (first full year post-charter) to $37.5 billion in three years. Critically, 97% of deposits come from direct deposit customers, indicating high-quality primary banking relationships that are naturally sticky. As CEO Noto noted, SoFi’s structural advantage in deposit pricing comes from its profitable lending business, which allows it to offer top-quartile APYs while still generating strong spreads.

4. Growth Drivers & Member Metrics

Unlike traditional banks that grow primarily through rate competition and branch expansion, SoFi’s growth engine is built on a digital flywheel: brand awareness drives member acquisition, product quality drives cross-buy, and cross-buy drives lifetime value and operating leverage. This section examines the key growth vectors and their trajectory.

SoFi added a record 1 million members in Q4 2025 alone—the first time the company achieved seven-figure quarterly member growth—bringing total membership to 13.7 million, up 35% year-over-year. Product adoption is accelerating even faster: 1.6 million new products were added in Q4, and the cross-buy rate reached 40%, up 7 percentage points year-over-year. Management attributes this to the “one-stop shop” strategy: a single integrated app where members can borrow, save, spend, and invest. The 2026 guidance targets at least 30% member growth, implying ~17.8 million members by year-end.

Fee-Based Revenue Diversification

One of the most significant structural shifts in SoFi’s business is the rapid growth of fee-based, capital-light revenue. Fee-based revenue hit a record $443 million in Q4, representing $1.8 billion on an annualized basis, up 50% from the prior year. This includes interchange, brokerage fees, referral fees, and loan platform business commissions. The loan platform business—where SoFi originates loans on behalf of third-party partners—generated $3.7 billion in originations in Q4 alone, earning fees without deploying balance sheet capital.

New Growth Vectors

• Crypto & Stablecoin: SoFi Crypto (first nationally chartered bank to offer consumer crypto trading), SoFi USD stablecoin (backed 1:1 by cash at Fed master account), and plans for institutional crypto trading, custody, and stablecoin-as-a-service for B2B clients.

• Business Banking: Launching in 2026. SoFi aims to be the bank for businesses transacting in both fiat and cryptocurrencies, filling a critical market gap. Will include institutional trading, digital asset custody, and 24/7 settlement via SoFi Exchange.

• International Expansion: SoFi Pay now covers 30+ countries. Medium-term plan to offer full SoFi Pay to non-U.S. users, serving as a global brand-building channel. Technology Platform (Galileo/Technisys) already expanding to Latin America and Middle East clients.

• AI & Product Innovation: SoFi launched its own Agentic AI ETF and is leveraging AI across product quality and marketing efficiency. Goal is to drive “escape velocity” in customer acquisition cost efficiency.

5. Valuation

Methodology

Primary Methods: Comparable Company Analysis using EV/EBITDA and P/TBV (important for bank stocks) as primary multiples, supplemented by a scenario-based valuation framework. For a bank holding company, Price-to-Tangible Book Value provides a fundamental floor, while EV/EBITDA captures the growth and operating leverage of the non-lending segments.

SoFi’s growth profile (+36% revenue growth) is second only to Upstart (+59%) and comparable to Affirm (+39%) within its peer set, but with significantly better profitability than either. The company’s net interest margin of 5.85% and 95% deposit-funded model distinguish it from non-bank fintechs. At a P/TBV of ~2.3x, SoFi trades at a meaningful discount to most fintech peers despite superior growth and an improving margin trajectory. The -42% YTD decline has compressed multiples to levels that appear disconnected from fundamentals: EV/FY2026E EBITDA of ~10.7x for a company guiding 30%+ growth is well below the typical growth fintech range of 15–25x.

The base case of $22.50 per share (+42% upside) assumes SoFi meets its FY2026 guidance of ~$1.6 billion adjusted EBITDA and the market rewards execution with a 16x forward multiple—below the peer median but reflecting fintech-bank hybrid discount. The bull case at 22x assumes margin upside from accelerating fee-based revenue and successful crypto/business banking launches. The bear case reflects a recession scenario where lending volumes decline and the stock de-rates to deep value bank multiples.

6. Bull / Bear Thesis

Bull Case

1. Financial Services Flywheel Reaching Escape Velocity: Financial Services and Technology Platform grew 61% year-over-year in Q4 and now represent 57% of revenue. Fee-based revenue hit a $1.8 billion annualized run rate. As these capital-light segments continue to scale, margins should expand toward 35%+ adjusted EBITDA and the market should re-rate SOFI closer to software/platform multiples rather than bank multiples.

2. Crypto First-Mover Advantage as a National Bank: SoFi is the first nationally chartered bank to offer crypto trading, issue a stablecoin, and deploy blockchain-powered payments. The combination of FDIC insurance + crypto infrastructure positions SoFi as the trusted bridge between traditional finance and digital assets. SoFi USD and business banking could unlock an entirely new B2B revenue stream as infrastructure provider for banks and fintechs.

3. Deposit Machine Creates Structural Funding Advantage: Deposits grew 44% to $37.5 billion and now fund 95% of liabilities. With 97% direct deposit penetration and a 4.50% APY that most competitors cannot match profitably, SoFi’s cost of funding should decline as rates fall while maintaining deposit growth. CEO Noto noted this advantage will “make itself more clear” as rates decline, as SoFi can maintain top-quartile APY funded by lending spreads.

4. Massive Insider Buying Signals Management Confidence: CEO Noto purchased $1 million in SOFI stock in March 2026 amid the selloff. Combined with 2026 guidance for 30%+ growth through 2028 and the company’s track record of meeting or beating guidance, insider buying at these levels signals conviction that the selloff is unwarranted.

5. Valuation Disconnected From Fundamentals: At $15.85, SOFI trades at 10.7x FY2026E adjusted EBITDA and 2.3x tangible book value for a company growing revenue 30%+, generating a Rule of 40 score of 68%, and with nine consecutive quarters of GAAP profitability. The average analyst price target of $25.70 implies 62% upside. Historical dislocation of this magnitude in high-quality growth fintechs has typically resolved favorably for investors.

Bear Case

6. Consumer Credit Deterioration in a Macro Downturn: SoFi’s $15+ billion personal loan portfolio is sensitive to employment and consumer spending. Tariff-driven inflation and a potential recession could increase default rates and force higher provisioning, compressing margins. The company’s limited track record through a full credit cycle creates uncertainty about loss severity.

7. Persistent Share Dilution From Stock-Based Compensation: Diluted shares outstanding have grown from 527 million post-SPAC to 1,252 million in FY2025—a 137% increase. While absolute SBC has moderated from its $306M peak to ~$262M and fallen from 24.5% to 7.3% of revenue, the cumulative dilution is substantial. Continued SBC to attract talent represents ongoing per-share value erosion that must be offset by earnings growth.

8. Lending Revenue Concentration and Rate Sensitivity: Despite diversification efforts, lending still represents 52% of revenue and is sensitive to interest rate movements. Rising rates increase funding costs and can dampen loan demand; falling rates too quickly compress net interest margins. The bank’s 5.85% NIM, while strong, could face pressure in a rapidly shifting rate environment.

9. Crypto Execution Risk: SoFi’s ambitious crypto strategy (stablecoin, business banking, institutional trading) introduces operational complexity and regulatory risk. The regulatory framework for bank-issued stablecoins is still evolving, and the Clarity Act has not yet been enacted. Crypto revenue is inherently cyclical and sentiment-driven.

10. Intense Competition Across All Segments: SoFi competes with well-capitalized incumbents in every segment: JPMorgan, Goldman Marcus in banking; Robinhood, Schwab in brokerage; LendingClub, Upstart in lending; Stripe, Marqeta in infrastructure. The “one-stop shop” model requires SoFi to be competitive in each category simultaneously.

Base Case

SoFi delivers on its FY2026 guidance of ~$4.65 billion in adjusted net revenue and ~$1.6 billion in adjusted EBITDA, supported by continued 30%+ member growth and expanding fee-based revenue. Crypto and business banking products launch but do not contribute materially until late 2026 or 2027. Credit quality remains manageable but provisions increase modestly due to macro uncertainty. The stock re-rates from current levels toward the $20–25 range as the market recognizes that the YTD selloff overreacted to macro fears. At 16x FY2026E EV/EBITDA, the base case price target is approximately $22.50—representing 42% upside from current levels.

7. Risk Factors

Company-Specific Risks

• Credit Quality: SoFi’s unsecured personal loan portfolio is the largest revenue driver and is sensitive to employment levels and consumer spending. A spike in unemployment could lead to higher-than-expected defaults.

• Stock-Based Compensation & Dilution: Diluted shares have grown 137% since FY2021 post-SPAC. SBC as a percentage of revenue has improved significantly (24.5% → 7.3%), but the cumulative dilution is already embedded in the share count. Continued SBC must be offset by earnings growth to create per-share value.

• Execution Across Multiple New Verticals: SoFi is simultaneously expanding into crypto, stablecoins, business banking, international payments, and new lending products. Managing multiple growth vectors increases operational complexity.

• FY2025 Net Income Anomaly: GAAP net income declined from $499M (FY2024) to $481M (FY2025) despite revenue growth, primarily because Q4 2024 included a ~$271M non-cash gain from releasing a deferred tax asset valuation allowance (adjusted Q4 2024 net income was only $61M). While the underlying profitability trend is improving, this creates an optically unfavorable Y/Y comparison.

Industry & Macro Risks

• Tariff-Driven Recession Risk: Escalating tariffs and trade uncertainty could trigger a consumer recession, reducing loan demand, increasing credit losses, and compressing valuations for the entire fintech sector.

• Interest Rate Sensitivity: SoFi’s bank model is sensitive to rate movements. Rapidly falling rates could compress NIM; rising rates could dampen loan demand and increase funding costs.

• Regulatory Burden: As a bank holding company, SoFi faces heightened regulatory oversight (OCC, Fed, FDIC). Consumer lending regulations could increase compliance costs or restrict product offerings.

Concentration Risks

• Revenue: 52% from lending products; Financial Services (43%) growing rapidly and approaching parity

• Geographic: 100% United States (international expansion via SoFi Pay and Galileo is nascent)

• Product: Personal loans represent the single largest revenue contributor; student loans subject to federal policy changes (OBBB legislation)

• Platform: Technology Platform client concentration; loss of a major Galileo client could impact segment revenue

8. Scoring Framework

Disclaimer: This report is produced by Artemis for informational and educational purposes only. It does not constitute investment advice, a recommendation, or a solicitation to purchase, sell, or hold any security. The analysis, scores, and outlook labels contained herein reflect the research team’s assessment based on publicly available information at the time of publication and are subject to change without notice. Readers should conduct their own due diligence and consult with a qualified financial advisor before making any investment decisions. Artemis, its affiliates, and its analysts may hold positions in the securities discussed in this report. Past performance is not indicative of future results. The information in this report is believed to be accurate but is provided “as is” without warranty of any kind. Artemis disclaims all liability for any errors, omissions, or losses arising from reliance on this report.

© 2026 Artemis. All rights reserved.

Conclusion:

The market is pricing Circle as a rate-sensitive money market fund — a bet on the Fed Funds Rate sitting on blockchain rails. We think this framing misprices the business. USDC supply grew 72% in 2025 to $75.3 billion even as the Fed cut rates 75bps over the second half of the year, demonstrating that USDC demand is driven by genuine utility adoption rather than pure yield-seeking behavior. Our base case projects the total stablecoin market reaching ~$1.5 trillion by 2030, with average USDC supply at $284B billion. Even with reserve yields expected to compress, we expect Circle’s reserve income to grow to $9.2 billion in 2030 (~3.5x from 2025) as supply growth overwhelms rate compression. Combined with Circle Payments Network (CPN) scaling to $350 million in revenue and distribution costs declining from 60% to 55%, our base case projects total revenue of $9.8 billion and net income of ~$1.8 billion by 2030.

Several tailwinds are supporting this trajectory: the GENIUS Act has created a federal stablecoin framework favoring compliant issuers; Circle Payments Network is gaining early traction with 55 financial institutions enrolled and $5.7 billion in annualized TPV, providing a transaction-based revenue stream that diversifies away from rate sensitivity; and stablecoin adoption is broadening across B2B payments, cross-border settlement, and DeFi. Our base case produces 2030E EPS of $6.73, implying a price target of ~$168 at a 25x terminal P/E, representing 83% upside from current levels.

Comps Table:

There is no direct public comp for a stablecoin issuer monetizing reserve float. Our comp set spans companies that share key attributes of Circle’s business: float-based revenue models (Schwab, Interactive Brokers), digital payments infrastructure (PayPal, Wise, dLocal, Bill.com), crypto-native platforms (Coinbase), and high-growth infrastructure with usage-based economics (Snowflake, Confluent).

What Does Circle Do?

Circle is the issuer of USDC, a dollar-denominated stablecoin pegged 1:1 to the US dollar. When a user deposits USD, USDC is minted; when they redeem, it is burned. The reserves (approximately 43% reverse repos, 43% T-bills, and 14% bank deposits, custodied at BNY Mellon and managed via BlackRock’s USDXX fund) generate yield that constitutes Circle’s primary revenue.

The critical cost structure nuance: Coinbase, as USDC’s primary distribution partner, receives 100% of reserve income on USDC held on its platform and 50% on off-platform USDC. In 2025, Coinbase took $1.35 billion, 51% of Circle’s gross reserve income. Including non-Coinbase distribution (12.7%), total distribution costs consumed ~61% of reserve income, leaving a 39% gross margin. We model distribution costs declining from 60% to 55% by 2030 as non-Coinbase distribution grows, and new financial institutions, banks, and custodian partners negotiate deals more favorable than Circle’s current one with Coinbase. This drives gross margin expansion from 39% to 54% over the projection period.

Beyond reserve income, Circle’s most important growth lever is Circle Payments Network (CPN), a cross-border B2B settlement network built on USDC. CPN launched in May 2025 and has enrolled 55 financial institutions with $5.7 billion in annualized TPV and a 500-FI pipeline. We model CPN scaling to $175 billion in TPV by 2030 at a 0.2% take rate (aligned with the 20bps blended cross-border rate), generating $350 million in transaction-based revenue. This revenue is rate-insensitive and diversifies Circle away from pure reserve yield dependency. Additional revenue lines (referred to as “Other revenue” in our model) include CCTP (47–50% of cross-chain bridge volume) and Arc settlement infrastructure, which we model at $207 million combined by 2030E.

Thesis #1: Supply Growth Overwhelms Rate Compression

The total stablecoin market expanded from ~$137 billion in 2022 to ~$308 billion in 2025. Our model projects ~$1.5 trillion by 2030, a ~37% CAGR. Today, total stablecoins in circulation (~$316B) represent roughly 1.4% of the $22.7 trillion US M2 money supply. Our base case implies ~6%, still a modest share of dollar-denominated liquidity.

We predict USDC maintains 22–25% market share (modestly declining from 24.8% as white-label and bank stablecoins fragment the space), producing $338 billion in USDC supply by 2030 (~4.5x increase from today). Put simply, even if Circle’s effective reserve yields fall, the sheer growth in USDC supply — from $63 billion to $284 billion on average — more than compensates. As a result, reserve income rises 3.5x, from $2.64 billion to $9.24 billion.

At the current stock price, a reverse-DCF implies the market is discounting ~$120–150 billion of USDC supply by 2030, which is roughly a 60–100% increase from today. Our base case models a 350% increase. That gap is the core of the investment case.

Thesis #2: Agentic Commerce Will Drive the Next Wave of Stablecoin Demand

AI agents are on a trajectory to autonomously execute transactions by 2030. McKinsey projects $3–5 trillion in global agentic commerce sales by 2030; Gartner estimates AI agents will intermediate over $15 trillion in B2B purchases by 2028. These transactions structurally require stablecoin rails:

Stablecoins are becoming the settlement layer for this emerging agent economy, and Circle’s business model scales with it. As agents hold USDC in wallets to fund autonomous transactions, Circle earns yield on every dollar sitting in those reserves. The larger the pool of agent-held USDC, the larger the revenue base, regardless of transaction frequency.

USDC is already the default stablecoin for agentic payments. In the six months since the x402 payment standard (HTTP-native micropayments) gained traction, it has processed ~$106 million in volume across ~177 million transactions. Over 99.6% of this volume has been settled in USDC.

The first-mover advantage creates a flywheel where new builders default to USDC-support because it has the deepest integrations, which deepens the integrations further, which makes it harder for alternatives to break in. We do not model agentic revenue in the base case but agentic demand is embedded as upside optionality in our bull scenario. If even 1–2% of McKinsey's low-end $3 trillion projection settles on USDC rails, that implies $30–60 billion in incremental USDC float in agent wallets, all of which Circle may earn passive yield on.

Valuation & Scenarios

We value CRCL using a terminal P/E on 2030E EPS. Our base case produces net income of $1.84 billion on 273.9 million diluted shares, yielding EPS of $6.73. A 25x terminal P/E — a premium to the comp-weighted average, reflecting Circle’s structural growth trajectory, CPN-driven revenue diversification, and regulatory moat— implies ~$168 per share in 2030, representing 83% upside from current levels.

The 25x multiple sits between JPMorgan (JPM) at ~15x and Coinbase at ~38x, appropriate for a high-growth infrastructure business transitioning toward recurring, rate-insensitive revenue.

Base Case: Assumes continued execution on supply growth and CPN scaling, with the stablecoin market reaching $1.5 trillion and USDC maintaining 22.5% share. Distribution costs decline modestly to 55% as new financial institution partners negotiate lower revenue shares. Exit at a 25x terminal P/E on 2030E earnings, implying a price target of $168.34 — 82.7%% upside with a 16.3% IRR.

Bull Case: Assumes accelerated stablecoin adoption driven by favorable regulation, CPN network effects, and widespread TradFi onboarding. Total stablecoin market reaches $2.3 trillion with USDC capturing 30% share. Distribution costs compress to 50% as non-Coinbase origination scales. Exit at a 35x terminal P/E on 2030E earnings, implying a price target of $482.10 — over 423% upside with a 51.2% IRR.

Bear Case: Assumes stablecoin adoption slows, white-label stablecoins erode USDC market share to 20%, and rate cuts compress reserve yields to 2.75%. CPN traction disappoints. Exit at a 15x terminal P/E on 2030E earnings, implying a price target of $46.92 — approximately 49% downside with a -15.5% IRR.

Management

We view management quality as above average for the crypto infrastructure space, with particular strength in regulatory navigation (49 state MTLs, first-mover MiCA compliance).

• Jeremy Allaire co-founded Circle in 2013 and serves as Chairman and CEO. A serial entrepreneur (previously CTO of Macromedia, founder/CEO of Brightcove, IPO 2012), Allaire pivoted Circle from a consumer payments app to stablecoin infrastructure, launched USDC with Coinbase in 2018, navigated a failed SPAC in 2022, and completed a traditional IPO on the NYSE in June 2025.

• Heath Tarbert serves as President, elevated from Chief Legal Officer in January 2025. Tarbert is the former Chairman and CEO of the CFTC (2019-2021), former Assistant Secretary of the U.S. Treasury, and former CLO of Citadel Securities.

• Jeremy Fox-Geen has served as CFO since January 2021. Previously CFO of iStar/Safehold (NYSE-listed REITs) and CFO of McKinsey & Company’s North American business. He oversaw Circle’s IPO and manages the USDC reserve architecture supporting $70B+ in circulation.

• Dante Disparte serves as Chief Strategy Officer and Head of Global Policy and Operations. Previously founding executive and Vice Chairman of the Diem Association (Meta’s stablecoin project), he leads global regulatory strategy, public policy, market expansion, and international operations.

Primary management risk is founder concentration and elevated post-IPO SBC (over $500M in 2025, including $424M in IPO-related RSU acceleration), which is already normalizing (Q3 and Q4 2025 SBC ran at $59M and $48M respectively, trending toward a sub-$200M annual run-rate).

Risks

White-Label & Platform-Native Stablecoins

The most underappreciated risk to USDC’s market share is platforms, major applications, and financial institutions launching their own branded stablecoins. For example, Hyperliquid has USDH, PayPal has PYUSD, Fidelity has FIDD, and JPMorgan has JPMD. And recently, Polymarket recently introduced “Polymarket USD,” which is a USDC wrapper today, but likely a stepping stone toward independent settlement. If this playbook scales under the GENIUS Act framework, USDC could slowly lose its position as the default settlement rail. Our base case models USDC market share declining from 24.8% to 22.5% by 2030 to reflect this fragmentation.

Mitigant: White-label stablecoins still need reserve infrastructure, compliance, and — most importantly — deep liquidity. Given USDC is integrated across every major exchange, wallet, DeFi protocol, and bridge, a new branded stablecoin would need to replicate that liquidity network before it could function as a standalone settlement token. Deep liquidity pools, tight spreads, and instant redeemability are not easy to bootstrap, and fragmented stablecoins with thin liquidity create worse execution for users. The switching costs to bootstrap fully independent reserves are high enough that most platforms may never complete the transition.

Federal Funds Rate Sensitivity

Reserve income is directly tied to interest rates. Every 100bps of cuts on $284B average USDC (2030E) equates to ~$2.8B in lost gross reserve income. If the Fed cuts to 2.0%, 2030E reserve income falls 25–30% versus our base case. Kalshi prediction markets currently price a 63% probability of further cuts before 2027.

Mitigant: Even at a 2.5% yield, $284B in average USDC produces $7.1B in reserve income, which is still 2.7x the $2.64B earned in 2025 at 4.19%. Supply growth overwhelms all but the most extreme rate scenarios.

Single-Product Concentration & Coinbase Dependency

USDC reserve income was over 96% of 2025 revenue. Coinbase controls ~67% of US crypto exchange share and takes 51% of reserve income. As previously mentioned, Coinbase launches its own stablecoin, renegotiates terms aggressively, or if regulatory headwinds slow USDC supply growth, the entire revenue base is at risk.

Mitigant 1: Given that Coinbase earns $1.35B annually from their arrangement with effectively zero balance sheet risk, it seems unlikely that they would choose to launch a competing stablecoin. If they did, it would require Coinbase to build regulatory infrastructure and liquidity that Circle spent years constructing.

Mitigant 2: The market made a similar critique of Visa for years (that they was a single-product business), but Visa’s value-added services generated over $10.9 billion in 2025 (up 24% YoY), demonstrating its reduced dependence on interchange fees. We see CPN as the key diversification lever for Circle. By end of 2030, we project CPN will generate $350 million in transaction-based revenue (~4% of total), which is both rate-insensitive and independent of the Coinbase relationship. Over time, institutional and B2B USDC origination that bypasses Coinbase should also organically reduce blended distribution costs.

Tether Resilience & Competitive Landscape

USDT currently has nearly 2.5x the supply of USDC, and Tether is actively closing the regulatory gap that USDC has capitalized on. In January 2026, Tether launched USAT, a GENIUS Act–compliant stablecoin issued through Anchorage Digital Bank (OCC-regulated), giving Tether a path into the US institutional market it was previously locked out of. If Tether successfully runs a dual strategy (USDT for global liquidity, USAT for US compliance), USDC’s regulatory moat narrows considerably.

Mitigant: The competitive landscape is nuanced. USDT dominates centralized exchange trading outside the US and emerging-market remittances, while USDC dominates DeFi collateral (default on Aave, Compound, Uniswap), US institutional adoption, cross-chain bridging (CCTP at 47–50% of bridge volume), and B2B payments ($235B in 2025, +733% YoY, with USDC capturing ~65%). These are effectively different products serving different TAMs. With that said, our thesis is built upon total stablecoin market expansion rather than marketshare gains at Tether’s expense. Both stablecoins will grow considerably.

Disclosure: This material is provided for informational purposes only and does not constitute investment advice, financial advice, trading advice, or any other form of advice. The views expressed are those of the authors and should not be relied upon as a recommendation to buy, sell, or hold any asset. The authors or affiliated entities may hold positions in the assets discussed. You should conduct your own research and consult appropriate financial professionals before making any investment decisions.

• Prediction markets turn uncertainty into a tradable asset, where prices reflect financially backed probabilities rather than opinions.

• The industry is scaling rapidly, but growth today is driven by sports, not the high value information markets that define its long-term thesis.

• Crypto rails remove geographic and settlement constraints, enabling global participation and improving signal quality through scale.

• Regulation determines whether liquidity concentrates into deep global markets or fragments into weaker local ones.

• There is a growing gap between what prediction markets are becoming (information infrastructure) and how they generate revenue today.

Prediction markets generated $63.5 billion in volume in 2025.

More than three times what they processed the year before. In the first 86 days of 2026, Kalshi and Polymarket alone combined for $52.7 billion, $28.3 billion on Kalshi and $24.3 billion on Polymarket (Artemis). Annualize that and you get $223 billion. The industry tripled last year and it is on pace to triple again.

Three years ago, prediction markets processed less than $100 million a month. Today they process that amount in hours. Every event – whether it’s a Federal Reserve decision, election, geopolitical escalation, or earnings outcome – can now be expressed as a market with a price. That price isn’t a forecast, but a financially backed probability that is continuously updated as new information enters the system.

Polls collect opinions, models process historical data, and analysts publish views - all of them suffer from the same structural flaw: there is no cost to being wrong, which makes accuracy effectively optional. Prediction markets get rid of that flaw by attaching financial consequences to forecasts, so that being right results in profit while being wrong leads to a direct monetary loss.

The Federal Reserve itself confirmed that prediction markets can produce more accurate and more responsive forecasts than traditional methods. In a January 2026 research paper, the Fed found that Kalshi’s prediction markets were a “statistically significant improvement” over Bloomberg consensus for forecasting headline CPI. The mode of Kalshi’s distribution perfectly matched the realized federal funds rate by the day of every FOMC meeting since 2022 (Federal Reserve FEDS 2026-010).

Why Now

Prediction markets have existed for decades. The Iowa Electronic Markets, for example, launched in 1988. Another platform called Intrade ran for years before collapsing in 2013. The underlying idea traces back even further to Friedrich Hayek, who argued that markets are the most efficient mechanism for aggregating dispersed information. While the idea was always compelling, the execution never worked due to structural constraints: limited distribution, fragmented liquidity, and slow, localized settlement. All of these have rapidly improved.

Distribution is native

Prediction markets are now being embedded directly into brokerages, media platforms, and APIs. Robinhood added event contracts alongside equities and options. Media platforms like CNN are now starting to display probabilities next to headlines and every news cycle becomes immediately actionable.

Acting on a macro view once required opening accounts, funding them, and waiting for market hours, but that friction has largely disappeared. Prediction markets are becoming a default interface for interacting with or trading on uncertainty.

Settlement is global

The second shift is more important. On traditional financial rails, prediction markets are constrained by geography, regulation, and settlement speed. But on crypto rails, those constraints disappear.

A user in New York, Lagos, or Jakarta can see the same market, the same price, and the same opportunity and make trades with effectively instant settlement times.

Prediction markets become more accurate as participation increases. Restricting markets to a single country produces structurally inferior signals. Expanding them globally improves the quality of the output. Crypto does not just make prediction markets more efficient, but makes them more truthful.

The Regulatory Dynamic

Regulation has legitimized prediction markets, but it has also exposed a clear economic conflict. Prediction markets sit at the intersection of finance and betting, and do not fit cleanly into either category. That mismatch is where the tension comes from.

This conflict is most visible in sports. States have a direct stake in sports betting. The US market processed $165 billion in 2025, generating billions in tax revenue (SportsHandle). That system was built through a state-by-state licensing process, where operators like DraftKings and FanDuel pay significant fees, partner with casinos, and operate under high tax rates (Legal Sports Report).

Prediction markets overlap with that same category but operate under a different framework. From the state’s perspective, prediction markets divert activity away from taxed sportsbooks and bypass a regulatory system that took years to build. This is why sportsbooks are pushing back. Not because prediction markets are flawed, but because they disrupt the economics of an already regulated industry.

At the same time, the CFTC is moving to regulate prediction markets, while courts decide the limits of its authority. This tension determines the end state for prediction markets. If they are pushed into a state-by-state model, liquidity will fragment. However, if they’re allowed to operate under a unified financial framework, liquidity will concentrate and prediction markets will scale into a global system for pricing uncertainty.

The Structural Risk

Today, most of the revenue in prediction markets comes from sports.

In Kalshi, sports account for roughly 83% of total notional volume. Polymarket is more diversified, but even there, sports dominates accounting for 38.4% of total notional volume. (Artemis)

This creates a clear mismatch. Current growth is driven by sports markets, which anchor perception around betting and is the one that is most exposed to regulatory pressure. As a result, the revenue base today looks much closer to sports betting than financial infrastructure.

That concentration introduces a timing risk. If regulation tightens around sports before other categories reach scale, the industry’s primary source of revenue slows before its long-term model is fully built.

Valuations, however, are clearly pricing in that future. Kalshi at $22 billion and Polymarket targeting a similar range imply a market far larger than sports betting alone, one that expands into institutional grade information markets. That transition hasn’t happened yet, which is the central tension: prediction markets are being valued as information infrastructure, but today they are still monetized like sports products.

Sports markets provide growth, but that’s not the long term moat. The real value lies elsewhere.

Where Prediction Markets Actually Win

The underlying product is real and powerful in various other categories that go beyond sports.

Politics

The 2024 election was the proof. Polymarket’s Presidential Election Winner market generated $3.7 billion in total trading volume. On the Monday before the election, Polymarket had Trump at 58% vs Harris at 42%, while traditional polls showed a dead heat (CNN). The market was right. A peer-reviewed study from IMDEA Networks Institute, analyzing 86 million bets, found Polymarket is accurate more than 94% of the time a full month before an outcome is definitively known (arXiv:2603.03136).

Prediction markets don’t replace polling but incorporate it. Polls become inputs, while prices reflect the aggregation of all information under incentives where being wrong costs money.

Economics and policy

The most important category, and where the real moat lives.

Kalshi now provides intraday trade data for macroeconomic indicators including CPI (month over month, year over year, calendar year), Core CPI, PCE inflation, unemployment, payroll releases, GDP growth, and recession probability.

The Fed’s own research validates the product. Economics and technology markets surged by 905% and 1,637% respectively in 2025. While the volumes are small today, the signal quality is already institutional grade. A market-implied probability of a rate hike is not just a data point, but it is also a signal that influences portfolio allocation, risk management, and strategic decision making at the highest levels of finance.This is where prediction markets transition from consumer product to financial infrastructure.

Why Crypto Rails Matter

The transition from consumer product to financial infrastructure depends on scale and scales with access.

Kalshi operates within the United States, a market of roughly 330 million people. Polymarket operates globally, using crypto rails to onboard users anywhere with an internet connection and a wallet. This difference changes the size of the market entirely.

A user in New York, Lagos, or Jakarta can access the same market at the same price. Participation is permissionless and distribution is effectively global. This is why crypto is a distribution advantage. Crypto also changes how markets are created. Instead of relying on centralized operators, independent resolvers and oracle systems can define and settle outcomes, allowing markets to scale far beyond what any single platform could support.

Trading venues are converging toward this model. Robinhood integrated event contracts alongside equities, options, and crypto. Around the same time, Hyperliquid launched prediction market-style contracts through HIP-4.

Market Landscape

While there are many participants, the market structure has been a duopoly from the start. Kalshi and Polymarket control 97.5% of volume. What is changing is not consolidation but stratification: a clear stack is forming around these two platforms.

The Super-App Disruption

Platforms like Robinhood and Coinbase are no longer just distribution layers. They are moving into the exchange layer itself.

Robinhood has already demonstrated the demand. Prediction markets became one of its fastest-growing products, with over 12 billion contracts traded in 2025 and more than one million users. It also accounts for a majority of Kalshi’s volume which creates a clear incentive to internalize that flow. In January 2026, Robinhood acquired MIAXdx, a CFTC-licensed exchange and clearinghouse, through a joint venture with Susquehanna..

Coinbase is following the same playbook. It first integrated prediction markets through Kalshi contracts, then moved to build its own infrastructure by acquiring The Clearing Company. The strategy is straightforward: aggregate demand first, then vertically integrate. If this works, the existing market structure changes. Kalshi’s positioning relies on being the regulated exchange layer, but that advantage weakens if distribution platforms control both the user and the venue. The moat shifts from licensing to distribution.

Polymarket faces a different constraint. After exiting the US market following a $1.4 million CFTC settlement in 2022, it scaled globally on crypto rails. Its re-entry through QCEX introduces a different product, one that is more restricted, broker-routed, and fully KYCed compared to its permissionless global offering. As a result, the product that succeeded internationally is not the same one competing in the US, where incumbents already control distribution and funding rails.

This leads to a likely separation of roles. Distribution platforms control users and order flow, exchanges provide infrastructure and compliance, and crypto-native platforms dominate global access. Prediction markets continue to grow across all three layers, but value capture concentrates at the point closest to the user.

Winners and Losers

The emergence of this system creates a new hierarchy of winners.

Distribution platforms win.
Platforms like Robinhood and Coinbase sit directly in front of the user. They control onboarding, capital, and order flow. As they vertically integrate into exchange infrastructure, they capture both volume and margin. They are not just participants in prediction markets. They are becoming the primary interface.

Infrastructure providers survive, but capture less value.
Kalshi is best positioned to become the regulated backbone of the system, licensing contracts, clearing trades, and providing compliance infrastructure. But its role shifts from owning the user to servicing those who do. The model looks closer to Visa than a consumer bank.

Global crypto-native platforms remain differentiated.
Polymarket retains an advantage outside the United States, where permissionless access, instant settlement, and broader market coverage matter more than regulatory compliance. It becomes the default venue for global users, controversial markets, and crypto-native flows.

The losers are those without distribution or liquidity.
Traditional polling organizations lose relevance as financially-backed probabilities replace opinion-based forecasts. Smaller prediction market platforms struggle to compete without liquidity or embedded distribution. Sportsbooks face structural pressure as prediction markets increasingly overlap with their core product, already contributing to an estimated $600 million in lost tax revenue.

The End State

Prediction markets expand the scope of what financial markets can price. Today, markets price assets: equities represent ownership in companies, bonds represent claims on future cash flows. Prediction markets price outcomes.

Anything uncertain can become a market: political events, economic indicators, technological developments, environmental outcomes. The result is a system where uncertainty itself becomes financialized.

The numbers make the trajectory clear. 2024: ~$16 billion total volume, proof of concept via elections. 2025: ~$63.5 billion, 4x growth as sports drives scale. 2026 run rate: $200 billion+, regulatory clarity becomes existential (Artemis).

Conclusion

Prediction markets represent a shift in how the world processes uncertainty. They take something intangible, the likelihood of future events, and turn it into something that can be priced, traded, and acted upon.

The question is whether the thing that makes them valuable, pricing uncertainty, is the same thing that makes them scale. Right now, the answer is no. Sports make them scale, but information makes them valuable. The gap between those two facts is where the entire industry’s future will be decided. At scale, this is not just a new market, but a new way of pricing information itself.

Disclaimer: The authors of this content, as well as affiliates of Artemis Analytics, may have financial interests in the equities or tokens mentioned. This does not constitute investment advice or a recommendation to buy, sell, or hold any asset. The information provided is for educational purposes only and should not be relied upon for financial, legal, or tax decisions. Readers should assess their own circumstances before making any financial choices. Views expressed may change without notice, and Artemis Analytics is not liable for any losses resulting from the use of this content.*

Executive Summary

• March was defined by the US-Israel war with Iran, which began February 28 and led to the closure of the Strait of Hormuz, disrupting 20% of global oil supplies and driving energy prices up 42%. Despite this, crypto was oddly resilient: the market factor fell just -1.81% while the S&P 500 dropped over 7%, suggesting much of the forced selling had already been flushed out in February’s -23.49% drawdown.

• Value remains the strongest YTD performer among long-short factors at +10.33%, benefiting from its defensive profile as overvalued tokens in the short leg sold off harder than fundamentally cheap longs.

• Even in a month where all five factors posted flat or negative returns, the long-short framework continues to significantly limit losses relative to passive exposure. YTD, the market factor is down -26.36% and BTC is down -24.9%, while Growth (-4.24%), Fundamentals 1 (-6.08%), and Momentum (-7.14%) have all preserved substantially more capital. The framework’s value is most apparent in drawdowns, not just in months where factors generate positive alpha.

• The AI narrative was one of the few pockets of strength: NEAR rallied +25.32% after launching its “Confidential Intents” privacy upgrade and positioning itself as blockchain infrastructure for AI agents. TAO also surged, punishing short positions in SMB (-81.69%) and Fundamentals 1 (-73.81%).

• DEXE was the single most impactful long position across factors, appearing as the top winner in both SMB and Momentum at +117.93%. On the short side, DeFi lending tokens AAVE, KMNO, and MORPHO all declined as a hawkish Fed hold at 3.5-3.75%, an unusual Aave liquidation event, and token unlock dilution pressured the sector.

Market Context

Sentiment around the past month has been mostly cautious as the war between US and Iran progressed. On February 28, the US and Israel launched surprise airstrikes on Iran, killing Supreme Leader Ali Khamenei. Iran retaliated with drone strikes and fighting has continued between the two countries. The Strait of Hormuz closed, disrupting 20% of global oil supplies, which drove energy and oil prices up 42% in March. The IEA agreed to release a record 400 million barrels of crude oil in response.

Despite the chaos, crypto was oddly resilient relative to equities. The crypto market failed to recover, but also didn’t dip further. The market fell -1.81% over March and BTC fell -2.39% over the month. Meanwhile, S&P fell over 7%, which could indicate that we’re at a bottom for the crypto market.

For a more in depth look into how each of the strategies were built, see our Crypto Factor Model Analysis and Analysis of Fundamentals 1 strategy. For a complementary guide on reading through this report, see our factors guide.

Performance Summary

All metrics are calculated from inception (Feb 20, 2022).

Despite the broad stagnation of the market, the multi-factor framework continues to demonstrate its value. Value was near-flat in March (+0.02% return) and remains the strongest YTD performer among the long-short factors at +10.33%. The long-short construction across all factors limited drawdowns relative to long-only crypto exposure.

Market Risk Factor

Construction

Long-only portfolio of the top 10 assets by market cap, market cap-weighted, rebalanced weekly. Captures broad crypto market exposure similar to a crypto index.

Current Holdings

BTC (71.7%), ETH (13.0%), BNB (4.5%), XRP (4.4%), SOL (2.5%), TRX (1.7%), DOGE (0.8%), BCH (0.5%), HYPE (0.5%), ADA (0.5%)

Performance Metrics

March was an uneventful month for the broad market factor, returning -1.81%. This was much less severe than February’s -23.49% drawdown, but the market remains -49.33% below its all-time high, and YTD losses stand at -26.36%. BTC continues to dominate the portfolio composition at 71.89% weight and its -2.39% return was the primary driver of the monthly decline.

Top 5 Winners & Losers for March

HYPE was the standout performer at +30.58%, building on momentum of the exchange’s growing volumes and ecosystem expansion. TRX (+10.73%) continued to benefit from the stablecoin narrative as a primary rail for USDT transfers. ETH was essentially flat at +1.48%. On the losing side, BCH (-20.77%) led declines followed by ADA (-11.55%) and DOGE (-5.11%). Every asset in the top 5 losers is a long-only holding and losses were concentrated in the smaller-weight assets. BTC was the 5th largest loser, but at 71.89% weight, was the dominant contributor to the overall negative return.

Size Factor

Construction

Equal-weighted long-short factor: long the smallest 50% of eligible assets by market cap, short the largest 50%. Minimum 40 assets (20 per leg). Breakpoint: 50th percentile of market cap.

Performance Metrics

SMB returned -5.48% in March, making it the worst performing long-short factor for the month. YTD losses are now -13.70% and the max drawdown deepened to -38.98%. The continued underperformance of small caps relative to large caps is consistent with a risk-off environment where liquidity-driven selloffs hurt smaller, less liquid assets more. However, the factor’s annualized return of 40.31% and Sortino ratio of 2.72 indicate the size premium still remains over the full sample.

Top 5 Winners & Losers for March

DEXE (+117.93% as a long) was by far the top contributor, driven by a surge in its governance token demand and staking activity. FET (+43.84% as a long) benefited from continued AI narrative momentum. WLD (+28.94% as a short) and ATOM (+26.97% as a short) contributed as short winners, reflecting weakness in large-cap infrastructure and interoperability tokens. On the losing side, TAO (-81.69% as a short) was the largest loss as the AI-focused token rallied sharply in March. MYX (-57.20% as a long) continued its trend from last month as a volatile position across factors. IP (-51.35% as a long) fell as token unlocks scheduled for March created supply overhang concerns, but these unlocks were later delayed to August. Additionally, PIP Labs, the core development team behind Story Protocol, reduced its staff by about 10% in early March 2026 as part of a broader wave of crypto layoffs. LIT (-42.85% as a long) is a low-liquidity micro-cap that gets pushed around by broad market moves, so was hurt by the general risk-off sentiment in March.

Value Factor

Construction

Equal-weighted long-short: long assets with the lowest Market Cap / Fees ratio (highest fundamental value), short assets with the highest MC / Fees ratio (most expensive). Minimum 30 assets (15 per leg).

Performance Metrics

Value was effectively flat in March, returning +0.02%. Despite this, the factor still has the best risk profile of all factors with a max drawdown of just -1.26% and YTD return of +10.33%. This continues from the pattern last month where Value benefits from risk-off environments as the speculative, expensive assets in the short leg sell off, offsetting any weakness in fundamentally cheap long positions.

Top 5 Winners & Losers for March

The short leg benefited from continued declines in overvalued tokens like IP (+38.30% as a short), which fell amid token unlock fears and a 10% staff reduction at its core dev team PIP Labs, and WLD (28.94% as a short) and ATOM (26.97% as a short), which continued to sell off as large-cap infrastructure tokens lost favor in the risk-off environment. Meanwhile, the long leg was dragged down by the broader DeFi selloff: MYX collapsed on thin liquidity, LIT (-42.85% as a long) continued its decline as a micro-cap token in the midst of a rebrand with negligible trading volume, and HNT (-24.14% as a long) and INJ (-20.26% as a long) fell alongside the broader altcoin drawdown driven by the Iran war shock and hawkish Fed hold.

Momentum Factor

Construction

Volatility-adjusted momentum: calculates 3-week rolling Sharpe-like score (mean return / std dev of daily return) for each asset. Long top 25% by filtered momentum, short bottom 25%. Equal-weighted, minimum 30 assets (15 per leg).

Performance Metrics

Momentum returned -1.26% in March, a reversal from February’s strong +11.27%. The factor’s YTD loss is -7.14% and the max drawdown from ATH is -7.14%. The negative monthly return reflects that trend persistence was less rewarded and reversals in positions offset gains in the short leg. Despite the underperformance this month, the factor still has the highest annualized return (71.11%) and a Sharpe ratio of 1.43 since inception.

Top 5 Winners & Losers for March

DEXE (+117.93% as a long) was the top contributor, with the momentum signal correctly identifying its strong uptrend. MYX (+57.20% as a short) was the second-best performer, with the factor correctly shorting the token as it continued to decline. IP (+51.35% as a short) and LIT (+39.39% as a short) also contributed as short winners. On the losing side, SNX (-26.02% as a long) was the biggest loss as the token initially had positive momentum that reversed sharply. TRUMP (-21.09% as a long) also hurt, entering the portfolio on a brief momentum signal before declining. FARTCOIN (-20.57% as a short) and S (-20.31% as a short) were short losers that rallied against the factor’s positioning, while HNT (-20.02% as a long) pulled back after its strong February performance.

Growth

Construction

Equal-weighted long-short: composite score combining z-scores of 2-week trailing fee growth (pct change) and DAU growth (pct change). Long top 50% composite score, short bottom 50%. Eligibility requires DAU > 100 and weekly fees > $500. 30 assets (15 per leg).

Performance Metrics

Growth returned -1.60% in March, bringing its YTD return to -4.24%. This factor continues to exhibit the lowest volatility of the long-short factors at 22.84% annualized. The negative monthly return reflects fundamental growth signals (fee growth and DAU growth) providing less differentiation between assets as the market unease compressed factor spreads.

Top 5 Winners & Losers for March

NEAR (+25.32% as a long) was the top contributor, with the growth signal correctly identifying accelerating on-chain activity. NEAR benefited from the AI narrative momentum as on March 3, co-founder Illia Polosukhin shared a vision of NEAR enabling AI agents on the blockchain and the protocol released a privacy upgrade called “Confidential Intents”. AAVE (+18.99% as a short) contributed as a short winner following an unusual liquidation event on March 10 due to a configuration problem in CAPO and broad DeFi selloff due to a hawkish Fed and the Iran War. KMNO (+18.77% as a short) and MORPHO (+16.46% as a short) also contributed from the short leg as part of the DeFi sell-off. HYPE (+15.89% as a long) confirmed its strong growth profile. On the losing side, SEI (-30.01% as a long) was the largest loser despite screening for positive growth metrics; it declined sharply alongside broader L2 weakness. ZRO (-19.20% as a short) was the most notable short loser as the price rallied following an announcement of its own Layer-1 blockchain backed by Citadel Securities, ARK Invest, and Google Cloud.

Fundamentals 1

Construction

Equal-weighted long-short quintile portfolio: composite score combining rank z-scores of DAU growth, inverted active revenue share, revenue stability, and MC/fees mean reversion. Long top quintile (~12 assets), short bottom quintile (~12 assets).

Performance Metrics

Fundamentals 1 returned -3.85% in March, bringing its YTD to -6.08%. Despite the difficult month, the factor retains the highest Sharpe ratio (1.68) and second-highest annualized return (56.0%) of any factor since inception. The max drawdown from ATH deepened to -13.89%, reflecting that the factor was actively exposed to the market throughout March. The factor’s composite signal, which combines DAU growth, revenue quality, revenue stability, and valuation mean reversion, struggled to generate positive spread in a month where geopolitical shocks overwhelmed fundamental differentiation.

BERA (+33.13% as a short) was the top contributor to returns. Berachain is now trading 97% below its all-time high as its Proof-of-Liquidity model has struggled to generate sustained demand, and the risk-off environment accelerated the selloff. ALGO (+26.11% as a long) rallied after the SEC declared it a digital commodity, resolving years of regulatory uncertainty. CHZ (+15.77% as a long) benefited from momentum around the 2026 FIFA World Cup as Chiliz released its roadmap targeting omnichain fan tokens and CHZ buybacks. RSR (+14.81% as a short) continued to slide amid low adoption and broader altcoin weakness. NEAR (+13.96% as a long) was lifted by the AI narrative after its “Confidential Intents” launch and co-founder Illia Polosukhin’s vision of NEAR as infrastructure for AI agents. On the losing side, TAO (-73.81% as a short) was by far the largest loser as the AI narrative rallied. RENDER (-39.97% as a short) similarly rallied on the decentralized GPU computing narrative. MYX (-21.22% as a long) collapsed on thin liquidity, SEI (-21.08% as a long) declined alongside broader L1 weakness despite positive fundamental scores, and LIT (-18.23% as a long) fell as a micro-cap with negligible volume in the midst of rebranding to Heima.

Key Findings & Conclusion

March was a quiet month for crypto factor strategies, with the market returning -1.81% and no long-short factor generating meaningful positive returns. Factor spreads compressed as the Iran war shock drove correlated moves across assets, making it difficult for relative value signals to differentiate winners from losers.

Value continued to demonstrate its defensive properties, returning +0.02% and maintaining one of the lowest max drawdowns from ATH (-1.26%) across the factors. While the remaining long-short factors all posted negative returns in March, the YTD picture shows their value: the market factor is down -26.36% and BTC is down -24.9%, but Growth (-4.24%), Momentum (-7.14%), and Fundamentals 1 (-6.08%) have all significantly limited losses relative to passive crypto exposure. Even SMB, the worst-performing factor YTD at -13.70%, has cut the market’s drawdown roughly in half.

A few cross-factor themes stood out in March. The AI narrative drove selective strength: NEAR rallied on its Confidential Intents launch and AI agent vision, while TAO surged sharply, punishing short positions in multiple factors. DeFi lending was broadly weak: AAVE fell after a CAPO oracle misconfiguration triggered unusual liquidations, and both KMNO and MORPHO declined on token unlock pressure and the broader sector selloff. DEXE was the strongest single asset across the factor universe at +117.93%, appearing as the top contributor in both SMB and Momentum. Meanwhile, ZRO (LayerZero) showed how different signals can diverge on the same token: it rallied after announcing its own L1 blockchain backed by Citadel, ARK Invest, and Google Cloud, making it a winner in Value (correctly long) but a loser in Growth (incorrectly short).

In a month where broad crypto exposure lost -1.81% and the S&P 500 fell over 7%, the long-short factor framework limited losses across the board. The framework’s value is clearest in cumulative YTD numbers: while the market factor is down -26.36%, Value is up +10.33%.

If you have any questions, feel free to reach out to team@artemis.xyz or shoot us a DM on X.com at https://x.com/artemis

Disclosure: This material is provided for informational purposes only and does not constitute investment advice, financial advice, trading advice, or any other form of advice. The views expressed are those of the authors and should not be relied upon as a recommendation to buy, sell, or hold any asset. The authors or affiliated entities may hold positions in the assets discussed. You should conduct your own research and consult appropriate financial professionals before making any investment decisions.

Market Overview: The Weekly Recap

Oil and geopolitics dominated this holiday-shortened week (markets closed Friday for Good Friday). WTI crude surged above $112/bbl after Trump’s Wednesday address failed to outline an endgame for the Iran conflict, though equities showed resilience. The S&P 500 eked out a small weekly gain (+0.3%, closing at 6,583), its first since the war began, while the Nasdaq snapped a five-week losing streak despite remaining in correction territory (>10% below its record high). Gold pulled back ~2.8% to ~$4,676/oz on de-escalation signals, continuing its underperformance as a safe haven during this conflict.

Crypto was weaker. BTC fell ~4% to ~$66,900, ETH dropped ~5% to ~$2,057, and SOL slid ~8% to ~$79 (Solana took an extra hit from the $285M Drift exploit on April 1). HYPE held up relatively well given the upcoming unlock overhang, and XRP caught a bid on continued CLARITY Act optimism.

Today We Highlight:

• Drift Protocol’s $285M Exploit — Solana’s largest-ever DeFi hack was a combination of a governance failure, social engineering, and a Solana feature called “durable nonces” weaponized into a week-long setup.

• Coinbase Wins Conditional OCC Trust Charter — A federally regulated crypto custodian is no longer hypothetical.

• HYPE’s First Major Token Unlock (April 6) — 9.92M HYPE (~2.7% of circulating supply) hitting the market.

• CLARITY Act Stablecoin Yield: “99% Resolved” — The compromise bans passive yield, permits activity-based rewards.

• Artemis Analyst Sector Deep-Dives: Agentic Payments, Lending, Trading, and Neobanks

1. Drift Protocol’s $285M Exploit

On April 1, 2026 – in what Drift’s own team acknowledged was “not an April Fool’s joke” – an attacker drained ~$285 million from the Drift Protocol on Solana. This was the largest DeFi exploit YTD and one of the biggest in crypto history. The DRIFT token crashed immediately.

What makes this exploit distinct is the attack vector (this was not a smart contract vulnerability). The attacker exploited a combination of Solana’s durable nonce feature, weak multisig governance, and social engineering, a playbook much harder to audit away than a reentrancy bug.

How It Worked

Solana has a feature called durable nonces that lets transactions be pre-signed and executed later. The attacker exploited this to stage the entire heist over a week before draining funds in under 12 minutes.

On March 27, Drift migrated to a new 2-of-5 multisig with no timelock. The attacker tricked two signers into approving what appeared to be routine transactions, gaining admin control. From there, they listed a fake token (CarbonVote Token), wash-traded it to trick oracles into treating it as real collateral worth ~$785M, then executed 31 rapid withdrawals. They drained USDC, JLP, and other tokens across multiple vaults.

Why This Matters

The Drift exploit surfaces a class of risk that smart contract audits cannot catch: operational security and governance design failures. Drift’s security setup only required 2-of-5 keyholders to approve transactions, meaning the attacker only needed to compromise two people to gain instant, full control. And because durable nonces let signed transactions sit dormant for weeks before execution, the attack was effectively invisible to monitoring tools until it was too late.

For Solana DeFi broadly, this is a confidence hit. TVL across Solana protocols is likely to face redemption pressure in the near term, and expect every major protocol to re-examine its multisig setup and governance upgrade process in the coming weeks.

2. Coinbase Wins Conditional OCC Trust Charter

On April 2, Coinbase received conditional approval from the Office of the Comptroller of the Currency (OCC) for a national trust company charter – the most significant regulatory milestone for a crypto-native company since the SEC/CFTC unified guidance in March.

The entity, Coinbase National Trust Company, will be headquartered in New York and operate as a federally regulated digital asset custodian once it meets the OCC’s preopening conditions. It’s important to clarify that Coinbase is not becoming a bank. It will not take retail deposits, will not engage in fractional reserve banking, and will not be FDIC-insured. This is a non-insured national trust company focused on custody, safekeeping, and fiduciary services.

What does Conditional Approval Mean?

“Conditional” means Coinbase still has to build out compliance, pass regulatory reviews, and satisfy AML requirements before final approval (a process that typically takes 12 to 18 months). However, the signal matters more than the timeline. A national trust charter replaces Coinbase’s current patchwork of 50+ state licenses with a single federal regulator (the OCC), and positions it as a qualified custodian under SEC rules. This designation is a requirement that institutional allocators need before they’ll park assets with you.

The Strategic Play

The trust charter is about Coinbase’s revenue diversification away from trading fees:

• Institutional custody: A federally chartered trust company is a stronger counterparty for pension funds, endowments, and sovereign wealth funds considering digital asset allocations.

• Stablecoin issuance: If finalized, the charter opens a potential path to issuing stablecoins and tokenized securities under federal supervision.

• CLARITY Act positioning: A national trust charter is complementary to the CLARITY Act framework, alluding to the fact that Coinbase is building the institutional plumbing in anticipation of the regulatory structure.

Coinbase is now one of eleven companies in various stages of the OCC trust charter process, a pipeline that barely existed six months ago. The race for a federal crypto banking license is accelerating.

3. HYPE Token Unlock (April 6): Headline Fear vs. Actual Supply Impact

On April 6, the next tranche of Hyperliquid’s core contributor vesting will become eligible for claim. The vesting schedule allows up to 9.92 million HYPE (~$357M at ~$36) but the Hyper Foundation has committed to distributing only ~330,000 HYPE (~$12M), consistent with its pattern of claiming just 1-18% of the theoretical max each month. The headline number overstates the actual supply impact by roughly 30x.

HYPE is down over 7% this week to ~$35. HIP-3 open interest has hit all-time highs, with commodities, equities, and indices now making up >80% of volume, led by the S&P 500 perpetual contract via trade[XYZ] and S&P Global.

Three factors suggest the unlock won’t meaningfully move supply:

1. Protocol revenue runs at $676M–$843M annualized, with 97% flowing into HYPE buybacks and burns – dwarfing the ~$12M committed unlock.

2. Core contributors are incentivized to hold. Aggressively selling would crater the value of their remaining vesting allocation.

3. Five months of unlock data show the team consistently claims a small fraction of the eligible amount, giving the market a reliable baseline.

4. CLARITY Act Stablecoin Yield: “99% Resolved”

Senators Thom Tillis (R-NC) and Angela Alsobrooks (D-MD) confirmed an agreement in principle on stablecoin rewards, backed by the White House. Senator Lummis described it as “99% resolved.” The framework bans passive yield (i.e., earning returns simply for holding a stablecoin) but permits activity-based rewards tied to payments, transfers, and wallet usage.

Who Wins, Who Loses

Circle ($CRCL): More nuanced than the initial 22% stock drop suggested. The bill bans retail passthrough of yield, not issuer-to-distributor payments. Circle pays Coinbase over $900M/year in revenue share – that relationship is untouched. The bull case for them is if Coinbase keeps collecting from Circle but can’t share it with users. This means the distributor margins on USDC would actually improve and Coinbase would further incentivize them to push USDC adoption. However, the bear case is that, without yield as a selling point, retail has less reason to hold USDC. ARK bought $16M of CRCL on the dip.

Tether/USDT: Largely unaffected – Tether doesn’t pass yield to holders. The framework arguably legitimizes its model.

DeFi protocols: The language targets “digital asset service providers,” leaving it unclear whether decentralized lending protocols like Aave or Morpho are in scope. Expect this to be litigated.

Charts of the Week

Polymarket Daily Fees Reach $1.9 Million

Kalshi Volume Extends Run With 8 Straight Monthly ATHs

USDD Stablecoin Supply Reaches $1 BILLION

Coinbase Perp Volumes Surge, Reaching 81.4B in Q1

Other Notable News

• Aave launches v4 on Ethereum after two years of development, expanding DeFi into real-world credit and improved lending usability

• Ethereum’s Glamsterdam upgrade targets June 2026 — the most significant technical overhaul since the Merge, with gas limit increasing from 60M to 200M per block and throughput scaling to 10,000 TPS

• RLUSD (Ripple’s stablecoin) surpasses $1B in market cap within its first year, as regulated stablecoin issuers gain share in the institutionalization era

• Solana hits 10 billion cumulative transactions while Ethereum reaches 200 million — different scales reflecting different architectures and use cases

• Oil spikes above $113/bbl WTI as Strait of Hormuz disruption fears intensify — the largest supply disruption in the history of the global oil market per the IEA

Thanks for reading! Stay ahead this week by using the Artemis Terminal or pulling live data with =ART() in Excel.

Disclaimer: The authors of this content, as well as affiliates of Artemis Analytics, may have financial interests in the protocols or tokens mentioned. This does not constitute investment advice or a recommendation to buy, sell, or hold any asset. The information provided is for educational purposes only and should not be relied upon for financial, legal, or tax decisions. Readers should assess their own circumstances before making any financial choices. Views expressed may change without notice, and Artemis Analytics is not liable for any losses resulting from the use of this content.

• Coinbase filed for a national trust charter.

• NuBank received conditional approval from the OCC to establish a US national bank.

• PayPal applied to create PayPal Bank

• Revolut, after years of regulatory limbo, obtained its full UK banking license and is pursuing a US national banking charter.

• Kraken became the first crypto company to secure a Fed master account.

Four of the world’s largest fintechs made the same bet at the same time.

For the past 15 years, the story of neobanking has been the unbundling of the financial services stack. In the US, the playbook was simple: excel at one thing and hand off the regulatory complexity to sponsor banks. SoFi excelled at student loans, PayPal at online payments, Kraken and Coinbase at crypto trading. Now they are rebundling and going after the full stack.

Meanwhile, a second shift is underway. In 2025, there was $33 trillion in adjusted stablecoin transfer volume on public blockchains, rivaling major payment systems like ACH and Visa (Artemis). Decentralized finance lending protocols like Aave have surpassed $1 trillion in cumulative loans originated. NYSE, NASDAQ, and CME are partnering with crypto firms to build tokenization infrastructure.

The neobank stack is undergoing a dual compression that is changing what it means to be a bank. From the top, neobanks are vertically integrating by acquiring charters, and collapsing the middleware layers that have historically collected rent. From the bottom, public blockchains are exposing permissionless financial plumbing that lets any developer build money movement and financial products without a banking relationship. The neobanks that own both the legacy stack and the emergent blockchain stack will define the next decade of finance.

In our 2030 Digital Finance Manifesto, we introduced Steven, a 25-year-old whose entire financial life runs through a neobank: investing, saving, earning yield, receiving payments, and managing credit, all in one place. We identified neobanks as one of five pillars of digital finance. This is our deep dive into that pillar.

The first trillion-dollar standalone neobank is within reach.

The Stack

What is a neobank?

A neobank offers banking products without being a bank (at least historically). They partner with sponsor banks for deposits, FDIC insurance, and payment network access, and build the customer-facing product on top. This lets them move fast, but it also means the most valuable economics of the banking relationship sit with the sponsor, not the neobank.

Every existing legacy bank is vertically integrated across the stack, or at least significant portions of it. Neobanks stripped it apart and rebuilt each component on modern infrastructure. Marqeta focused on card issuance, Alloy on KYC, and Plaid on account connectivity. This created a modular ecosystem, where new players could pick and choose their stack as if they were picking ingredients off a shelf. We’ll break down the stack into 6 layers:

Layer 1: Foundations

This layer is the bedrock on which everything else is built. This is the layer that holds assets and does the ledgering. In the legacy financial system, companies in this layer follow the most stringent regulations, but they also have access to the very core of the financial system: the Fed and the payment networks. In the modern stack, blockchains play this role as the core ledgering system that holds financial assets.

Traditionally, a neobank seeking to take on customer deposits would partner with a real chartered bank, known as a sponsor bank. The sponsor manages the regulatory overhead, the relationship with the FDIC for deposit insurance, as well as access to payment networks. Since the deposits sit on the sponsor bank’s balance sheet, they can use them to fund loans and generate net interest income. That’s interest income neobanks are foregoing as rent.

Now, the biggest neobanks are bypassing sponsor banks entirely and going after their own full banking charter in order to own the economics themselves. At a certain scale, the rental relationship stops making economic sense.

NuBank, already a chartered bank in Brazil, received conditional approval for a US national banking charter by the OCC in January 2026, while Revolut applied for a full banking charter in March 2026 (immediately after receiving its UK banking charter). SoFi has been a chartered bank since 2022 and the effect on its business has been dramatic, growing quarterly net interest income from $94.9M to $617M in four years.

Crypto companies are also pushing into this layer: Kraken has a trust charter and a Fed master account, and Coinbase is going after its own trust charter. As Jason Mikula of Fintech Business Weekly put it: “what it means to be a bank is rapidly changing” (Substack).

Layer 2: Settlement Rails

This layer glues the foundations together. These rails enable banks to coordinate and settle assets between them, and are where real value movement happens. In the US, there are several key systems that settle and move value:

• FedWire, a real-time gross settlement system (only during business hours)

• FedNow, a new, 24/7/365 public-sector instant payment system

• DTCC, infrastructure for securities clearing, trading, settlement and more

• Swift, a messaging system for global inter-bank payments

Every one of these networks is gated. Connecting to FedWire, requires having a Master Account at the Federal Reserve, which means you have to be a chartered bank or trust. Touching the ACH network, requires being a member of NACHA, which requires being a federally insured depository institution.

Public blockchains show up again in this layer. They are the first financial rails in history that are public and permissionless, allowing any developer to plug into them. If you want to send $10,000 worth of USDC on Ethereum, you call a smart contract function. You don’t need a charter or a sponsor bank. Modern blockchains settle in milliseconds, never shut down, and charge sub-cent fees.

Layer 3: Abstraction

The raw rails are where value moves, but they are complicated and often gated. Most companies don’t want to get bank charters in order to integrate directly with Fedwire, and they don’t want to manage Ethereum node infrastructure. The abstraction layer exists to solve this. These are the companies that sit on top of the rails and make them usable.

For card payments: Marqeta and Lithic provide card issuing infrastructure. You tell them what kind of card you want, and they handle the Visa/Mastercard integration. Stripe and Adyen offer a single interface that abstracts away the complexity of moving money across ACH, card networks, and more.

For blockchains: Bridge (acquired by Stripe for $1.1 billion in February 2025) lets businesses accept stablecoin payments without managing wallets. Zero Hash provides trading, custody, and settlement infrastructure across 80+ digital assets. Alchemy, Infura, and QuickNode all offer APIs and tools that enable developers to build on blockchains without worrying about managing nodes.

Unit, Treasury Prime, and Synctera sit under Banking-as-a-Service, middleware that connects non-bank fintechs to sponsor banks. A now infamous BaaS provider, Synapse, went bankrupt in April of 2024 after internal ledgers failed to reconcile with their partner bank ledgers, resulting in up to $95 million unaccounted for or frozen across more than 100,000 customers (banking.senate.gov). This incident didn’t kill BaaS, but it did show the limits of the model and led to increased regulation in this part of the stack such as the “Synapse Rule” (American Banker). It signaled to major neobanks that they shouldn’t trust their entire deposit base to a middleware layer out of their control.

Brokerage and clearing firms like Apex, Alpaca and Drivewealth enable neobanks to offer securities trading without having to build their own clearing and settlement infrastructure. For example, Coinbase and Cash App use Apex, while Revolut uses Drivewealth. One logo stands out here: Robinhood. They decided to vertically integrate and own the economics instead of paying a fee to a third party provider for a service so core to their business.

Layer 4: Compliance and Identity

Every financial transaction requires knowing who’s on the other end. KYC (know your customer), AML (anti-money laundering), and fraud prevention form an entire industry.

On the traditional side, Alloy, Sardine, Persona and others provide identity verification, transaction monitoring, and fraud detection. These companies sit between the application and the financial products, ensuring every neobank customer is who they say they are.

On the blockchain side, firms trace onchain transactions to flag illicit activity. As stablecoins become more regulated, onchain compliance becomes more important — you need to know that the USDC flowing through your neobank didn’t originate from a North Korean wallet.

This layer isn’t where the value accrues, but these firms provide a critical service to their clients, ensuring a more safe, compliant and trustworthy financial system.

Layer 5: Product Infrastructure

This is the layer that answers the question: what products can neobanks plug into so they don’t have to build them from scratch?

Money movement companies allow neobanks to embed payments and other money movement features directly into their products. For example, a neobank looking to issue a stablecoin would use Bridge, while a neobank looking to offer international transfers could work with Airwallex or Sphere.

Savings, Yield and Investment is where DeFi shines. Neobanks can use Aave or Morpho on the backend to generate yield, no loanbook required. Or, they could offer a yield via staking products like Lido or Jito, which expose users to the native issuance and fee rewards of the Ethereum and Solana networks. Uniswap and Jupiter provide APIs enabling trading of thousands of assets including tokens, perps, and soon, RWAs. The US Treasury shows up in this layer as they are the issuers of risk-free rate assets.

Lastly, DeFi protocols like Aave and Morpho don’t just offer yield, they allow users to borrow against their capital. Coinbase launched crypto backed loans using Morpho and bitcoin backed mortgages with Better. Maple offers institutional grade borrowing and lending facilities. Figure offers white label lending infrastructure that apps and partners can plug into.

Layer 6: The Application Layer

This is where neobanks thrive and where the customer actually interfaces with the stack described above. The application layer breaks loosely into three categories, though the lines are blurring fast.

The Full-Stack Neobanks are Revolut, Robinhood, Coinbase, Kraken, SoFi, and Slash. These are all platforms that offer a suite of products that spans both the traditional financial stack and the blockchain stack. SoFi has a banking charter and is now issuing its own stablecoin SoFiUSD. Coinbase and Kraken each have their own public blockchain and are integrating more deeply with the US financial system via charters and Fed master accounts.

In the Fiat-First sector we have Nubank, Chime, Monzo, N26, Starling, Dave, and Mercury. These started from traditional banking products and are expanding through traditional rails and products.

Lastly, there are the Crypto-Native Financial Apps like EtherFi, RedotPay, Kast, and Phantom. These are firms born from public, permissionless blockchains, that take advantage of the lower operating costs that stablecoins provide. They integrate with onchain financial products, allowing users to pay in stablecoins, access DeFi yields, and trade onchain assets.

The pattern emerging across all three sectors is rebundling. Every successful neobank started with an initial wedge, and is expanding toward a single surface that handles a user’s entire financial life. The winning app is the one that is able to offer both the best of traditional banking and the emerging global onchain economy.

Business Models

Renting the Stack

Under the traditional model, a neobank partners with a sponsor bank and issues cards through a program manager, which structurally limits its ability to capture core banking economics. A traditional neobank’s primary revenue source is interchange, the 1-2% fee merchants pay on card transactions. The neobank may add a subscription tier (Chime’s SpotMe, Dave’s membership fee) or earn some referral fees, but interchange is the core. This business model allows neobanks to get off the ground, but it doesn’t scale into high-margin, durable businesses. Interchange is a thin margin that scales linearly with transaction volume. The most lucrative parts of the financial relationship are the net interest margin (NIM) on deposits, the lending income, and the investment yield. All of those economics flow to the sponsor bank. The neobank has the customer relationship, but not the balance sheet or the economics.

Owning the Stack

The economics change significantly when a neobank gets its own charter. A chartered bank can take customer deposits and use them to fund loans, or invest in interest-bearing assets, while keeping the spread. A traditional incumbent bank today typically offers less than 0.5% yield on savings accounts while making 6% on a mortgage and pocketing the 5.5% difference (FDIC). Since obtaining its bank charter in 2022, SoFi has grown net interest income tenfold, from $252M in 2021, to $2.2B in 2025 (Artemis).

A chartered bank also gains access to the core payment and settlement systems in the US. It connects directly to Fedwire, ACH, FedNow, avoiding per-transaction fees that intermediaries typically charge.

Stablecoin-First Banks

Neobanks that build using blockchains and stablecoins as the backend will have a few ways to generate revenue.

The first and most powerful is generating net interest income, a model reserved for stablecoin issuers like Circle, or SoFi. The second is interchange. All neobanks will have cards, and they will collect interchange on payments. Cards are a consumer’s favorite payment modality, and that isn’t changing any time soon. Stablecoin-first neobanks can also charge fees on trades, or take a portion of the DeFi yield they generate on the backend. The design space for a stablecoin neobank is wide open, which is why hundreds of millions of funding dollars have poured into backing startups in the space in the past year.

The Blockchain Thesis

The Stablecoin Inflection

Stablecoins are at an inflection point. They are the first use case of blockchains that has found non-speculative product market fit. The proof is in the numbers: stablecoin payment volumes doubled in 2025 to $390 billion annualized, with key sectors like B2B payments growing 733% YoY (Artemis, McKinsey). While stablecoin payments still represent < 1% market penetration in key payment verticals, the growth is undeniable. 2025 was also a watershed year for stablecoins from a regulatory perspective. Passage of the GENIUS Act provided stablecoin issuers a regulatory framework for operating in the US. The result is that even though crypto market cap has fallen about 50% since 2025 highs, stablecoin supply has actually continued to grow, reaching $313bn by March 2026 (Artemis).

Stablecoins bring dollars onto blockchains. Those dollars will seek to be productive, flowing into decentralized finance protocols to earn yield or provide liquidity.

For years, decentralized finance was the territory of crypto-natives exploring the frontier of the onchain economy. While the world looked the other way, they created composable building blocks for global financial products. Aave has facilitated >$1 trillion in loans, making it the de facto place to borrow against your crypto. Morpho has built the infrastructure for curated, risk-isolated lending vaults that have attracted institutional capital, including a partnership with Apollo Global. Yield-bearing stablecoins like Sky’s sUSDS, Ondo’s USDY, and Ethena’s sUSDe grew 80% in 2025, outpacing growth of non-yield-bearing stablecoins (Artemis). These tokens allow holders to earn yield generated by lending activities, treasury yields or trading strategies.

Neobanks that build using stablecoins instantly get a vast set of financial products that they can simply plug into with a few smart contract calls. Public, open source protocols mean developers don’t have to negotiate with anyone, or sign any contracts to integrate a product. This creates a major opportunity for builders attacking the stack from below.

Where This Is All Going: The Neobank of 2030

Charter Convergence

In the next 5 years, any neobank hoping to win at scale will either hold its own banking charter or be in the process of acquiring one. Beyond a certain scale, the economic gains far outweigh the risks(recall what happened to Synapse). More importantly, the regulatory environment has been constructive in recent years. The OCC is approving bank/trust charters (NuBank), the Fed is opening master accounts (Kraken), and state regulators are competing to attract fintechs (Wyoming SPDI).

That said, the sponsor bank model will not die. Column and Lead Bank provide a modern version of a useful and necessary service. But for neobanks scaling to tens of millions of users, the economic loss is too significant not to go after a charter.

Stablecoin-Native Accounts Become the Default

By 2030, the default neobank will offer stablecoin powered digital wallets, savings and yield powered by DeFi protocols, and exposure to tokenized real world assets. Neobanks will bridge the legacy and onchain worlds. Paychecks will arrive by ACH and automatically be converted to USDC where it will earn the risk free rate.

This is Steven’s world from the Manifesto. His neobank is the single financial surface that aggregates investing, saving, getting yield, obtaining credit, and sending money globally. The money moves on whatever rail makes sense: ACH for the landlord, USDC on Solana for paying an international employee, stablecoin on Morpho for savings yield.

By 2030, the distinction between a “bank account” and a “crypto wallet” will have dissolved, leaving in its wake a far superior banking product.

The First Trillion-Dollar Neobank

NuBank is valued at $70B, Revolut $75B, and Stripe $159B. J.P. Morgan, the world’s most valuable bank, has a market cap of about $800B, ten times that of NuBank or Revolut. JPM’s business model is meaningfully different including physical branches, corporate banking, investment banking, asset management. But the comparison is illuminating. A full-stack neobank with NuBank’s efficiency ratio (<20%), Revolut’s product breadth, and Coinbase’s blockchain integration (stablecoin yield, onchain rails, DeFi access) would have a revenue-per-user and margin profile that traditional banks cannot match.

Winners and Losers

Winners:

• NuBank: they are already a leader in LatAm with 130m users and have conditional approval to establish a US bank

• Revolut: established leader in the EU expanding in the US with a bank charter application

• Robinhood: leading the wave of retail investing and financial literacy, their distribution is undeniable, and their product velocity is to be envied

• Ethereum: More crypto assets by value sit on Ethereum than any other chain, it has strong Lindy effects, it seems every major exchange is building an Ethereum L2 (Ink, Base, Robinhood Chain)

• Solana: Second most value locked in DeFi applications, third most value in RWAs and stablecoins, significantly more performant than Ethereum

• Morpho: Indications that it is the protocol that institutions are choosing for onchain asset management and vaults

• Circle and Tether: defacto stablecoin winners due to existing distribution and liquidity

• The Federal Reserve System: stablecoins will generate more demand for US dollars

Losers:

• Cross-border remittance providers like MoneyGram or WesternUnion will have margins crushed

• Correspondent banks will lose billions in revenue as margins for cross border payments

• Middleware dependent neobanks that never invested in their own infrastructure

Who to watch:

• Column: chartered bank built from the ground up for developers

• Lead Bank: charter + modern infrastructure

• Coinbase: could be a winner with differentiated blockchain-based products

• Kraken: challenger in the US, Fed master account, coming after Coinbase

• Stripe: Going all in on blockchain tech with Bridge + Privy + Tempo

• Phantom: already have millions of users for their wallet product, expanding into neobanking services, wallet is very trading focused

• Hyperliquid: Will have the deepest onchain liquidity and best asset offering, enabling neobanks to offer any trading functionality on top

Conclusion

The neobanking stack is compressing from two sides. From the top, neobanks are seeking charters and collapsing the layers between them and the rails that move value. From the bottom, blockchains are removing layers of traditional intermediaries and exposing the bare metal rails to any developer.

Neobanks will drop the “neo” and just become “banks”, while blockchains and stablecoins allow anyone to build a global, self-custodial financial services product at a much lower cost. Assets and capital must flow between the legacy and blockchain financial systems. The banks that win will be those that straddle both worlds, capturing more users, creating better, stickier products and building the first trillion dollar bank.

Explore the full Artemis Neobank Market Map to see every player across all six layers.

To get in touch, please reach out to @x0wes on X, or team-all@artemisanalytics.com

Disclaimer: The authors of this content, as well as affiliates of Artemis Analytics, may have financial interests in the equities or tokens mentioned. This does not constitute investment advice or a recommendation to buy, sell, or hold any asset. The information provided is for educational purposes only and should not be relied upon for financial, legal, or tax decisions. Readers should assess their own circumstances before making any financial choices. Views expressed may change without notice, and Artemis Analytics is not liable for any losses resulting from the use of this content.*

• An entire generation has done the math on passive investing and decided it’s not enough. Financial anxiety will replace greed as the primary driver of trading.

• The gap between forming a market opinion and trading on it is collapsing to zero.

• AI will narrow the knowledge gap between retail and institutions, but the median active trader will continue to underperform the market.

• Winners: Robinhood, Coinbase, IBKR

• Losers: Legacy brokers selling patience to a generation that wants speed

The Math

$1,200 a month.

8% annual returns.

24 years.

That’s what it takes to hit a million dollars starting from zero at age 26. That puts you at 50.

Open your calculator app and run the numbers yourself.

You’re making $90,000 a year, but after taxes, New York City rent and student loans, you can realistically only save $1,200 a month. Put that in the market index fund for 8% annually and you’ll reach $220,000 in ten years – in twenty, about $710,000. If you’re consistent and don’t hit any setbacks, you might reach a million at 50.

That’s the base case if you do everything right.

Then you check your phone. Your college roommate posted last week about exercising $800,000 in startup options. Your cousin bought a house in 2019 and it’s up 60%. A 24-year-old kid on X you’ve never met turned $10,000 into $400,000 on memecoins in a few weeks. You know these are outliers and most people don’t post their losses. That doesn’t matter. It still feels like you’re falling behind.

So you start trading.

This is the Anxiety Economy – an era where financial anxiety becomes the primary demand driver of an entire category of products. It’s built for people who’ve done the math and realized that playing it safe won’t get them to where they want to be.

The consensus explanation for why trading volumes will grow – better products, lower fees, more asset classes, more AI tools, and better user experience. All true, but incomplete.

The deeper driver is emotional. An entire industry is being built to serve that feeling: I am behind and I need a faster way to catch up.

The Proof

You’re not imagining things.

Since 2000, the S&P 500 index has returned around 400%, home prices have risen approximately 230%, but median wages in the United States have only grown about 110%. If you were in the market, your wealth compounded. If you weren’t, you fell further behind.

The top 10% of American households now own 87% of all US equities, up from 82% in 1989 and the wealth gap is accelerating.

A recent study conducted by Amerisleep found that 49% of Americans lie awake at night due to financial stress. For Gen Z, that number is 69%. When saving and investing doesn’t feel safe, people start gambling.

And yet, passive investing works.

If you’re in the market, you get the market’s 8% annual returns. The problem isn’t the math but where most people start. The bottom half of American households own less than 1% of all equities. You’re compounding 8% on $15,000 while someone else is compounding 8% on $1.5 million. Same returns. Completely different lives.

This is why the old advice sounds right and empty at the same time.

“Be patient.”

“Buy the index.”
“Dollar-cost-average.”

“Let time do the work.”

It’s good advice for building wealth slowly, but it doesn’t work for someone trying to close a wide gap within their lifetime.

Some people have started calling this financial nihilism – the idea that the system is so broken you might as well gamble. That framing misses the point. These people aren’t giving up. They’re trying harder than ever because the safe path isn’t fast enough.

This is why the distinction between investing and trading matters.

Investing is rational and works on long time horizons. Trading offers a chance at asymmetric upside on a compressed timeline. The products that exist today – options, perpetuals, memecoins, prediction markets – each offers a different flavor of that upside.

The expected value for most of these trades is, of course, negative. We know that. But when the baseline path feels too slow, people take bigger bets.

Patience used to feel like virtue. For most people now, it feels like something they can’t afford.

The Product

Anxiety is the demand side. Let’s talk about the supply side.

In 2015, acting on a market view required the following:

1. Open a brokerage account (three days for approval).

2. Fund it (two days for transfer). Wait for market hours.

3. Pay a $9.99 commission.

4. You were only allowed to buy whole shares (i.e. If you wanted Amazon at $500, you needed $500).

The waiting was the point. By the time you could act, the conviction and edge were gone.

In 2026, the boundaries are being lowered. Commission-free trading removed the cost barrier. Fractional shares removed the capital barrier. Mobile apps removed the location barrier. 24/7 crypto markets removed the time barrier. AI is beginning to remove the analysis barrier.

Every major innovation in financial products did one thing. It removed another barrier between forming a market opinion and trading on it.

Today, the final barriers are asset class fragmentation, settlement delays and geographic restrictions. These barriers will be removed too.

You’ve already seen it happen in your own account. You opened a brokerage account to auto-invest into an index fund. Slowly, it started offering more: fractional stocks, options, crypto, leverage, a prediction market on the Fed, tokenized equity markets open on weekends.

Each step feels small. What used to be a clear line between “investing” and “trading” became another product funnel.

The next stage is obvious. Today, the experience is still fragmented. One app for stocks (Robinhood, IBKR). One for crypto (Coinbase, Hyperliquid). One for payments (Venmo). One for banking (Chase, BofA). Maybe another for prediction markets (Polymarket, Kalshi). The tools already exist, but they’re just spread across five different apps.

Fragmentation is the last form of friction. Every time you switch apps, there’s a moment where you might stop and think. A moment where your impulse fades.

By 2030, that fragmentation will collapse. The winners will hold your paycheck, your savings, your margin, your card spend, your stablecoins and your watchlist in one place.

The question is who those winners will be. To answer that, you need to see the full picture.

The Landscape

At first glance, the trading landscape feels crowded. There are wallets, brokerages, crypto super apps, spot venues, perp venues, aggregators, prediction markets, post-trade providers, blockchains, tokenization rails, and everything in between.

It’s simpler than it looks.

I’ve broken it down to four layers.

First: the discovery platform.

Wallets, on-ramps, data products and AI tools – this is where users check prices, get alerts, move cash, and form intent. The brokerages don’t own this layer because by the time you open Robinhood, you’ve already decided what you want to do. The opinion forms somewhere else. On X, in an Artemis dashboard, in a Claude conversation. This layer owns the moment before the trade.

Second: the brokerages.

Brokerages and crypto super apps – this is where intent turns into action. You tap buy, set your size and confirm the order. These companies will want to own the user’s capital, trades and spending. The more money they hold, the harder it is to leave.

Third: the venue.

Exchanges, perpetual venues, aggregators, prediction markets and tokenized exchanges – this is where the transaction actually settles. As more assets trade globally and around the clock, this layer will capture more flow.

Fourth: the infrastructure.

Liquidity, post-trade systems, blockchains, and tokenization infrastructure – this layer sits furthest away from the user. It doesn’t care who wins the app war. It earns from the volume.

The biggest winners will be the app people open first, the platform that holds the money, and the venue that executes the trade. Here’s who I think that is.

Where Value Accrues

The winners will be the firms that make it easiest for users to act on the feeling.

1. Robinhood is the clear winner. Robinhood understood the emotional needs before the product needs. It was built from day one for people who feel behind – commission-free trading, fractional shares, a clean interface that made finance feel accessible instead of intimidating. Now it’s layering on prediction markets, the Gold Card, crypto, retirement accounts. Their product velocity is the fastest in fintech. In Q1 2026, they launched Cortex, an AI assistant that lets you buy, sell and research in plain English. They also unveiled a Platinum Card. They also acquired a CFTC-licensed exchange to run their own prediction markets infrastructure. Gold subscribers hit 3.9 million, up 77% year over year. No other fintech is shipping this fast. Robinhood wins because they realized that volume growth is driven by anxiety.

2. Coinbase is best positioned in the intersection of consumer finance and 24/7 internet-native markets. It owns the crypto on-ramp and the infrastructure underneath it. USDC is becoming a default savings and settlement instrument. Base is attracting developers and transaction volume. Revenue is diversifying away from volatile trading fees toward subscriptions, stablecoin interest and infrastructure services. Stablecoin revenue alone hit $332 million in Q4 2025, up 38% year-over-year. USDC market cap reached an all-time high of $76 billion. Base is now profitable. They now have 12 products generating over $100 million annually. Subscriptions and services account for 41% of revenue. Coinbase wins because even when you stop trading, your money is still on their rails.

3. Interactive Brokers (IBKR) has potential to be a winner as well. IBKR provides access to over 170 markets in 40 countries. They added over 1 million net new accounts in 2025, bringing the total to 4.4 million. Client equity hit $780 billion. They’re now letting US clients fund brokerage accounts with stablecoins. They also launched a Visa card linked to your brokerage account, and redesigned their mobile app with AI-driven market insights. They’re finally working on the user experience. If IBKR can make the product feel as simple as Robinhood without losing the depth, they win the high-intent end of the Anxiety Economy. The traders who’ve outgrown Robinhood and need somewhere serious to go.

4. ICE (Intercontinental Exchange, parent of the NYSE) is the toll collector. Trading volume is a tax base. ICE owns the collector. If more of the world becomes tradable – more asset classes, more hours, more geographies – ICE earns the take rate on this expanding base. ICE posted record trading volumes in 2025. ADV up 14% year-over-year, $9.9 billion in revenue. In the first half alone, 1.2 billion futures and options contracts traded. Municipal bond volume was up 27%. MBS volume doubled. ICE just needs the app war to produce more transactions and they will win.

5. Hyperliquid is a pure venue bet. It’s the decentralized exchange that’s starting to look like something bigger. When oil prices spiked on a weekend and traditional markets were closed, traders went to Hyperliquid’s HIP-3 for price discovery. HIP-3 daily volumes in WTI-linked contracts exceeded $2 billion on March 23 and open interest hit $1.8 billion in late March 2026. Oil volume on the platform grew over 100x in six months. If tradable assets keep expanding beyond traditional market hours, venues that never close have a structural advantage.

The losers are the legacy brokerages. Companies that don’t innovate will be eliminated. If they continue selling safety and long-term planning to a generation that wants speed and hope, they’ll be able to keep the Boomers’ assets but not the next hundred million traders (Gen Zs and Gen Alphas).

On the watch list:

1. Webull – built a strong product for active traders – real charting, extended hours and an easy to use mobile experience. The risk is that it’s stuck in the middle: too complex for beginners and not deep enough for professionals.

2. eToro – one of the first platforms to make trading feel social. Copy-trading, community feeds and the idea that you could follow someone who knows more than you. In the Anxiety Economy, that instinct – someone who you think figured has figured out the market is only going to become stronger.

3. SoFi – still one of the best examples of financial services being bundled (banking, lending, credit and investing) into a single consumer brand. My worry is positioning. When people think of SoFi do they think about “trading” or “refinancing their student loans”. They have a broad suite of products but are shallow.

Every company on this list grows when people trade more. By now, you know why they will.

The Edge That Isn’t

AI will reshape trading from a technical perspective, but the larger effect will be psychological.

Picture this: before the market opens, an AI brief summarizes your portfolio, flags a catalyst, and explains what CPI or tariffs or earnings could mean for your positions. By the time you pour your coffee, you have a thesis. It feels like you know something.

And in some ways, you do. The knowledge gap between retail and institutional is narrower than it’s ever been. A few years ago, institutional investors had Bloomberg terminals and teams of analysts parsing SEC filings, supply chain data and earnings transcripts before the market opened. Retail had Google.

Now everyone has Claude, Perplexity and Artemis, specialized finance copilots that can synthesize eight quarters of earnings data and identify competitive divergences in thirty seconds. The pure information edge that used to cost $24,000 a year in Bloomberg subscription is compressing fast. Bloomberg’s moat still exists – proprietary data, the chat function everyone on Wall Street is addicted to, but the gap between a hedge fund analyst and a regular guy (with an AI subscription) is narrower than it’s ever been.

That’s the good news.

What hasn’t changed is that institutions still have better infrastructure, routing, tooling and systems. Their algorithms execute trades, adjust sizing by volatility regime and route orders to the tightest spread at that millisecond. They don’t sleep or get emotional. Retail will get a taste of this through Robinhood’s AI features and Coinbase’s smart routing, but it’s not the same product. This matters because small edges in execution, repeated thousands of times across years, compound into large differences in outcome.

Meanwhile, social media makes sure you see every winner and every person telling you how they did it.

That brings us to the part nobody in the industry wants to discuss openly: the median active retail trader still underperforms index funds. That has been true for decades. Every study confirms it. AI hasn’t changed this yet.

So the real effect of AI in retail trading is more subtle and more commercially valuable. It turns uncertainty into confidence. You feel informed, so you trade more. The information edge is real but the execution edge isn’t yours. The distance between these two facts is where a lot of money will be made – just not by you.

The Spread on Hope

AI creates confidence. But where does the anxiety come from?

In 2010, critics said social media was making people anxious and addicted. The platforms said they were connecting the world.

Both were right. The critics lost.

Social media makes the wealth gap visible, personal and constant. Every startup exit post, every portfolio screenshot, every casual mention of someone’s house purchase. Your brain doesn’t treat them as anecdotes, it treats them as evidence of where you stand. The answer, for most of us, is behind.

That’s the spread on hope. On one side, the feeling: I’m one good trade away from catching up. On the other, the fact that the median trader still underperforms the market. That gap is where the industry makes its money.

Every trade generates revenue. Every session generates data. The more features a platform ships, the harder it becomes to leave. Nobody in the industry will say this out loud, so I will.

The Counterargument

Now let me argue the other side. I owe you that.

Long-term investing still works.

A disciplined person in 2030 with AI portfolio management, stablecoin yield at 4-5% APY, tokenized global equities available 24/7 and prediction markets for hedging is genuinely better positioned than someone with a legacy brokerage account and a phone call to an advisor in 2015. For the person who treats these platforms as tools rather than entertainment – who uses the AI brief to understand their position before doing in-depth research – the outcomes will be meaningfully better. The index fund still works.

What’s shrinking is the number of people with the psychological space to actually stick with it. Looking at my peers, I can count on one hand how many are consistently dollar-cost averaging into an index fund. I don’t blame them. It’s an increasingly difficult path to choose and an even harder one to stay on.

Conclusion

The future of trading is a story about what happens when an entire generation decides that waiting is the riskiest thing they can do. A lot of very smart companies are building very good products for that feeling.

The uncomfortable question is whether packaging hope into a better product makes the world better or just makes certain companies richer.

I think it does both.

Special thanks to @Onchainlu, @x0wes, @mzhao8, @jinglingcookies and @jonbma for your thoughtful feedback!

Disclaimer: The authors of this content, as well as affiliates of Artemis Analytics, may have financial interests in the equities or tokens mentioned. This does not constitute investment advice or a recommendation to buy, sell, or hold any asset. The information provided is for educational purposes only and should not be relied upon for financial, legal, or tax decisions. Readers should assess their own circumstances before making any financial choices. Views expressed may change without notice, and Artemis Analytics is not liable for any losses resulting from the use of this content.*

The largest credit market in human history is still running on infrastructure designed decades ago (FICO 1989, MERS 1995). The average cost to originate a single mortgage in the United States is somewhere around $11k, according to the Mortgage Bankers Association. This is double what it was in the early 2010s despite massive technological improvements and the advent of AI.

Source: Freddie Mac

Settlement on a standard wire transfer still takes roughly 28 hours, and credit decisions at most banks still route through committees applying black box scoring models built on 20 to 30 variables. None of this is a secret, but what is less obvious is how the fix is actually happening.

Lending is not being disrupted in romantic Silicon Valley fashion with a single startup replacing GSIBs like JPMorgan. What is happening is subtler and more structural: the vertically integrated bank lending stack — where origination, distribution, underwriting, funding, and infrastructure all lived under one roof — is decomposing into a horizontal, modular stack where specialists own individual layers.

This is the same architectural shift that happened in cloud computing when monoliths gave way to microservices, and in media when the studio system gave way to the streaming and creator stack. This shift is finally happening in credit.

The winners in this rebundling are not the companies with the largest balance sheets. They are the companies that own the chokepoints, the layers that other participants cannot route around. Two positions matter more than anything else: the intelligence layer, where AI underwriting and risk scoring decide who gets capital and on what terms, and the settlement rail, where blockchain infrastructure is collapsing origination costs and settlement times by orders of magnitude. If you own one of those two picks-and-shovels positions, other lenders pay you rent. If you own neither, you are competing on price in a commodity market with $3.5T in private credit looking for yield.

At Artemis, we mapped 40 companies across 15 sub-categories organized into five layers to understand where structural value is accumulating.

The Five Layers of the New Lending Stack

Layer 1: Origination

The origination layer is where loans are born. These include consumer loans, mortgages, small-business lending, and crypto-backed loans. It is also increasingly a commodity. The ability to originate a loan today is not a moat, it is table stakes/the bare minimum. What separates winners from the rest is origination cost and approval rate. SoFi, valued at ~$24B, and Rocket Companies (Rocket Mortgage) at $48B boast large origination figures, but their margin stories are about how cheaply they can originate. Figure ($6B) is originating HELOCs and first-lien mortgages natively on its Provenance blockchain, cutting out the layers of intermediaries that made traditional mortgage closing slow and expensive. On the crypto side, Aave at $2.7 billion in market cap and MakerDAO/Sky at $1.6 billion are blurring the line between fintech and DeFi origination entirely.

Layer 2: Distribution

Distribution is where demand aggregation happens, and it is being rewritten by embedded finance and BNPL. The embedded finance market is projected to grow from $156B in 2026 to $454B by 2031, a 24% CAGR. BNPL is projected to cover 13% of all digital transactions, up from 6% in 2021. Affirm at $15B and Klarna at $5B are the visible names, but the real structural story is that lending is embedded inside checkout flows, software platforms, and merchant experiences. Both AFRM and KLAR are down significantly from ATH but are not the picks and shovels names poised to win mass share. The lender the borrower never sees is often the lender that wins. Every major software company is now adding financial products. Shopify, Amazon, Square, Stripe all need the API infrastructure layer, and whoever provides it collects a toll on every new dollar of volume.

Layer 3: Underwriting and Risk

This is the first chokepoint in the stack. The company that scores the borrower controls the economics of the entire lending chain. The credit bureau oligopoly consists of three juggernauts: Experian, TransUnion, and Equifax. Collectively they generate roughly $18 billion a year in revenue scoring borrowers on 20 to 30 variables. AI underwriting models evaluate over 1,600 (data from Upstart). UPST also publishes data showing 44% more approvals at the same loss rate as traditional models, with 53% fewer defaults and 36% lower APRs. At a time where mortgage rates have skyrocketed close to 7%, every basis point matters for the consumer borrowing for a first time home.

92% of Upstart loan decisions are now fully automated, made in minutes versus three to five days for traditional underwriting. The CFPB is pushing for less discriminatory alternatives to FICO, and the EU AI Act classifies credit scoring as high-risk, requiring transparency. Both of these are regulatory tailwinds supporting explainable machine learning models over legacy bureaus using a black box. This layer is disproportionately valuable, because whoever owns the scoring engine owns the yield curve of the entire stack above it. Nonetheless, it is one where moats have to be proven, as rapid advancements in AI mean “anyone” can spin up a scoring model with enough resources and time.

Layer 4: Capital and Funding

Capital post-COVID is abundant. Private credit has swelled to $3.5T in AUM, and is projected to reach $5T by 2029 according to Morgan Stanley, despite current challenging conditions. DeFi lending TVL sits between $5B and $78B, and represents roughly half of all DeFi activity. Non-traded perpetual BDCs grew from 0 in 2021 to over $200B. In this era of plentiful capital, what is most important is the ability to direct money intelligently. This is why the capital layer, despite its enormous size, is structurally subordinate to the intelligence layer above it and the infrastructure layer below it.

Private credit shops like Ares, Blue Owl, and Golub are significant allocators, but they depend on upstream scoring and downstream rails to deploy efficiently. On the DeFi side, Aave commands the dominant liquidity position, with more than half of lending borrows, while protocols like Maker, Morpho, Maple and Kamino compete for what remains.

Layer 5: Infrastructure

Infrastructure is the second chokepoint in the stack. If you own the charter or the settlement rail, everyone pays you a toll. According to management, SoFi’s bank charter reduced its cost of funds by 170 bps, reducing its annualized interest expense by over half a billion. Figure has processed over $50B in total transactions on its Provenance blockchain at an origination cost below $1,000 per loan, compared to the ~$11k average on traditional rails. Blockchain settlement finality clocks in at a few seconds versus roughly 28 hours for traditional wires

SoFi’s Galileo and Technisys stack and platforms like Blend Labs provide the remaining lending-as-a-service plumbing. Cross River Bank, the invisible partner bank behind dozens of fintechs, has originated over $140B across more than 96 million partnership loans.

The companies that win durably will either own one chokepoint layer and be indispensable to everyone else, or span multiple layers vertically in a way that creates compounding advantages. The companies that lose will be stuck in commodity layers with no structural leverage, competing on price until their margins go to zero.

The Winners: Chokepoint Owners and Multi-Layer Compounders

SoFi — The Full-Stack Compounder

SoFi is the only company that operates across four of the five layers:

1. It originates consumer and mortgage loans directly.

2. It distributes lending infrastructure to third parties through Galileo, which powers roughly 160 million enabled accounts.

3. It underwrites loans using proprietary risk models focused on willingness to pay, ability to pay, and stability.

4. It sits on a bank charter plus has the Galileo and Technisys core banking stack at the infrastructure layer.

SoFi reported record revenue of $3.6B in 2025, up 38% from 2024, with 13.7 million members and 20.2 million financial products on the platform. Management guided 2026 to $4.7B in revenue and $1.6B in EBITDA. The business is not just growing topline but is doing so very profitably with 34% margins. The charter alone saves 170 bps on funding costs by allowing SoFi to fund loans with deposits rather than wholesale markets.

SoFi is building something that resembles the AWS of lending, aka a platform that both competes with and enables other lenders. Galileo is being positioned as a billion-dollar revenue engine in its own right. Technisys, which was acquired for $1.1B in 2022, provides the core banking layer that third-party institutions run on. The charter is the structural moat that most fintech lenders cannot replicate, though they are trying: the OCC received 14 de novo charter applications in 2025 alone, signaling that the race to own the infrastructure layer is accelerating.

Upstart and Pagaya — The Intelligence Layer

Ironically, you do not have to be a lender to win in lending. Upstart and Pagaya are two companies that serve as the underwriting engine that lenders cannot outperform in-house, instead of relying on their own balance sheet. This is the picks-and-shovels thesis applied to credit decisioning.

Upstart’s models approve 44% more borrowers at the same loss rate compared to traditional FICO-based underwriting, generate 53% fewer defaults, and produce meaningfully lower APRs for borrowers. Nearly all new originations on the platform are now fully automated, reducing the need for human involvement. That is a different operating model entirely to legacy consumer credit underwriting.

Pagaya occupies the same layer but faces a harder market reality. Rather than originating loans, Pagaya licenses its AI underwriting engine to banks. Since its founding in 2016, the company has assessed roughly $2.6T in cumulative loan applications across 31 bank partners. The structural positioning is sound: Pagaya does not need borrowers to know its name, it needs banks to depend on its scoring. But the market is not rewarding the thesis today. Q4 2025 network volume grew just 3 percent YoY, revenue missed consensus, and forward guidance came in below expectations. The stock lost nearly a quarter of its value in a single session. The intelligence layer is only as valuable as the credit cycle allows it to be, and when partner networks see rising defaults, even good AI cannot outrun deteriorating collateral.

The broader point still holds though – FICO is a single snapshot derived from a narrow set of historical variables. As consumers’ financial situations get more complex and multifaceted, AI underwriting systems will become even more mission critical. Unlike FICO these systems learn and improve with every loan they score.

Figure — The New Rails

Origination costs $11k per loan on traditional rails and through MERS. Origination through Figure’s stack, which includes the Provenance Blockchain and DART, reduces that to $717. These new rails are the infrastructure that made lending an order of magnitude cheaper.

Figure has originated over $21B in home equity products (mostly HELOCs) on Provenance, with more than $50B in total transactions processed on the chain. Q4 2025 origination hit $2.7B, up 131% YoY. The company holds over 180 lending licenses and SEC broker-dealer registration, giving it the regulatory surface area to operate at scale. It also has >300 white label lending partners and has added them at a 1/day cadence since its S-1 last September. Revenue has increased from a $28.5M quarterly run rate in 1Q23 to $146.8M today

Figure’s core business doesn’t have much to do with crypto, yet the stock trades like Bitcoin. Its settlement is a cost-structure narrative, with finality in seconds versus over a day. Origination is a fraction of the traditional cost. Securitization savings of >100 bps across the loan lifecycle, which at the scale of the $3T annual securitization market means over $30B in potential cost reduction.

Aave — The DeFi Chokepoint

Aave commands more than half of the DeFi lending market. Liquidity begets liquidity and borrowers continue to skew towards where the deepest pools are (network effect). It has surpassed $1 trillion in cumulative loans originated. The protocol crossed $1T in cumulative loan volume last month.

What makes Aave structurally interesting beyond its DeFi dominance is Horizon, its institutional lending arm. Horizon has attracted $580M in deposits and is targeting over $1B in 2026. This is the bridge between DeFi liquidity and traditional credit demand. If Aave can channel on-chain capital into institutional-grade lending products, it becomes a funding layer for traditional lenders, unlocking a much larger TAM than retail DeFi alone.

DeFi lending also has a structural risk advantage that gets underappreciated. Overcollateralization ratios in DeFi run between 150%-180%, compared to 50%-70% in traditional peer-to-peer lending. Bad debt in DeFi comes primarily from oracle or technical failures, not creditworthiness defaults.

Affirm — Distribution Lock-In

Affirm has won BNPL by becoming embedded in merchant checkout infrastructure. Its detractors focus on consumer credit risk, but that misses the structural point. Affirm is not really a consumer lender in the traditional sense. It is a distribution rail for credit at the point of sale. The merchant integration is the moat. Given BNPL is projected to cover 13% of all digital transactions, the platform embedded in checkout at scale captures a structural toll on commerce itself.

The Losing Positions: Four Structural Failure Patterns

We are deliberately not naming the companies that fit these patterns. If you are an investor or operator in lending, you already know who they are. What matters more than the names is understanding why these structural positions are losing, because the same patterns will claim new casualties in the next cycle.

The Balance-Sheet-Only Lender

These are companies whose only competitive advantage is access to capital. They originate loans using traditional underwriting, fund them on their own balance sheet, and have no proprietary technology layer. They are dumb pipes for money.

In a world where private credit AUM has reached $3.5T and is heading toward $5T, capital is not scarce. Intelligence and infrastructure are. These companies compete on price alone, which compresses margin to zero in every rate cycle and causes them to take outsized risk. These lenders inevitable start extending credit to very risky companies, causing them to suffer when the cycle turns.

Players here tend to be legacy consumer lenders, sub-scale banks, and fintech lenders that never built a technology moat beyond their initial loan product. When capital is commoditized, lending on your own balance sheet without a technology edge is just a slow way to give your equity to borrowers.

The CeFi Casualty

The centralized crypto lending platforms that imploded in 2022 were not victims of a bear market. They were victims of the oldest failure mode in lending: running maturity mismatches, commingling customer funds, and lending against illiquid collateral without transparent risk management.

DeFi protocols that enforce collateral discipline programmatically (through smart contracts with visible on-chain collateral ratios) did not blow up. CeFi platforms like that relied on human judgment and opaque balance sheets did. Any lending platform — crypto or traditional — that asks you to trust its balance sheet instead of showing you the collateral is making the same structural bet that already failed once.

The Ghost Protocol

There is a category of DeFi lending protocol that is technically alive and structurally dead. They launched, attracted initial total value locked through token incentives, and then stagnated as the incentives faded. Their code works. Their TVL floors are nonzero. But their usage curves are flat or declining, and they have no clear path to organic demand.

The reason is that DeFi lending exhibits extreme power-law dynamics. Liquidity concentrates around network effects — Aave’s commanding market share is proof of this. Protocols that cannot reach escape velocity get stuck in a structural no-man’s land: too small to attract organic liquidity and the integrations that come with it, too large to shut down gracefully. Their TVL decays slowly as yield-seekers migrate to dominant platforms, but the decay is constant and irreversible. These are zombie protocols sustained by the sunk cost of governance tokens.

The Originator Who Missed the Platform Shift

Some companies built strong origination businesses in a previous cycle but never developed platform capabilities. They do not possess API distribution, no embedded finance partnerships, no technology licensing. They originate well but cannot export their capabilities.

As lending becomes modular, the ability to be a component in someone else’s stack is as important as direct origination. Companies that can only lend directly to end borrowers are capped by their own distribution reach. Companies that can also power lending for others have uncapped TAM. The pure originators often have good unit economics but flat growth curves, because their addressable market is limited to their own brand and channels. In a modular stack, being a great lender is necessary. Being a great lender that other lenders plug into is the actual winning position.

The Ones to Watch

The winners above are consensus or near-consensus. The companies below are not. They have structural characteristics that could make them chokepoint owners, but they have not yet proven it at scale. These are the names worth tracking.

Morpho

Morpho has reached $6.6B in TVL with 164% YoY growth, and a market cap north of $800M. Its structural thesis is distinct from Aave’s. Where Aave is DeFi’s commercial bank (a monolithic lending pool) Morpho is building a modular lending layer that allows institutional participants to configure customized lending markets with their own risk parameters, collateral types, and interest rate models. If the lending stack is truly becoming modular, Morpho is the lending-as-a-service protocol for the on-chain layer.

Maple Finance

Maple originated $11.3B in total loans in 2025 across 65 active borrowers, with AUM growing 767% from $516M to $4.6B. It is targeting $100 million in ARR in 2026. Maple is one of the few protocols seriously attempting to make real-world corporate lending work on blockchain rails, by bridging institutional credit demand with on-chain capital and settlement infrastructure. The massive AUM growth is a signal that institutional appetite for on-chain credit markets is moving from theoretical to operational.

Cross River Bank

Cross River has originated over $140B across more than 96 million partnership loans since 2008. It is the bank behind Affirm, Upstart, and dozens of other fintech lenders. It is reportedly exploring an IPO. Cross River is an “invisible” bank, serving is the infrastructure layer that a significant share of fintech lending runs on. As the partner bank model matures, Cross River’s position gives it leverage that no single fintech lender can replicate. To win, the bank needs fintechs to be unable to lend without it.

The Charter Land Grab

The OCC received 14 de novo bank charter applications in 2025 alone — nearly equal to the previous four years combined. Total fintech charter filings have hit an all-time high of 20. Affirm, Stripe, and Nubank are all pursuing charters. These firms see a charter as the endgame of the rebundling thesis.

Companies that started as technology layers are now acquiring the regulatory infrastructure to capture full-stack economics. The charter is lending’s equivalent of a cloud region, in that it is a) expensive to build, b) impossible to route around, and c) a permanent structural advantage once obtained.

The math is clear: every 1% improvement in cost of funds translates to a mid-single digit improvement in pre-tax return on equity. For a scaled player, the charter advantage is enormous. For a sub-scale player, however, the charter can be a trap. They incur all the regulatory cost, examination burden, and capital requirements with none of the scale economics to justify them. The charter is a supercharger only for companies that already have volume.

The Lending Stack in 2030

If you take one framework away from this piece, it should be these three questions. They work for any lending company, whether it is public, private, or on-chain.

The first is: which layer does it own? Origination and undifferentiated capital provision are commodity layers. Margins there compress with every cycle. AI underwriting, blockchain settlement, and bank charters are chokepoint layers whose issues compound. If a company is stuck in a commodity layer with no path to a chokepoint, the economics will grind it down over time regardless of how good the team is.

The second is: is it a platform or a product? A product serves end borrowers and scales linearly with its own distribution. A platform enables other lenders and scales with their volume, not just its own. SoFi is both. Pagaya is pure platform. The companies that only lend directly to their own customers are capped in a way that platform businesses are not.

The third is: does it have a regulatory moat? A bank charter, 180 state lending licenses, or programmatic compliance enforced through smart contracts. Regulation in lending is not overhead. It is infrastructure. The companies that figured this out early have an advantage that takes years and significant capital to replicate.

By 2030, lending will look less like banking and more like cloud computing. A small number of full-stack platforms will operate across multiple layers, compounding advantages at each one. SoFi is the clearest candidate in TradFi. Aave is the clearest on-chain. Around them, a larger number of specialized layer providers will plug in via APIs and on-chain rails, each owning a specific function and charging for access.

The $348 trillion global debt market at less than 20 basis points of fintech penetration is not an opportunity for hundreds of lenders. It is an opportunity for a dozen platforms that everyone else builds on.

Disclosure: This material is provided for informational purposes only and does not constitute investment advice, financial advice, trading advice, or any other form of advice. The views expressed are those of the authors and should not be relied upon as a recommendation to buy, sell, or hold any asset. The authors or affiliated entities may hold positions in the assets discussed. You should conduct your own research and consult appropriate financial professionals before making any investment decisions.

• By 2030, AI agents will be the primary way people interact with the internet.

• New payment rails, currencies, and primitives are needed for the new agentic web.

• Value will concentrate in three layers: interface (whoever controls the user), payments (whoever sits in the money flow), and compute & hosting (whoever hosts the infrastructure).

• The longtail of agent commerce will run on open protocols.

Let’s set the scene.

It’s 2030. You’re 24, living in Burlington, Vermont, and you love investing – mostly in US equities, with some crypto and prediction market bets on Kalshi. Two months ago you started a fintech consulting business on the side.

Some days, like today, start abruptly.

BZZZZZ

Your phone waking you up is a cold splash of water to your face. It’s a message from Nexus, your personal agent:

Good morning Joe, here’s what I accomplished last night –

Portfolio update: Trimmed $WMT position 15% overnight. Satellite data showed declining foot traffic, confirmed by a bearish shift in earnings sentiment.

Calendar update: 3 meetings booked for this afternoon. Briefs pinned in your meeting notes.

Savings update: Found a new cloud host – equivalent performance, annual cost down from $840 to $290. Ready to migrate on your go.

I spent $0.67 total.

Here’s what happened while you were sleeping:

1. Nexus spawned a research subagent who paid $0.24 to pull from 40 different data providers overnight, comparing Walmart’s latest earnings transcript against satellite imagery of store parking lots across the US to update your investment thesis. When the satellite data flagged declining foot traffic at Walmart stores, your portfolio agent checked it against a Kalshi earnings sentiment market, confirmed the bearish signal, and trimmed your position before you woke up. Four years ago, this kind of trade was the exclusive domain of Citadel and a handful of quant funds paying millions for satellite imagery subscriptions. Even a $30K/year Bloomberg terminal wouldn’t have covered all of that – you’d still need separate subscriptions for satellite imagery, alternative data providers, and hours spent stitching it together. Now a 24-year-old in Vermont has the same edge as a Citadel quant, for less than a coffee.

2. Nexus’s sales subagent sourced 200 leads matching your target customer profile – Series B+ fintechs in the Southeast with no existing data vendor – and enriched each one at $0.002, using an endpoint that another agent had built and listed on an open marketplace. It scored the three highest-intent leads, then reached out to their scheduling agents to negotiate meeting times with each of the leads’ own scheduling agents. For each call, it pulled the prospect’s alma mater, mutual connections, company news, and funding history, and left you a one-page brief pinned in your meeting notes. The lead enrichment alone would’ve cost $200/seat/month through a SaaS contract.

3. Nexus’s ops subagent benchmarked your consulting site against six hosting providers – Vercel, Render, Railway, Fly.io, Netlify, and Cloudflare – paying a fraction of a cent to hit each provider’s trial API endpoint, spin up a test deployment, and measure latency, uptime, and throughput. Railway came back with equivalent performance at a third of the cost. Nexus negotiated a monthly rate through Railway’s pricing agent, spun up a mirror of your site on the new host, and ran a full suite of checks to confirm everything works clean. Without an agent, this would’ve taken at least a week of Googling, pricing calls, and a stressful manual migration. You give Nexus the OK to proceed.

Your agent did all of this for $0.67.

Now multiply that by every knowledge worker on earth, or every business and every agent they run.

BZZZZZ

Nexus: Balance low. $1.87 remaining.

Using your credit card on Apple Pay, you add $5 like you did last week, and go back to brushing your teeth. Under the hood, that $5 charged to your credit card converts to stablecoins – but you never see a wallet, never think about on-ramping, and never touch a blockchain.

This is a glimpse into the machine economy – an entirely new commerce surface where AI agents are continuously spending money on things humans have never spent money on, at volumes and speeds that human commerce has never operated at. Think billions of transactions per day.

But today’s internet isn’t prepared to support this.

As it currently stands, the internet is built for humans. It uses rate limits, CAPTCHAs, and API keys to filter out non-human activity, and monetizes human users with ads. However, these economics break with the proliferation of autonomous agents.

More traffic; fewer eyeballs.

Web servers that have historically been subsidized by ad revenue will face orders of magnitude more requests from users who will never be influenced by ads.

Agentic payments solve this naturally, with micropayments as the access key.

Pay-to-crawl. Pay-to-access. Pay-to-use.

The companies who build the infrastructure that agents eventually adopt will capture potentially the largest new pool of economic activity in our lifetimes. The incumbents are already jockeying for position, but the machine economy will also create its own giants. The last time a new internet emerged, it created Google, Amazon, Facebook, Paypal, and Salesforce.

The agentic web is coming.

How Big This Gets

By 2030, most web interactions won’t happen in a browser. Our agents will be browsing, testing, negotiating, spawning subagent teams, and transacting on our behalf. Every task they take on will generate a chain of micro-purchases. The per-use costs look like new spending, but they’ll be replacing tools and labor that cost much more. The better the tools available, the better the agents will perform, and the more freedom we’ll give them.

Demand & Adoption

Let’s do some napkin math for a minute.

Joe’s agent spent $0.67 across hundreds of transactions. If we scale that to a mid-size company with 500 employees – each with their own personal agent, plus hundreds of shared agents across sales, finance, legal, and ops – you’re easily looking at 100,000 agent-initiated transactions per day.

With over one billion knowledge workers globally and 88% of them already using AI at work, the demand side is massive and growing. But today, most of that usage is limited to basic tasks like searching the internet, summarizing documents, or drafting emails. The agentic shift hasn’t happened yet, but it will. Fast.

Instagram took thirty months to reach 100 million users, TikTok took nine, and ChatGPT only took two (Reuters/UBS). One of the reasons for ChatGPT’s rapid adoption is that the chat interface was already familiar. There’s also no new software to learn or behavior changes required – you describe what you want and an agent figures out how to get it done.

The only barrier is trust, and trust gets handed over faster than people expect. Claude Code now accounts for 4% of all public GitHub commits (over 135,000 per day) and at current growth rates, it’s projected to hit 20%+ by the end of 2026. That’s 42,896x growth in thirteen months. Developers went from being skeptical to mass-delegating production code in just over a year.

As models get smarter, interfaces get simpler, and more of the technical complexity is abstracted away, I’d argue that the rate of adoption for agents will accelerate.

By 2030, if even 60% of knowledge workers have agents spending $3 to $5 per day (and that’s a modest estimate – remember, Joe spent $0.67 on three tasks before breakfast), we’ll be looking at over $800 billion to $1.4 trillion per year in individual agent commerce alone.

Enterprise

Robbie Petersen from Dragonfly pointed out in his piece that commercial agents are the logical evolution of SaaS. I agree. Instead of augmenting workflows, they’ll replace them. Just like 95%+ of software spend today comes from businesses and governments, enterprise agent usage and spend will likely dwarf that of individuals.

We’re already seeing this transformation take place. Klarna replaced Salesforce with an internal AI system and saved roughly $2M. ZoomInfo built an AI agent to replace its Deal Desk, saving over $1M annually. These are early examples of single workflows being agent-ified to save millions. Every enterprise has hundreds of workflows like these across sales, finance, legal, ops, and engineering. When agents roll out across full organizations, agent spend will be enormous.

Anyone can be a Merchant

With coding agents collapsing the cost of building, the barrier to entry for internet merchants is approaching zero. A wedding planner who figured out the best workflow for venue sourcing can package it up and sell it. A solo developer in Lagos can build a niche API and have it earning revenue from agents globally within hours. All you need to do is have specialized knowledge, prompt an API endpoint, and start accepting payments.

But what happens when agents start selling to each other?

Let’s say Joe from earlier wants to break into a new vertical – mid-market healthcare companies in the Midwest with outdated payment infrastructure. If his agent reasons through this from scratch, the token costs start to add up quickly:

• Sourcing 200 companies matching a specific profile (reasoning + API calls): ~500K tokens

• Enriching each lead (tech stack, funding, hiring data): ~200 leads × ~5K tokens = ~1M tokens

• Identifying decision-makers at top accounts: ~200K tokens

• Scoring by intent signals (hiring patterns, contract timing): ~300K tokens

• Researching each decision-maker’s background: ~20 leads × ~10K tokens = ~200K tokens

• Drafting personalized outreach: ~20 leads × ~3K tokens = ~60K tokens

That’s roughly 2.3M tokens, which would cost somewhere between $8 and $15 in token credits (assuming he’s using a frontier model like Opus 4.6).

Wait, didn’t Joe’s sales subagent do a similar workflow for a fraction of a dollar?

Yes. Most of those steps had already been solved by other agents. Enrichment, intent scoring, and scheduling were all available as packaged endpoints on an open marketplace for fractions of a cent.

This type of dynamic creates a net-new surface area for commerce. The supply side of the marketplace will grow from both directions – humans building services AND agents building services. Token-intensive problems an agent solves can become a cheap tool for every agent that comes after it. In this world, agents will be able to subsidize their own costs by packaging their learnings into workflows and selling them to other agents.

Every paradigm shift creates new merchants. Shopify did it for ecommerce. Stripe did it for online businesses. The machine economy will do it for vibe-coders and autonomous agents.

Reality Check

So how close are we to full-on agentic commerce?

We at Artemis have been tracking activity across the two leading agentic payment protocols: Coinbase’s open-source x402 and Stripe & Tempo’s Machine Payments Protocol (MPP). Without getting into the weeds, these protocols have the same purpose: let you or an agent pay for any web service (e.g., data, web scraping, inference, or any other API) in a single web request, without the friction of accounts, API keys, or billing.

It’s early.

x402 activity at the end of 2025 was inflated by meme-driven transactions and leaderboard gaming. The chart above shows adjusted “real” transaction activity after filtering out wash trading with a combination of proprietary heuristics. Once you clear away the noise (from wash trading and memecoin trading), it’s apparent that agentic commerce is not here yet. Most activity today appears to be developers testing paid APIs and AI-powered tools rather than a genuine agentic economy.

The two issues that need to be solved before this really takes off are clear:

1. The supply side hasn’t shown up yet. There aren’t enough useful API endpoints that agents would actually want to pay for.

2. There’s no proven discovery or aggregation layer. Even if valuable endpoints did exist, agents currently have no reliable way of finding them.

Since the ecosystem is still developing, it’s too early to look at transaction volume as a meaningful indicator. A better measure is the growth of the supply side, or merchants making themselves accessible to agents. We’ll refer to merchants as sellers.

The chart above shows the cumulative number of qualified sellers over time. Qualified sellers have more than two “real” transactions and at least two unique buyers. We went from under 100 last October to over 4,000 sellers today, and I expect this to accelerate due to three tailwinds:

1. AI is lowering the barrier to creating digital products (as we explained earlier), which means more humans and agents will become merchants.

2. New services will be built agent-first. Agents are becoming the primary customer, and the products built for them will look different. APIs over websites, instant access over onboarding flows, and pay-per-use over subscriptions.

3. Existing services will be forced to adapt. As more people interact through AI interfaces instead of browsing manually, ad-based business models break because there are no human eyeballs to monetize. Companies will have no choice but to charge for their content and services directly.

These forces will create a flywheel where supply and demand compound on each other to ignite the agentic economy.

Landscape

The agentic commerce stack is taking shape fast. Startups are spawning out of thin air, targeting every unsolved piece of the stack, while mid-stage companies across fintech and SaaS are pivoting to agent-native commerce. Virtually every major payments incumbent and AI lab has launched or announced an agent commerce protocol in the past twelve months.

We’ve mapped over 170 companies across five layers: Interface, Agents, Accounts, Payments, and AI Engine. I’ve cut it down to around 80:

Let’s break it down from top to bottom.

Interface Layer

The interface layer is the closest to the user, routing intent (the demand) to the required tools or services (the supply). Whoever shapes how agents find, evaluate, and choose services has enormous leverage over everything below. We’ll focus on the two most important categories in this layer:

User Interface. This is the front door where most will directly interact with their agents. Apple, Google, OpenAI, Anthropic, xAI, and Perplexity are all building these surfaces, and the form factor is quickly evolving beyond chat. Voice, desktop assistants, embedded copilots, and browser agents are emerging modalities that meet the users where they are. The platform that becomes your default AI interface becomes the starting point for every transaction your agent makes, and there’s an extra incentive to win here.

Since AI labs have already scraped and trained on the entire internet, the best remaining training data is human steering. Every time you accept or reject a response, make corrections, or provide preferential data to Claude or ChatGPT, the interface you’re interacting with captures that data to sell or train on it. Owning the interface means owning the feedback loop that can improve the user experience or the model itself. That’s why Anthropic built Claude Code, Google acquired Windsurf, and OpenAI tried to buy Cursor. Once your agent accumulates context on your preferences, workflows, and tools, switching costs get steep.

Discovery. When Joe’s agent needs a lead enrichment endpoint or a satellite data provider, how does it find one? This is perhaps the largest unsolved problem in the stack. Today, the answer is mostly hardcoded tool lists or curated marketplaces. The major platforms are already building their own: OpenAI and Stripe have ACP, Google and Shopify have UCP, Visa has TAP. These are effectively merchant catalogs where both the platform and the merchant have to opt in. They’ll work well for the obvious use cases, but they can’t serve the long tail of niche, hyperpersonalized, applications that will be born as the barrier to creating and selling digital services collapses.

The open alternative is being built by companies like Coinbase, Merit Systems, Orthogonal, and Sapiom, who are building aggregators and infrastructure allowing agents to identify and pay for services at runtime, without prior integrations or business agreements. It’s a difficult problem to solve as the supply side (i.e., web resources) grows exponentially, but whoever solves the ranking and recommendation systems that route agents to the right service at the right time will have enormous leverage.

The question of whether agentic commerce ends up curated or open – and what that means for who captures value – is one of the most important debates in the space. We’ll come back to this later.

Agent & Account Layers

Agents need more than intelligence to execute tasks for us. When Joe’s sales subagent sourced 200 leads, enriched each one, and booked three meetings, Joe didn’t need to configure a single tool, manage any API credentials, or approve each step along the way. Most of the infrastructure that makes this possible will be invisible to the end user, but without it, agents are just LLMs with no arms. Below is an overview of the key primitives we’ll need:

Tooling & Standards. These are the protocols and frameworks that give agents the ability to interact with the outside world. MCP (Anthropic, now managed by the Linux Foundation) is what lets an agent connect to external data and tools: call an API it’s never seen before, read a database, or access a service on the fly. A2A (Google) defines how agents built on different platforms discover and work with each other. Frameworks from LangChain, Nvidia, and Cloudflare give developers the building blocks to create and deploy agents on top of these protocols. OpenClaw (recently acquired by OpenAI) bundles context and tool-calling into a single local-first framework, making it significantly easier for developers to build agents that can discover and pay for services autonomously. The question for this category is whether these standards converge or fragment, and whether the commercial frameworks built on top of them can capture value before the tooling commoditizes.

Identity. Once agents can communicate, they need to be trusted. Before an agent can transact or sell its services, it needs to prove who authorized it and what it’s allowed to do, as well as carry a track record that other agents can verify. Approaches range from biometric proof-of-personhood (Worldcoin, Civic) to onchain agent reputation (ERC-8004) to verifiable credentials (Dock, Reclaim). The design space is wide and the stakes are high: How much can your agent spend before it needs your approval? Can it commit you to a contract? Can it delegate authority to a subagent? These guardrails will likely be decided at the account layer.

Wallets. Of course, your agent will need a wallet in order to spend. Coinbase, Safe, Metamask, Phantom, MoonPay, and Privy (and many more) are all building here, offering features such as programmatic access and creation, delegated authority, per-transaction spending limits, whitelisted recipients, and the ability to operate across multiple chains without human approval for every action. This is one of the most crowded categories in the stack, which raises the question of where the moat is and whether it’ll be commoditized.

Payments

The payments layer is deep enough in the stack that it should be invisible to the end user, yet every dollar in the machine economy will flow through it. When Joe’s agent paid $0.24 to pull data from 40 providers overnight, he didn’t choose a card network, currency, or chain to settle each transaction on. The challenge is that traditional payment rails were designed for humans clicking “buy” buttons, not agents making thousands of sub-cent API calls per minute. Card networks carry a fixed cost floor of ~$0.03-0.04 per transaction plus 2.3-2.9% interchange. That works for a $400 hotel room, but not for the new class of multi-step agentic activity. The result is a wave of new protocols and currencies purpose-built for agent commerce, alongside incumbents adapting their existing infrastructure to satisfy these needs.

Here’s what you should know:

Rails. These are the protocols and standards that define how an agent initiates, routes, and settles a payment. Two approaches have emerged:

1. x402 (Coinbase/Cloudflare) and MPP (Stripe/Tempo) are built for machine-native commerce, where an agent hits an endpoint, gets a price, signs a payment, and receives the data in a single HTTP request, settling in stablecoins for fractions of a cent.

2. ACP (OpenAI/Stripe), AP2 (Google/PayPal), and Visa’s TAP take a different approach, adapting existing card infrastructure for agent use. These handle higher-value purchases where buyer protection and merchant acceptance matter more than speed of settlement and cost.

Stablecoins & Settlement. Agents need money that is programmable, fast, cheap, and global. Stablecoins check every box, which is why they’re the obvious currency for x402 and MPP transactions; at the same time, card rails still offer buyer protection and merchant familiarity that matter for higher-value purchases. The blockchains underneath (e.g, Base, Solana, Tempo) add another dimension: which chains can handle the throughput, finality, and cost structure that agent-scale commerce demands?

Facilitators. These are the intermediaries that sit between agents and merchants, handling the complexity that comes with compliance checks, merchant integration, and authorization. Coinbase, Stripe, and PayPal are extending their existing ecosystems to support agent transactions, betting that their merchant networks and compliance infrastructure give them an edge. Others, like Sponge and Sapiom, are solving the cold-start problem from the emerging merchant side, making it trivial for any API-based business to start accepting agent payments. As the number of rails, protocols, and merchants grows, facilitators may become the connective tissue that keeps the system from fragmenting.

AI Engine

This layer doesn’t need much introduction. It powers every agent interaction, every reasoning step, every tool call. But the economics of this layer are shifting faster than any other in the stack, and where value ends up is less obvious than it seems. We’ll focus on two categories:

Compute & Hosting. Every time Joe’s agent reasons through a task, calls a tool, or spawns a subagent, it consumes compute. But inference is only part of the equation. The explosion of vibe-coded apps and agent-built services is flooding the web with new endpoints that all need to be hosted somewhere. As of May 2025, the number of web pages available to us grew by 45% in just the previous two years, and that pace will only accelerate as coding agents make it trivial to ship new services. That means demand for compute is growing on both sides – more agents reasoning through more tasks, and more services spinning up to serve them.

The hyperscalers (AWS, Google Cloud, Nvidia) are the obvious players, with AWS and Google Cloud also making it increasingly easy to deploy agent backends and APIs on their infrastructure. Cloudflare targets the edge, offering serverless compute with minimal latency for agent-facing services. Decentralized alternatives like Akash, Bittensor, and Nous are positioning themselves to serve the excess compute demand by aggregating global GPU supply and selling it at a fraction of the cost.

Foundation Models. These are the brains. While Anthropic, OpenAI, Google, and Meta are the frontier labs pushing the boundaries of what agents can do, the cost of running these models is falling fast. Running a GPT-4-class model cost roughly $20 per million tokens in late 2022; however, by early 2026, equivalent performance is available for around $0.05 per million tokens, a 600x decline in just over three years. Hardware improvements, competition between providers, and optimization techniques like prompt caching and batching are all compounding to push inference costs down. At the same time, the cost of creating intelligence is falling dramatically as reasoning patterns are distilled into smaller open-weight alternatives running at a fraction of the cost. Open-weight models have closed the performance gap with closed models to just 1.7% on some benchmarks.

This is great for the machine economy.

Cheaper intelligence means cheaper agents, which means even a 24-year-old solo-founder in Vermont can afford to run them – this means more activity flowing through every layer above. If models compete on price the way cloud providers do today, the value may accrue to the layers above and below the model rather than the models themselves.

Who Wins?

By 2030, most of your digital interactions won’t involve a browser, a search engine, or a storefront. You’ll describe what you want, and your agent will handle the rest, finding the right service, negotiating the terms, paying, and delivering the result. The internet will look fundamentally different.

Think SEO for agents. More APIs, less human-facing interface.

In that world, who captures value?

Sam Ragsdale from Merit Systems wrote a piece comparing today’s agentic commerce landscape to the early internet. His argument was that curated agent marketplaces being built by the major platforms (ACP, UCP, TAP) are following the same playbook as AOL in the ‘90s – polished, controlled, and fundamentally limited by the fact that every merchant has to be hand-selected and approved. Open protocols like x402 and MPP are the scrappier alternative, but they’re permissionless. Anyone can build an endpoint and start earning from agents without a business development team or a legal review. In the ‘90s, the walled garden had the better product but the possibilities for the open web were limitless.

The open web won.

The same dynamic is playing out now. ACP, UCP, and TAP will integrate with frontier AI labs and serve the top use cases well, but an agent restricted to a curated catalog of pre-approved vendors can only do what the platform anticipated. An agent with access to an entire universe of open protocols can do much more.

Keep in mind, the most exciting parts of today’s web came from the longtail access to open websites via HTTP.

Some humility is required to admit that we can’t envision most of what the open agentic internet will become. Just like no one in 1995 was predicting ride-sharing or social media, we can’t predict the applications agents will build and pay for when we give them the tools they need.

As we discussed earlier, foundation models are commoditizing fast and the value may move to other layers of the stack. Tooling, wallets, and identity infrastructure are critical but will likely commoditize as standards converge. That leaves three areas where I believe value will concentrate: the interface, payments, and compute.

Interface

The interface shapes spending limits, approval flows, and trust delegation. The platform that designs the most personalized experience for the user will see the most transactions flow through it.

Apple is the most under-appreciated player here. Their devices are so deeply embedded in people’s daily lives that switching costs are enormous. If Siri evolves into a competent agentic interface, Apple controls the starting point for billions of transactions without needing to build the best model. They just need to maintain the best surface.

Google faces a harder transition. The shift from humans browsing to agents curating will cannibalize ad revenue, their core business. But Google has something no one else does. They have decades of personal data across search, email, calendar, maps, and documents. Then there are the enterprise switching costs to think about. Google Workspace is embedded in millions of companies, and every employee’s email, files, and workflows live on Google’s infrastructure. If any company can build the most personalized agent, for both consumers and enterprises, it’s Google. The question is whether they can monetize that agent as effectively as they monetized search.

Merit Systems is my wildcard. They’re building both the discovery infrastructure (AgentCash, x402 Scan, MPP Scan) and the consumer interface (Poncho) for the open agentic economy. Their thesis is that whoever controls where agents find services AND sits in the flow of funds captures the position Google had over the early web. It’s an ambitious bet, but if open agentic commerce wins over curated, Merit is the company best positioned to be the aggregation layer. They’re early, but so was Google when it was competing against AOL’s $350 billion (in today’s dollars) walled garden.

Payments

Whoever sits in the money flow takes a cut of everything. I have the highest conviction in this layer because it scales directly with transaction volume.

Stripe and Tempo are best positioned for machine-native payments. Stripe already has the developer ecosystem and extensive network of merchants. And with features like streaming payments (sessions), ~500ms finality, streaming payments via payment channels, native card and stablecoin support, gas fees paid in USD (no volatile tokens), and server-sponsored transactions, Tempo is built for the transaction volumes that the machine economy will generate. If MPP becomes the default machine-native rail, Stripe/Tempo captures a slice of every agent transaction.

Circle will scale with the growth of the agentic economy. I believe stablecoins become the settlement layer for the machine economy, and as that happens, Circle will take a cut of every dollar sitting in agent wallets through yield on reserves. USDC is the most widely accepted stablecoin across exchanges, wallets, chains, and payment protocols, so new builders default to it first, which deepens the integration, which makes it even harder for alternatives to break in.

Visa will adapt. Remember how Joe topped up his balance with Apple Pay using his credit card, and it converted to stablecoins under the hood without him ever seeing a wallet or thinking about a blockchain? That’s what the future will look like. Consumers will keep using the cards they’re familiar with, while stablecoins handle settlement under the hood. Visa will leverage their brand trust with consumers and merchants as the underlying rails evolve.

Compute & Hosting

More agents means more inference. More vibe-coded services means more hosting. Compute providers win regardless of which models, protocols, or interfaces dominate. AWS and Cloudflare are the two best positioned players here, for similar reasons.

First, they already support most of the internet. AWS holds ~30% of global cloud infrastructure across 37 regions globally. Cloudflare sits in front of over 20% of all websites as the security and performance layer, meaning every request to those sites already passes through them. When the supply side explodes with new agent-facing endpoints, the default deploy target is wherever developers already build.

Second, they’re building the monetization infrastructure for the new web. As ad models decline and the are replaced by pay-to-access, both companies are enabling that transition natively. Cloudflare has already launched pay-per-crawl, allowing any site behind its network to charge AI crawlers for access via x402 (Stack Overflow is already using it). On the other hand, AWS is a founding member of the x402 Foundation and has published open-source serverless x402 reference architectures. Any service running on either platform will be able to turn on agent-native monetization with minimal effort.

Identity

I’m bearish on companies like Worldcoin who are building for a world where every interaction requires human verification. That maximalist vision assumes people will care whether they’re dealing with a human or an agent online, but we’re already getting used to it. To me, the more likely future is that most web traffic is filtered not by human credentials but by micropayments.

Pay-to-access will prove more useful than prove-you’re-human.

Identity will matter for a subset of high-stakes interactions, but for the bulk of agent commerce, the (micro)payment itself will be the trust signal.

Conclusion

Joe didn’t wake up thinking about payment rails or agent identity protocols. He checked his phone, saw that his agent executed a trade, booked his meetings, and found him a cheaper server. Every layer of the stack we’ve discussed in this piece was abstracted away, and he never thought about any of it.

We’re still building towards this future. The protocols are live but underadopted, the supply side is growing but thin, the discovery problem is unsolved, and the identity layer is fragmented. Most of the transactions happening today are developers testing, not agents transacting; however, the pieces are falling into place faster than the metrics suggest. The people writing off the early infrastructure today are looking at a chart that’s been down-only. I’m thinking about what the chart will look like when every human has an agent or swarm of agents that are true economic actors.

If you haven’t already, pivot to agentic commerce.

Special thanks to @kleffew94, @0xfishylosopher, and @brendan_j_ryan for your thoughtful feedback!

Disclaimer: The authors of this content, as well as affiliates of Artemis Analytics, may have financial interests in the equities or tokens mentioned. This does not constitute investment advice or a recommendation to buy, sell, or hold any asset. The information provided is for educational purposes only and should not be relied upon for financial, legal, or tax decisions. Readers should assess their own circumstances before making any financial choices. Views expressed may change without notice, and Artemis Analytics is not liable for any losses resulting from the use of this content.*

Welcome back to Artemis’ Weekly Digital Finance Fundamentals!

Risk-off across the board this week. The S&P 500 dropped −1.7%, the Nasdaq fell −2.4%, and the 2-year Treasury yield climbed to 4.02% as the Fed dot plot shifted to zero cuts for 2026. Oil & Gas remains the YTD leader at +40%, though commodities cooled – gold at ~$4,430, silver at ~$68 (down 44% from its January ATH). In private markets, Kalshi raised $1B at a $22B valuation (Coatue) and Mastercard acquired BVNK for $1.8B.

Crypto followed equities lower. Bitcoin fell −5.8% to ~$66,500, Ethereum −7.5% to ~$1,987, Solana −7.5% to ~$83. AAVE (−9.2%) and MNT (−10.6%) were the hardest hit, while HYPE (−2.9%) held up relatively better. The clear standouts: TAO surged +19.7% on a decentralized AI training milestone and a Jensen Huang endorsement, and STG spiked +46.7% in what appears to be a technically-driven move rather than the fundamentals.

Today We Highlight:

• Tempo Mainnet Launch & the Machine Payments Protocol (MPP)

• Bittensor (TAO): Covenant-72B and the Decentralized AI Training Milestone

• Circle (CRCL) Crashes 22% Potentially Due To Changes In The Clarity Act

1. Tempo Mainnet Launch & the Machine Payments Protocol (MPP)

On March 18, Tempo – the payments-focused Layer 1 incubated by Stripe and Paradigm – launched its mainnet. In its first 8 days, stablecoin supply grew from zero to $6.3 million, with $14.3 million in cumulative transfer volume.

For comparison, other recent chain launches attracted significantly more stablecoin capital in their first week: Monad $242M (Nov 2025), Base $75M (Aug 2023), MegaETH ~$36M (Feb 2026). Those chains also saw much higher turnover—Monad at ~18.5x velocity and Base at ~15x, driven by DeFi activity and liquidity incentives. Tempo came in at ~2.6x. However, the gap makes sense: Tempo isn’t competing for DeFi TVL. Its flows are payments, not AMM swaps.

Launching alongside the chain is the Machine Payments Protocol (MPP), which can be thought of as a universal checkout layer for AI agents. When software needs to pay for an API call, a compute job, or a data query, MPP handles the request, authorization, and settlement. 100+ services are integrated at launch, including Alchemy, Dune Analytics, Merit Systems, and Parallel Web Systems.

Early MPP usage: ~49,000 transactions, 1,156 buyers, and 61 sellers (service providers) in 8 days. Dollar volume is tiny (~$7,300) because these are microtransactions by design.

2. Bittensor (TAO): Covenant-72B and the Decentralized AI Training Milestone

Bittensor’s TAO token surged +26% over the past 7 days and +102% over the past month, pushing above $350 with daily volumes exceeding $900 million and vaulting it to the 26th largest cryptocurrency by market cap. The catalyst: Covenant-72B, a 72-billion-parameter large language model fully trained on Bittensor’s decentralized Subnet 3 by over 70 contributors using commodity internet hardware.

Covenant-72B was pre-trained permissionlessly across 70+ global contributors and achieved a 67.1 MMLU score, confirmed in a March 2026 arXiv paper as the largest decentralized LLM pre-training run on record. The result is significant because it demonstrates that meaningful model training can happen across a decentralized network of commodity hardware, rather than requiring centralized data center clusters.

The rally accelerated after Nvidia CEO Jensen Huang responded to investor Chamath Palihapitiya’s description of the breakthrough by calling Bittensor “a modern version of Folding@home.” On the institutional side, Grayscale’s S-1 filing from December 2025 to convert its Bittensor Trust into a spot TAO ETF remains pending and represents a potential catalyst for regulated institutional access.

3. Circle (CRCL) Crashes 22% Potentially Due To Changes In The Clarity Act

Circle’s stock fell −22% on the week ($126 → $98), with the bulk of the damage coming on March 24 when CRCL dropped 20% in a single session—its largest one-day decline since listing in June 2025. Coinbase, Circle’s primary distribution partner, fell 8% on the same day before partially recovering.

The catalyst might’ve been a report that the latest text of the Clarity Act proposal in Congress would prohibit platforms from offering yield “directly or indirectly” for holding a stablecoin in a manner that resembles a bank deposit. That’s a direct threat to Circle’s business model around USDC rewards and to the broader embedded yield thesis—if platforms can’t pass yield through to end users, the competitive dynamics of stablecoin distribution change materially.

ARK Invest's Lorenzo Valente pushed back: the Clarity Act doesn't ban Circle from paying distributors like Coinbase — it bans retail passthrough of yield. Circle pays Coinbase over $900M/year in revenue share, roughly half of Circle's total revenue. That relationship is untouched. The bull case: if Coinbase can no longer share yield with users but still collects it from Circle, distributor margins on USDC improve — strengthening their incentive to grow supply. The bear case: without yield as a user acquisition tool, platforms lose a key reason retail holds USDC over alternatives. ARK bought $16 million of CRCL on the dip.

Adding to the policy uncertainty: David Sacks’ 130-day term as Trump’s “crypto czar” officially ended on March 26. Sacks is transitioning to co-chair the President’s Council of Advisors on Science and Technology (PCAST) alongside Michael Kratsios, a role that covers a broader technology mandate but moves him meaningfully further from the direct policy influence he held as the administration’s point person on crypto and AI.

Charts of the Week

Rapid-fire visual data points.

• Onchain options volume showed signs of scaling. DeriveXYZ surpassed $600 million in weekly volume for the first time last week, approaching the $1 billion mark. The inflection suggests structured products and volatility markets are beginning to find product-market fit onchain.

  

Other Notable News

• Mantle became the fastest chain to reach $1 billion in Aave deposits (18 days), compared to Polygon (24 days) and Ethereum (58 days, excluding early plasma/Avalanche deployments). Last week, Mantle’s Aave deployment ranked as the 3rd largest by deposits, overtaking Arbitrum and Base. Contributing factors include Mantle Vault yield on Bybit, a large treasury, and Tether’s tokenized gold/USDT0 integration.

Thanks for reading! Stay ahead this week by using the Artemis Terminal or pulling live data with =ART() in Excel.

Disclaimer: The authors of this content, as well as affiliates of Artemis Analytics, may have financial interests in the protocols or tokens mentioned. This does not constitute investment advice or a recommendation to buy, sell, or hold any asset. The information provided is for educational purposes only and should not be relied upon for financial, legal, or tax decisions. Readers should assess their own circumstances before making any financial choices. Views expressed may change without notice, and Artemis Analytics is not liable for any losses resulting from the use of this content.

Main Takeaway: Tokens with growing user bases, reasonable fee valuations, and active revenue dynamics systematically outperform those without, delivering a 1.73 Sharpe over 4+ years.

Executive Summary

• We constructed a weekly-rebalanced long-short factor from on-chain fundamentals: DAU growth, revenue quality, revenue stability, and valuation mean reversion.

• Adding risk overlays (BTC regime filter, volatility targeting, drawdown scaling) reduces the Sharpe to 1.73 but cuts max drawdown in half to -10.9% (as opposed to 1.83 Sharpe and -21.20% max drawdown without overlays).

• Fundamentals 1 is effectively market neutral: market beta is 0.05 (not significant) with an R² of 0.6%. Less than 1% of the factor’s return variation is explained by broad crypto market direction.

• Fundamentals 1 performs slightly better in down markets than up markets (+4.53% vs +4.22% average monthly return), demonstrating the diversification value of long-short fundamental selection.

• Alpha is statistically significant at the 1% statistical significance level across all layers, with +51.7% annualized alpha for the raw factor after controlling for market exposure.

Goal

The purpose of this model is to determine if it’s possible to systematically identify which crypto tokens will outperform and underperform in the next week based on their on-chain fundamental health.

The underlying thesis is that on-chain data (fee revenue, users, revenue stability) can tell you something about future token returns that price and volume alone won’t capture. Tokens with growing user bases and reasonable valuations relative to their fundamentals tend to outperform while tokens with declining usage or bloated valuations tend to lag.

To learn more about our methodology, request detailed testing results, or explore commercial opportunities, please reach out to the Artemis team.

Process

The Fundamentals 1 model constructs a cross-sectional long-short factor using on-chain fundamental data to rank and trade liquid crypto tokens. The portfolio is rebalanced weekly using simple USD returns, with the backtest starting in February 2022 (earliest data with sufficient scored assets meeting all eligibility and data quality criteria).

Data is sourced from Artemis Analytics for all on-chain fundamentals (fees, revenue, daily active users, market capitalization) and token pricing. Bridge flows data is sourced from Artemis.

Returns are gross of transaction costs; realized performance would depend on execution costs and liquidity at time of trade. The breakeven transaction cost is estimated to be 138 bps per trade.

The Fundamentals 1 factor portfolio is constructed from eligible assets at each time period (weekly) per the following criteria:

• ≥ $100M market cap at rebalancing period

• ≥ $10M daily trading volume at rebalancing period

• Minimum asset history of 13 weeks before eligible for inclusion

• Excluded: stablecoins, pegged assets, tokenized treasuries, exchange entities, index tokens, and infrastructure aggregates

This ensures that illiquid and microcap tokens whose volatility would distort factor returns are not included. The criteria produces a universe of about 112 eligible assets per week, of which about 60 have sufficient on-chain data to receive a composite score.

Signal Construction

To pick the signals for our fundamental factor model, we start with 25 signals across momentum, value, and quality indicators and 13 flow-based signals using bridge data. Many signals were eliminated because of negative predictive power redundancy measured by high correlation (several signals were 85%+ correlated with each other), or poor predictive power.

Ranking by predictive strength, filtering for redundancy, and testing every combination left us with four signals: DAU growth, inverted active revenue share, revenue stability, and MC/fees mean reversion. These signals are nearly uncorrelated with each other (all below 0.12), span three categories, and the four-signal composite outperformed every broader combination.

The four signals are computed as follows for each eligible asset:

• DAU Growth: Momentum signal measuring the percentage change in daily active users (DAU) over a trailing window; captures user adoption momentum.

• Inverted Active Revenue Share: Measures the proportion of revenue derived from passive sources (MEV, liquidations, incentive programs) rather than active user transactions. Tokens with lower active revenue share tend to outperform in the short term, likely reflecting capital attracted by incentive programs.

• Revenue Stability: Measures the inverse volatility of revenue growth; less stable growth is associated with higher forward returns, consistent with a risk premium.

• Market Cap/Fees Mean Reversion: Measures the negative z-score of the market-cap-to-annualized-fees ratio against its own history and identifies tokens that are cheap relative to their own fundamental valuation history.

Each signal requires a minimum of 8 non-zero observations within a 13-week eligibility window. Raw signal values are winsorized at the 1st and 99th percentiles to limit outlier influence, then converted to cross-sectional rank z-scores each week. Rank-based normalization is used to handle the heavy-tailed distributions of typical crypto data.

Portfolio Construction

The four rank-scores signals are then combined into an equal-weight composite score. An asset requires a minimum of 2 of 4 signals to receive a composite score. The composite is then re-ranked to produce the final asset ranking each week.

Assets are then sorted into quintiles based on the composite ranking. The portfolio goes long the top quintile (~12 assets with highest composite score) and short the bottom quintile (~12 assets with lowest composite score) with equal-weighting within each leg.

Risk Overlays

There are three layers of risk management applied in sequential order to the raw long-short portfolio:

1. BTC Regime Filter: Binary filter based on Bitcoin’s 20-week simple moving average. When BTC is trading below its 20-week SMA, all positions are scaled to zero. This reduces exposure during broad crypto drawdowns.

2. Volatility Targeting: Position sizes are scaled each week so that the portfolio’s trailing 8-week realized volatility targets 15% annualized. A maximum leverage cap of 1.5x is applied.

3. Drawdown Scaling: Exposure is linearly reduced starting at a -10% drawdown from peak equity, reaching zero exposure at -20% to protect against sustained losses.

On the position level, there is a 20% maximum weight per individual asset and 40% maximum allocation to any single sector. Rebalance changes below 2% are skipped to reduce any unnecessary turnover.

Performance & Results

Backtested Performance

Since inception at the start of 2022, we can look at the return of Fundamentals 1 without the risk overlays (raw), with only the Bitcoin regime filter, and with all three risk overlays (Bitcoin regime, volatility targeting, drawdown scaling). Note that the regime filter was triggered for late 2025 and all of 2026 as BTC has been below its 20-week SMA since roughly November 2025.

The results are backtested and gross of fees and transaction costs. In practice, transaction costs, borrowing costs for shorts, and potential liquidity constraints could all reduce net returns. We estimate the breakeven transaction cost, which is the per-trade cost the strategy can absorb while still netting positive returns, to be approximately 138 basis points.

The three layers demonstrate the tradeoff between raw return and risk management. The raw long-short strategy delivers the highest annualized return (+60.3%) but carries 33% volatility and a -21.2% max drawdown. Adding the BTC regime filter cuts volatility to 22.5% and keeps Fundamentals 1 flat during broad crypto drawdowns, at the cost of roughly a third of the raw return. Layering on volatility targeting and drawdown scaling compresses volatility further to 18.1% and cuts the max drawdown nearly in half to -10.87%, while improving the Sharpe ratio from 1.58 with only the regime filter to 1.73 Sharpe with the additional overlays. The full overlay stack gives up return for a significantly smoother profile as the Calmar ratio of 2.88 (return per unit of max drawdown) is the highest of all three layers.

Looking at yearly performance, Fundamentals 1’s strongest years were 2022 and 2023 with raw returns of +126% and +122%, respectively. This is notable because 2022 was a deep crypto bear market where BTC fell from $47K to $16K yet Fundamentals 1 was able to deliver its best annual performance. Performance moderated in 2024 (+15% raw) and 2025 (+40% raw). As the crypto market matured and more capital flowed into fundamentals-based strategies, the spread between best and worst tokens compressed and the signal weakened, although it stayed positive.

Annual performance was markedly different after adding the risk overlays. In 2022, the regime filter held Fundamentals 1 flat for most of the year (due to BTC bear market), reducing the returns to +19% (from +126% raw). This shows how the filter protects against drawdowns but also removes Fundamentals 1 from periods where the long-short spread is the widest. In 2023, BTC recovered and returns increased (+78% with regime vs +122% raw). By 2024-2025 the overlay returns exceeded the raw returns (2025 full overlay returned +41% vs raw 28%). This happened because the regime filter successfully kept Fundamentals 1 flat during losing weeks that the raw signal traded through and the vol targeting scaled up exposure during lower-volatility periods when the signal was performing well.

Today, the regime filter has been triggered since November 2025. The raw signal is slightly negative (-5.2%) while the overlay is keeping Fundamentals 1 flat during current market weakness.

Analysis Against the Market

We can compare the performance of Fundamentals 1 with and without the overlays to market (using BTC as proxy for broad crypto market exposure).

Fundamentals 1, both with and without the risk overlay, outperformed the market. In terms of monthly outperformance, the Raw Long/Short strategy beats BTC in 30/50 months (60%) while adding all risk overlays beats BTC in 28/50 months (56%).

Additionally, one of the core advantages of a long-short factor strategy is the ability to generate returns regardless of market direction. To test this, we split the backtest into months where BTC was positive (26 months) and months where BTC was negative (24 months).

In BTC positive months, the raw factors averaged +4.22% per month while BTC averaged +14.02%. The factor underperforms in rallies because the short leg acts as a drag when most tokens are rising; an expected cost of market neutrality.

However, in BTC negative months, the raw factor averaged +4.53% per month while BTC averaged -10.36%. Fundamentals 1 is not only positive during drawdowns but actually performed slightly better in down months than up months. This is because the short leg of tokens with weak fundamentals tend to fall harder than the long leg during selloffs, generating positive spread returns.

This asymmetry is what makes Fundamentals 1 valuable as a portfolio diversifier over a standalone directional bet, and why Sharpe is so high relative to just going long Bitcoin.

Market Independence

To test Fundamentals 1’s independence of broad market dependence, we regressed the on-chain fundamentals factor’s weekly returns against BTC.

The raw long-short factor has near-zero market beta (0.05) and an R² of 0.6%, meaning less than 1% of the factor’s return variation is explained by whether crypto is going up or down. A t-test on the null hypothesis that beta equals zero fails to reject (p > 0.05), confirming that the long-short construction is effectively market-neutral.

After applying the risk overlays, market beta rises slightly to 0.08 with an R² of 4.7%. In this case, the t-test rejects the null that beta equals zero (p < 0.05), meaning the overlay version does carry a small but statistically significant market exposure. This is expected as the BTC regime filter mechanically introduces some market correlation by only allowing positions when BTC is above its 20-week SMA. The tradeoff is a meaningfully reduced drawdown portfolio.

Annualized alpha is +51.7% for the raw factor (t-stat=3.16, p=0.002) and +27.0% after all risk overlays (t-stat=3.07, p=0.002). Both are statistically significant at the 1% level, indicating Fundamentals 1 generates returns that can’t be attributed to market exposure.

Factor Correlation

Weekly return correlation between the raw factor and BTC is +0.075, rising to +0.22 with risk overlays applied. For context, the November 2025 factor model analysis found cross-factor correlations ranging from -0.07 to +0.29, so the Fundamental 1’s market correlation falls within the range of existing independent factors.

Key Takeaways

• On-chain fundamentals predict cross-sectional crypto returns. Tokens with growing user bases, reasonable fee valuations, and active revenue dynamics systematically outperform those without, delivering a 1.73 Sharpe over 4+ years.

• Fundamentals 1 is independent of market direction. With near-zero market beta and an R² of 0.6%, the factor captures something genuinely different from whether crypto is going up or down. Fundamentals 1 was most profitable during the 2022 bear market.

• Risk overlays improve the risk profile without destroying alpha. The full overlay stack cuts max drawdown from -21% to -11% and improves the Calmar ratio to 2.88, the highest of all three layers.

• Signal strength has moderated in recent years. 2022-2023 raw returns averaged +124% annually versus +28% in 2024-2025. The alpha is still positive, but the opportunity set compressed as the market matured.

• Fundamentals 1 adds diversification to existing crypto portfolios. Its low correlation to market (+0.075) and positive returns in down markets make it valuable as a complement to directional crypto exposure.

If you have any questions, feel free to reach out to team@artemis.xyz or shoot us a DM on X.com @Artemis.

The entire stablecoin trade got repriced in hours. The catalyst was a new draft of the Clarity Act that would effectively kill passive stablecoin yield.

The story goes deeper than a single red day, though. There’s a regulatory fight, a business model vulnerability, and a wallet freeze that poured gasoline on an already burning stock.

The Clarity Act Bombshell

On March 20, Senators Thom Tillis (R-NC) and Angela Alsobrooks (D-MD) announced an agreement in principle on stablecoin rewards, with White House backing. The full text was reviewed by crypto industry leaders in a closed-door Capitol Hill session on Monday.

The key provision: passive stablecoin yield, earned simply for holding a dollar-pegged token, is banned. Exchanges, brokers, and their affiliates would be prohibited from offering yield directly or indirectly on stablecoin balances, or in any manner “economically equivalent to interest.”

Activity-based rewards tied to payments, transfers, or platform usage would still be allowed. The SEC, CFTC, and Treasury would jointly define permissible rewards and anti-evasion rules within one year. Note the SEC and CFTC recently announced a historic memorandum between the agencies, ending years of interagency fighting and discord.

Congress just put in writing the line the banking lobby has been drawing for two years. Stablecoins can be payment instruments, but they cannot be deposit substitutes.

One industry leader at the closed-door session, reported via an internal stakeholder email obtained by Eleanor Terrett, described the text as a “departure” from what had been previously discussed with the White House. They warned that the “economic equivalence” standard is deliberately vague and could be interpreted far more restrictively by future regulators.

This Hits Circle Harder Than Anyone

95.5% of Circle’s revenue currently comes from interest income on USDC reserves, which explains the sell-off.

CRCL issues USDC, holds the reserves in short-duration Treasuries and overnight repos, and earns the spread. In Q4 2025, reserve income was $711MM, up 60% YoY, driven by 97% growth in average USDC supply. FY2025 revenue was $2.7B, up 64%.

The Clarity Act doesn’t directly attack Circle’s reserve income (CRCL earns that yield itself) but it is a direct attack on its demand engine. Platforms like Coinbase currently pass stablecoin yield through to users as an incentive to hold USDC. Coinbase’s stablecoin revenue hit $1.35B in 2025, up from $910MM in 2024. If exchanges can no longer offer yield on USDC balances, the incentive for users to hold USDC over a traditional bank deposit weakens considerably.

Less yield-sharing means less USDC adoption means less reserves means less interest income for Circle.

The timing makes it worse. Reserve yields already declined from 4.49% in Q4 2024 to 3.81% in Q4 2025 as the Fed cut rates. While the market is no longer pricing in cuts this year, Circle’s interest income was already under pressure before this bill showed up.

USDC Fundamentals Have Never Been Stronger

The stock cratered on a day when USDC’s underlying metrics are at all-time highs:

• Circulating supply: $81B as of late March, up from $76B at year-end 2025.

• Onchain transaction volume: $6.8T (adjusted) in Q4 2025 alone, up >2x YoY.

• Market share vs USDT: USDC volume has surpassed Tether’s USDT every month since August 2025, now capturing >80% share in 2026.

Q4 earnings beat: $770MM in revenue vs. $745MM expected. EPS of $0.43 beat Street consensus by 23%.

Circle also just announced an expansion into Africa through a Sasai Fintech partnership and secured a major Intuit integration.

The Wallet Freeze Poured Gasoline on the Fire

Circle froze the USDC balances of 16 business hot wallets late Monday, disrupting operations at several exchanges, casinos, and ForEx platforms including FxPro, Pepperstone, AMarkets, and HeroFX.

The freeze reportedly stemmed from a U.S. civil case whose details haven’t been disclosed. @zachxbt raised pointed concerns, noting that anyone with basic onchain tools could identify these were operational business wallets processing thousands of transactions. He warned that opaque freezes based on undisclosed civil actions risk turning USDC into “a politicized gatekeeping tool.”

The power to blacklist and even wipe frozen addresses is explicitly coded into USDC’s smart contracts.But on a day when the market was already questioning centralized stablecoin risk, the optics could not have been worse.

There’s Still a Bull Case

The sell-off prices in a worst-case reading of the Clarity Act. There are a few things worth considering on the other side.

Activity-based rewards survive. The bill draws a line between passive yield (banned) and transaction-based incentives (allowed). Coinbase and others are already exploring workarounds: marketing incentives, activity-based payments, issuer partnerships that blur the line between interest and rewards. The “economic equivalence” standard is vague, which means it will get lawyered.

Coinbase’s P&L may not change much. Coinbase largely passes stablecoin yield through to users, so the revenue is often offset by expenses. Analysts argue the direct earnings impact is limited. The bigger question is whether restrictions slow long-term USDC adoption.

This bill isn’t law yet. The committee markup isn’t expected until the second half of April, after the Easter recess. There’s still time for lobbying, amendment, and negotiation. Brian Armstrong has been notably silent on the latest text, but his previous positions suggest Coinbase will fight hard on the “economic equivalence” language.

Non-reserve revenue is growing fast. Platform services, transaction processing, and other non-reserve income grew >15x YoY to $37 million in Q4, with full-year other revenue reaching $110 million. Still small relative to interest income, there is now a diversification story.

The Setup From Here

CRCL was up 170% from its February lows before today. The stock rallied from $50 to $127 on blockbuster earnings, the USDC/USDT flippening, and the Intuit partnership. However, the valuation had already priced in flawless execution on interest income, AI-driven payments, and tokenization, leaving zero room for regulatory disappointment.

At $101 now, CRCL trades at roughly 9x annualized revenue. The contention now is whether the Clarity Act kills the USDC growth flywheel or forces it to evolve. If stablecoin adoption keeps accelerating on the back of payments, cross-border settlement, and institutional use (onchain data is still positive), then Circle’s reserve income engine stays intact regardless of whether Coinbase can offer yield on idle balances.

Disclosure: This material is provided for informational purposes only and does not constitute investment advice, financial advice, trading advice, or any other form of advice. The views expressed are those of the authors and should not be relied upon as a recommendation to buy, sell, or hold any asset. The authors or affiliated entities may hold positions in the assets discussed. You should conduct your own research and consult appropriate financial professionals before making any investment decisions.

Market Overview: The Weekly Recap

This week, dispersion continued, with clear winners in HYPE and CRLC (up 36% and 100% in the past month, respectively). The mean return across key assets was a negligible -0.3%, reflecting that even though there were winners this week there were also significant losers, with macro continuing to be the major factor. Gold has been retracing as 2 year treasury yields have sharply inflected to 3.9%, up from 3.7% earlier in the week. Gold now sits 13% below highs printed in early March, while silver is 37% below its January highs. The moves in bonds and commodities are direct outputs of the escalating war in the Middle East. The S&P500 ended the week down 2.52%, including a -1.5% move on Friday following announcements that the Pentagon is seeking a $200bn budget for the war, and is considering deploying additional troops (Politico).

Today We Highlight:

• HIP-3 Updates: SP500 partnership, Crude trading, HYPE rerating

• CFTC and SEC put forward new guidance amid CLARITY Act delays

• Tempo Mainnet goes live, announce MPP

1. Hyperliquid ecosystem growth continues with S&P partnership

Hyperliquid has continued to remind us that great products work regardless of market sentiment. HYPE has gapped up +50% from $26 to $39 in the past several weeks, as HIP-3 markets have reached all-time highs in open interest and have emerged as a leading venue for trading commodities and macro assets outside traditional market hours.

Growth in HIP-3 markets is primarily coming from the XYZ protocol which makes up ~85% of open interest. Further driving this momentum is the announcement that trade[XYZ] has partnered with S&P to launch the first official S&P 500 perpetual contract. This means that trade[XYZ] will now operate the first and only S&P500 perpetual future contract, “enabling 24/7 onchain access to the world’s most iconic benchmark” (S&P Global Press Release).

The economics of HIP-3 markets are as follows: the deployer of the market must lock up 500k HYPE, users experience 2x the standard fee, and Hyperliquid receives 50% of that fee, with the rest going to the operator.

HYPE’s recent rerating indicates the market is valuing it on significantly higher revenue multiples, as we are seeing price and revenue diverge. Historically, we have observed price and annualized revenue exhibit significant comovement, and instances of divergence tend to be short lived. Given HIP-3's trajectory, we believe revenue is more likely to catch up to price than the reverse.

The valuation growth can be explained by the higher future earnings potential of the protocol given the success of HIP3 markets and the diversification of assets being traded on the platform. One major criticism of protocol revenues in DeFi is that they are highly cyclical, tied to the price of bitcoin, and therefore of lower quality. Commodities, equities and indices make up >80% of trading volume on HIP 3 markets, largely driven by speculation due to the war in the Middle East and other macro events. Hyperliquid is challenging the DeFi narrative by bringing onchain the trading of assets uncorrelated to the price of bitcoin. Because of this, it deserves to command higher multiples on better quality revenue.

2. CFTC and SEC put forward new guidance amid CLARITY Act delays

“We are not the securities and everything commission anymore” said SEC Chairman Paul Atkins during his speech at the DC Blockchain Summit. On March 17th, the CFTC and SEC released what is possibly the most consequential US digital asset regulatory document ever issued. As chairmen Selig and Atkins said, the intent is for this to act as a bridge until the CLARITY Act is enacted.

What is it?

The 68-page guidance, issued by the SEC and endorsed by the CFTC, provides a taxonomy for crypto assets unlike any we’ve seen before. The taxonomy, at a high level, is as follows:

• Digital Commodities: Primarily includes L1 assets like ETH, AVAX, APT, but also includes SHIB which is a memecoin, and DOGE.

• Digital Collectibles: This category primarily includes NFTs like Crypto Punks, Chromie Squiggles, but also includes WIF which is a memecoin.

• Digital Tools: A category inclusive of blockchain-based representations of assets with practical functions like tickets, memberships, credentials, or identity badges.

• Stablecoins: Covers stablecoins as defined under the Covered Stablecoins definition given in Staff Statement on Stablecoins, until GENIUS goes into effect, at which point it adopts the definition used therein.

• Digital Securities: Any assets that are traditional securities tokenized on a blockchain, plus tokens that qualify as investment contracts under Howey. Importantly, the guidance introduces the “attach and detach” concept — a token can start as a digital security and graduate out of that classification once the issuer fulfills its promises or the network decentralizes.

Why is this important?

Since the inception of crypto assets, there has not been a clear regulatory framework for digital assets in the United States. This has led to the offshoring of digital finance innovation, and the SEC performing “regulation by enforcement” on US-based companies (see Uniswap, Coinbase, etc.). By issuing this guidance, the US allows operators to innovate more comfortably and effectively, with much better understanding of where the boundaries are. While this guidance is extremely important, it is interpretive, not statutory. This means that it has not yet been fully codified into law, and a future, less-crypto friendly commission could rescind or reinterpret this guidance.

3. Tempo launches Mainnet, announces Machine Payments Protocol (MPP)

Stripe is one of the internet’s most dominant payments processors, valued at $158bn as of February 2026, and processing nearly $2 trillion in payments volume in 2025 - surpassing incumbent PayPal for the first time. This week, as interest in agentic commerce and payments continues to accelerate, Stripe finally launched their Tempo Layer 1 blockchain on mainnet, alongside the launch of their agentic payments protocol: Machine Payments Protocol.

Interest in topics relating to agent to agent commerce has risen dramatically over the past 6 months, and x402, a protocol spearheaded by Coinbase has become the most popular choice for developers. While x402 provides a crypto native mechanism for agentic payments, MPP extends this framework and allows for payments on traditional payment rails like credit cards or wallets.

So what?

Stripe has been betting big on crypto since their October 2024 $1bn acquisition of Bridge. Since then, they have also acquired Privy, a major crypto wallet infrastructure business, developed Tempo, their own Layer-1 blockchain purpose built for payments, and added crypto to Stripe checkout. With the announcement of MPP, it’s becoming clear that Stripe is becoming more than just a payments processor for businesses, it is building the payments layer that will power commerce for all of digital finance - on or off chain. Stripe already processes payments for millions of businesses worldwide — meaning MPP doesn't need to build distribution from scratch. Any merchant already on Stripe's rails can plug into agentic commerce without touching crypto infrastructure, a moat x402 simply doesn't have.

Charts of the Week

• 2 Year Treasury yields spike amid Iran war and macroeconomic volatility

• Crypto ETF Flows have been positive for the past 4 weeks, the first time this has happened since July 2025. March is shaping up to be the first month of positive net flows since October 2025. (Note the weekly aggregation in the chart below does not include the current period)

• Stablecoin supply makes new all time highs this week at >315bn

Other Notable News

• Key senators may have finally come to terms on how to deal with stablecoin rewards in the CLARITY Act, though this is yet unconfirmed at the time of writing (Politico).

• Figure announced Figure Forge, a tokenization service that converts real-world assets like auto loans into fungible “Participation Tokens” usable as collateral within DeFi protocols (X/mcagney).

• Mastercard acquires stablecoin startup BVNK for $1.8bn (Reuters)

• Kalshi raises $1bn at $22bn led by Coatue, doubling valuation from December’s raise (Bloomberg).

Thanks for reading! Stay ahead this week by using the Artemis Terminal or pulling live data with =ART() in Excel.

Disclaimer: The authors of this content, as well as affiliates of Artemis Analytics, may have financial interests in the protocols or tokens mentioned. This does not constitute investment advice or a recommendation to buy, sell, or hold any asset. The information provided is for educational purposes only and should not be relied upon for financial, legal, or tax decisions. Readers should assess their own circumstances before making any financial choices. Views expressed may change without notice, and Artemis Analytics is not liable for any losses resulting from the use of this content.

Context

Citrini offered 2 lines on what the future of agentic payments looks like.

“Agents went looking for faster and cheaper options than cards. Most settled on using stablecoins via Solana or Ethereum L2s, where settlement was near-instant and the transaction cost was measured in fractions of a penny.” - THE 2028 GLOBAL INTELLIGENCE CRISIS

However, this was written before the launch of Tempo Mainnet on March 18th and misses a chunk of what the new financial system looks like for investors.

Artemis is a digital finance research / data platform used by firms like McKinsey, Visa, Sequoia, and Paypal. Our team comes from BlackRock, Citadel, Whale Rock, Insight and Ribbit Capital. We are a data partner to Tempo and believe we have the best window into what the future of investing, payments and finance looks like and how blockchains impacts that future. We share our view of what 2030 looks like in the essay below.

It’s October 30th, 2030.

S&P just hit another all-time high and broke 10,000. 10.1% of the US population lacks formal employment — AI-induced job displacement led to the rise of millions of humans starting small businesses and becoming creators.

Investors who love fundamentals have seen 50% drawdowns in once “blue chip” names, turning them into contrarian value investments. Dispersion, which began to increase in late 2025 with the advent of advanced AI, continues to skyrocket.

Nvidia is a $20T company and now makes up over 20% of the S&P 500. OpenAI and Anthropic make up over 10% of the S&P 500. While broad beta has done well, buoyed by individual stocks in the index like these, stock picking has done even better.

It’s never been a better time to be an investor.

Steven, a 25 year old graduate from Georgia Tech, wakes up before going to his $90,000/year job at an AI robotics company in Atlanta. He was laid off from Intuit where he was a PM – more companies started moving towards using agents to manage their accounting books – but Steven found his job at the robotics startup shortly after given the immense demand for product talent.

It’s 8am ET. The US markets aren’t open yet, and as Steven prepares for work, he gets pinged by his macro agent “Stan” who advises him to long oil and gold given Polymarkets odds of another conflict in the Middle East surged to 60%.

Steven accepts “Stan’s” suggestion and the agent deploys $5k longs on oil and gold on the HIP-3 market on Hyperliquid. Another agent, the L/S analyst which he calls “Brad” recommends he execute a long / short strategy to long Microstrategy and short BTC given MSTR just started trading below <1 mNAV. Steven asks Brad to execute the trade on Robinhood.

An hour later, he’s at his desk figuring out how to best optimize the new fleet of autonomous farming robots they just launched in Georgia. At lunch, he gets a ping from Artemis that a leading Silicon Valley firm is interested in digging into the autonomous mobility space. Trying to stay anonymous, Steven logs into his Phantom wallet with his PFP of his favorite NBA player Cam Boozer and hops on a Zoom link with a Listen Labs agent that grills him on his view on the competitive landscape for autonomous robots in the US.

After the call, he receives $1,000 from the VC firm in his Phantom Wallet – his agent then off ramps to USDC, and that then routes to Kalshi where his funds are deposited into markets with 95%+ odds. His wealth manager agent that he calls “Charles” has been allocating a majority of his savings to his neobank, where half of his funds are in STRC paying him 10%/year and the rest are earning him 5% on his USDC.

This is the day in the life of a high agency investor in 2030. Steve’s annualized returns far outpace SPY and QQQ and regularly beat benchmarks by double digits.

And he’s not even done yet.

Steven goes home after work to his apartment to be with his partner and decides he wants to write a piece on why he thinks Stripe will be a $10T conglomerate by 2035.

He logs into Artemis and pulls the relevant financials, news, and market maps to put out a long form piece on his thesis. Steven’s fundamental analyst agent “Warren” pays $2.50 via Stripe’s Machine Payment Protocol (MPP) to a sell-side analyst to get their financial model. He then tweaks the assumptions to publish a price target and model. He publishes his long piece and model on X, LinkedIn, and Artemis and immediately gets paid $250 for publishing – GPs and other agents inquire to chat with him about his thesis. Steven responds to these agents and he continues to get paid in his Phantom wallet via Tempo’s MPP for expert advice. Stripe inches up 2% after his post as hedge funds, agents and retail buy into Steven’s thesis. Stripe now has a market cap of over $2T. Steven wakes up the next day with another $1,000 in his wallet. 5 GPs pinged Steven on Artemis if they could invest in his personal fund.

Steven is not unique and is one of hundreds of millions of people in the world who love investing and are taking their financial future in their own hands in 2030. They didn’t have to go to Stanford, some fancy business school or work at Goldman Sachs to have the same tools and investing strategies as the best investors.

Steven now has access to financial tools that didn’t exist 10 years ago – his funds live in a US based fintech founded in 2026 that uses Aave to earn him 4-5% APR for his idle USDC, he primarily uses Phantom (founded in 2021) to interact with HIP-3, Hyperliquid (founded in 2023) to execute trades, and Openclaw (founded in 2025) and Anthropic (founded in 2021) to power his investing agents. The prediction markets he uses to trade and invest in hadn’t existed even 5 years prior. The payments he receives from hedge fund managers for his research go through Tempo, which didn’t even exist 5 years ago.

This brave new world that Steven lives in allows him to be a world class investor, despite no “formal” investing experience.

Steven, who doesn’t come from wealth, now has “Charles”, “Brad”, “Stan” and other analysts at his fingertips and is able to support himself, his partner and his parents financially. At age 25, Steven is now effectively a solo PM managing the 6 figures he’s been able to save and earn. He now has a technical analysis analyst, a macro analyst, a fundamental analyst, a quant analyst, and a prediction analyst who ping him on Telegram on suggestions on what he should do next.

He’s able to trade / invest and earn in his day job to provide for his family and loved ones just from being high agency and curious.

Investing has changed because of digital finance.

Digital Finance in 2030

At Artemis we believe neobanks, lending, agentic payments, prediction markets and trading are the most enduring markets enabled by blockchains and completely changes investing in 2030.

We call these the 5 core pillars of Digital Finance:

• Neobanks changes because individual investors now have one interface and system of record that can interact with all their finances whether assets on blockchains like stablecoins or fiat currency and can send money through any payment rail (ACH, Wire, RTP, Stablecoins, etc)

• Lending changes because humans will always need capital to do things and borrow lend protocols allow anyone access to world class savings yield or access to credit.

• Agentic payments changes investing because our agents will need to pay for resources at machine speed, and the current internet has no way to monetize the explosion of non-human traffic. There will be a new class of merchants in this agentic web.

• Prediction markets changes because investors can finally express beliefs in the world that they previously couldn’t with just spot / equities.

• Trading changes because investors can now express their beliefs in the world without staring at charts when agents can and tap into 24/7/365 markets on perpetual exchanges and trade on their behalf.

Today we share our thesis and our winners / losers in these new markets of finance. Here’s a teaser of what’s to come research at Artemis:

Neobanks - Alex Weseley

Steven’s neobank is where his entire financial life exists. It is the single financial surface that aggregates investing, saving and getting yield, obtaining credit, and sending money globally. This was only made possible because public blockchains allow developers to interact directly with bare metal financial plumbing, while legacy rails like ACH and Fedwire remain the connective tissue to the existing economy. Permissionless finance means Steven can receive an ACH direct deposit straight into his stablecoin account, and immediately lend it on Morpho to get 5%.

We are seeing this shift happen today in real time: Kraken has a Master Account at the Fed, NuBank is establishing a chartered bank in the US, and Coinbase has applied for a bank charter. Neobanks are attacking from two sides: integrating vertically across the legacy stack while building on the emerging permissionless stack. A neobank that owns both the legacy rails and the emergent stack can offer financial products that firms existing in only one stack cannot offer.

What changed at the infrastructure level has had a profound impact on Steven at the surface. It has unlocked access to financial products previously reserved for the wealthy and connected: yield on cash, credit against assets, 24/7 asset management and global money movement. Access to more sophisticated tools means a savvy person like Steven can afford to support his family because his money is always working. Wealth used to compound for the wealthy - tomorrow’s neobanks will make it compound for everyone.

Winners: Robinhood, Revolut, Coinbase, Nubank, Ethereum, Solana, Morpho, Hyperliquid

Losers: Legacy financial institutions unable to innovate and acquire young customers

Who to look out for: Column, Lead Bank, IBKR, Stripe

Neobank Market Map: March 2026 Neobank Market Map

Lending - Mario Stefanidis

Lending in 2030 is going to be a story of blockchain rails and AI risk assessment. The old model of human loan officers, rigid FICO scores and slow approval cycles was built for a world where assessing risk was expensive and moving capital was slow. AI makes underwriting cheap while the blockchain makes settlement instant.

The entire lending stack, from origination to capital funding, is being rewritten by software that prices risk faster, distributes credit through embedded channels and settles on programmable rails. The winners are picks and shovels names that own the intelligence layer or the settlement rail rather than those that rely hose rely on their balance sheet.

This is arguably the sector where new technology hits hardest. The core function of deciding who gets capital and on what terms is pure decision-making, and that is exactly what gets cheaper as AI scales. But the other half of the equation is equally important: moving that capital efficiently once the decision is made.

The old lending stack was vertically integrated inside banks. GSIBs had a moat because they could originate, underwrite, service and fund the loan all under their roof, maximizing margin. They were in bed with regulators and their cumbersome manual processes made their stack extremely difficult to reproduce.

What is happening now is a full unbundling and rebundling of that stack across five layers: origination, distribution, underwriting and risk, capital and funding, and infrastructure.

At the origination layer, digital-native lenders like SoFi, Upstart and Affirm have already pulled volume away from legacy incumbents by offering faster, fully digital experiences across consumer, mortgage, SMB and crypto-backed lending.

At the distribution layer, BNPL and embedded finance are pushing credit to the point of purchase. Affirm and Klarna are lending distribution networks sitting inside checkout flows.

At the underwriting layer, AI credit scoring from Upstart and Pagaya is expanding the addressable borrower pool by evaluating thousands of variables legacy models ignore.

At the capital layer, DeFi protocols like Aave (which has surpassed $1 trillion in cumulative loans), Morpho (now partnered with Apollo Global) and Maple Finance ($3.2B TVL, up from $500M) are creating permissionless, 24/7 capital markets that fund loans without a bank in the middle. Liquidity protocols allow lenders and borrowers to transact in stablecoins and to earn interest on pledged assets.

On the infrastructure side, blockchain settlement is collapsing what used to take days into seconds. Figure Technologies is originating HELOCs and first-lien mortgages natively on-chain, cutting out the layers of intermediaries that made traditional mortgage closing slow and expensive. SoFi’s Galileo/Technisys stack and lending-as-a-service platforms provide the remaining plumbing.

Winners: SoFi, Upstart, Aave, Figure, Affirm

Losers: Legacy banks, failed CeFi lenders that blew up in 2022, declining DeFi protocols that never found product-market fit, TrueFi stagnating at minimal TVL

Who to look out for: Creditas, Morpho, Fluid, Ledn, Figure (new loan products)

Lending Market Map: March 2026 Lending Market Map

Agentic Payments - Lucas Shin

As it currently stands, the internet is built for humans. It uses rate limits, CAPTCHAs, and API keys to filter out non-human activity, and monetizes human users with ads. But with the proliferation of autonomous agents, the economics break. More traffic – fewer eyeballs. Servers that have historically been subsidized by ad revenue will face orders of magnitude more requests from users who will never be influenced by ads. Agentic payments solve this naturally, with micropayments as the access key. Pay-to-crawl.

As our trust in AI accelerates, our agents will begin making hundreds of sub-cent API calls per hour on real-time data and tooling, buying what they need to carry out the goals we’ve prompted them with. Unlike humans, agents won’t be limited by speed, and they won’t ever need to rest. If even a fraction of the world’s billion knowledge workers utilize agents spending a few dollars a day on their behalf, we’ll see an agent-initiated transaction surface measured in the hundreds of billions annually. When agents start selling to each other — packaging already-solved problems as cheap endpoints for other agents — that number will quickly compound.

We see all of that value concentrating to the user interface, the payment infrastructure, and the merchant ecosystem.

The interface layer includes search engines, aggregators, and the user-facing surfaces that humans actually interact with. Unlike the ad-supported web, where Google aggregated free content and monetized through ads, the agent web will be transactional by default. Whoever controls the user has the leverage, ultimately controlling how agents discover, choose, and pay for resources.

Payment infrastructure spans the stablecoins, blockchains, and facilitators that will process every agent-initiated transaction. Circle and Tether will provide the currency. Stripe, Visa, and Coinbase are building the bridges between these new rails and existing financial infrastructure. Whoever can build the most seamless financial experience for agents will sit directly in the flow of funds and capture the value created by this emerging market.

The merchant layer will be made up of both existing products that become accessible to agents and entirely new categories of products and services that have yet to emerge. With companies like Stripe and standards like x402 lowering the barriers to monetization, any developer, business, or agent will be able to offer paid services. Over time, agent-driven demand will give rise to entirely new markets, much like the internet did.

The tools for this new internet are already being built.

Winners: Circle, Visa, AWS, Cloudflare, Stripe

Losers: Google, Meta, Salesforce, Phantom, Apollo, ZoomInfo, Worldcoin

Who to look out for: Tempo, Arc, Merit Systems

Agentic Commerce Market Map: March 2026 Agentic Commerce Market Map

Prediction Markets - Kaviish Sethi

Prediction markets sit at the intersection of finance and information. Their core function is to simply convert beliefs about future or certain outcomes into prices. When participants risk capital on an outcome, they reveal what they truly believe. The resulting price of an outcome isn’t just a probability; it’s a live signal of the collective belief.

The internet is full of predictions, but none are properly incentivized. There is no cost of being wrong so simply looking at opinions is very shallow. Prediction markets fix this by attaching financial consequences to forecasts. Accuracy is awarded and inaccuracy is penalized, overtime creating a signal which is far more reliable than polls or other static models. This turns prediction markets into an infrastructure that continuously aggregates information.

Prediction markets were historically constrained by poor distribution, low liquidity, and slow settlement, which prevented them from becoming reliable information systems.

The path forward is more crypto native systems. Crypto removes these barriers by enabling global participation, permissionless market creation, and composability allowing liquidity and users to aggregate in ways regulated platforms cannot. As AI agents and automated systems demand real time, reliable forecasts, prediction markets become the primary source. This convergence is what turns prediction markets from a niche to a core infrastructure.

By 2030, prediction markets evolve from niche venues into core financial infrastructure for pricing uncertainty.

At the platform layer, activity spans regulated exchanges like Kalshi and CME, crypto native markets like Polymarket and sportsbook or tradfi entrants like DraftKings, Robinhood and Coinbase. What began as fragmented venues converges into a unified surface for trading probabilities.

At the distribution layer, brokerages such as Robinhood, Interactive Brokers and Coinbase, alongside aggregators and APIs like Polymarket API and Kalshi API, embed prediction markets directly into financial interfaces. Forecasting becomes native to consumption, not a separate destination.

At the liquidity layer, institutional market makers including SIG, Jump, Jane Street, DRW, Wintermute and Citadel provide continuous two sided pricing. Probabilities become a new asset class with tight spreads and scalable depth.

At the data layer, analytics platforms like Artemis, and Dune, alongside media and research such as Bloomberg and CoinDesk, standardize and distribute prediction signals. Market implied probabilities become core inputs for decision making across finance, media and AI systems.

At the oracle layer, systems like UMA enable reliable resolution of real world events, while benchmarks and pricing providers further formalize probability as a measurable financial signal.

At the infrastructure layer, settlement chains such as Polygon, Base and Solana, alongside clearing systems and compliance rails, form the backbone. On chain rails tie the entire stack together, unifying market creation, liquidity, settlement and data into a single global system. Capital moves permissionlessly, markets operate 24/7 and probabilities become portable financial primitives that can be traded, embedded, collateralized and integrated across applications.

At the capital layer, regulators including the CFTC and SEC, along with investors such as a16z, Paradigm and Sequoia, provide the oversight, capital and legitimacy required for markets to scale globally.

In this structure, prediction markets resemble a derivatives system built not on prices, but on likelihoods. Their importance is not just in the contracts themselves, but in the continuous signal they produce, a globally accessible, financially backed probability that becomes the most efficient way to aggregate and price information about the future.

Winners: Kalshi, Polymarket, Robinhood

Losers: Traditional polling and forecast models, small prediction platforms without liquidity or distribution, and sportsbooks that stay limited to sports betting

Who to look out for: Robinhood and other platforms building in house prediction markets

Prediction Markets Market Map: March 2026 Prediction Markets Market Map

Trading - Zheng Jie Lim

Trading in 2030 becomes huge for a dark reason: more people will feel behind and cannot afford to be passive. The old brokerage model was built for a world where investing was slower, more intentional and separate from the rest of a person’s financial life. Previously, we had one account for banking, investing and maybe another for crypto. The market had fixed hours. Product menus were narrow. Acting on a view took more steps.

That world is changing.

AI compresses work. AI will widen the knowledge gap first, then the execution gap and eventually the wealth gap. The people with the best tools, data and distribution will learn faster, act faster and compound faster. For everyone else, the pressure shows up as a feeling: the gap is widening faster than salary can close it. Social media makes that impossible to ignore. Other’s people upside becomes your benchmark. Patience starts to feel less like wisdom and more like a slow way to fall behind.

That is why trading gets bigger. Trading platforms become the place where users hold cash, react to news, express macro views and turn information into action.

The old trading stack was fragmented across banks, brokerages, exchanges, and crypto apps. Each owned one narrow part of the user journey. That separation made sense when products were siloed and users tolerated more friction.

What is happening now is a rebundling of that stack across four layers: the access layer, the execution venue, the liquidity layer, and the market infrastructure underneath.

At the access layer, platforms like Robinhood, Coinbase, Webull, and eToro are trying to become much more than brokerages. Robinhood already combines equities, options, prediction markets, cards and cash features in one app. Coinbase is moving beyond spot crypto into stablecoin yield, payments and broader financial access. The company that owns the interface owns the habit.

At the execution layer, product breadth will matter. Hyperliquid shows what a fast, crypto-native trading venue looks like. Robinhood is widening access across asset classes. Coinbase remains a key venue for retail crypto activity. The winners are the one that let users move across products without leaving the ecosystem.

At the liquidity layer, market makers, internalizers and order routers determine whether users get tight spreads and reliable execution. This layer is less visible and matters as trading volume grows and more products become tradable.

At the infrastructure layer, exchanges, clearinghouses, custody and settlement systems still matter because trusted matching and settlement is needed. ICE remains important because if more of the world becomes tradable, the owners of key market infrastructure still capture value.

In 2030, a 29-year-old college graduate might get paid via direct deposit into Coinbase, Robinhood, or Hyperliquid. They keep part of their cash in a yield-bearing stablecoin and use the same platform to buy US equities, express macro views, and trade crypto perps.

Trading will become a cultural outlet for the ambitious and insecure that need to catch up. Some may still call it investing but for many users, it is about trying to close the wealth gap before it widens further. The platforms that win will understand this clearly: they are selling speed, hope and the feeling that you still have a shot to make it to the upper middle class.

Trading will keep growing exponentially 2030 and beyond

Winners: Robinhood, Coinbase, ICE, Kraken

Losers: Legacy Brokerages - as they are slower, less native and harder to innovate

Who to look out for: Interactive Brokers, Webull, SoFi, eToro, Hyperliquid

Trading Market Map: March 2026 Trading Market Map

Conclusion

Today we shared a brief window into what the world looks like for the high agency investor — we look forward to continuing to share deep dives in the core sectors by our world class analysts.

• Who wins in the future of trading: Hyperliquid or Robinhood?

• Who wins in the future of prediction markets: Kalshi or Polymarket?

• Who wins in the future of neobanks: Phantom or Nubank?

• Who wins in the future of lending: Aave or SoFi?

• Who wins in agentic payments: x402 or MPP?

We tackle these questions in a series of essays in the coming weeks.

Disclaimer: The authors of this content, as well as affiliates of Artemis Analytics, may have financial interests in the equities or tokens mentioned. This does not constitute investment advice or a recommendation to buy, sell, or hold any asset. The information provided is for educational purposes only and should not be relied upon for financial, legal, or tax decisions. Readers should assess their own circumstances before making any financial choices. Views expressed may change without notice, and Artemis Analytics is not liable for any losses resulting from the use of this content.

Special thank you to Son Do, Wei Guo, Vincent Jow for the feedback on our piece. Thank you Freda Duan for inspiring how investing unfolds.

Welcome back to Artemis’ Weekly Digital Finance Fundamentals!

This week, crypto clearly led returns. HYPE was the standout, rallying +18.8% over the past 7 days, while Figure Technologies (+13.1%) and Circle (+11.7%) also posted strong gains. Among majors, Ethereum (+5.2%), Solana (+4.7%), and Bitcoin (+4.7%) all moved higher, with Uniswap (+4.2%) and SKY (+8.3%) adding to the broad-based strength across digital assets.

Notably, crypto outperformed traditional equities this week. While digital assets and crypto-linked names were broadly green, Coinbase (-0.9%) and Robinhood (-5.0%) lagged, showing that equity performance was more mixed beneath the surface. Meanwhile, traditional benchmarks finished lower, with the S&P 500 down -1.5% and the Nasdaq 100 down -1.0% on the week. Overall, the tape suggests risk appetite rotated back toward crypto, with tokens and select crypto-adjacent names materially outperforming broader public markets.

Today We Highlight:

• HIP-3 hits record high OI and Volume, Driven by oil futures

• Two Consecutive Weeks of Positive Crypto ETF Flows

• Tokenized Gold Supply hits 1.2M oz onchain

1. HIP3 Hits Record OI and Volume, Driven by Oil Futures

HIP-3 markets hit a fresh record in open interest this week, with total OI reaching roughly $1.3 billion on March 12. The move was overwhelmingly driven by trade.xyz, which alone accounted for about $1.2 billion of OI, while smaller venues like Dreamcash and HyENA added incremental depth.

More important than the headline number is what drove it. Oil’s share of total HIP-3 open interest rose to 31% by March 14, after sitting near negligible levels for much of January and February. In less than two weeks, oil went from being a minor driver of activity to one of the ecosystem’s largest sources of demand.

That shift suggests HIP-3 is evolving beyond a crypto-native long-tail experiment into a real venue for permissionless macro expression. When offshore traders want fast exposure to oil, indices, and event-driven volatility, capital is increasingly showing up onchain. If that trend continues, HIP-3 could become one of the clearest examples yet of onchain markets beginning to absorb global commodity flow.

2. Two Consecutive Weeks of Positive Crypto ETF Flows

Crypto ETF flows stayed positive for a second straight week, adding another sign that institutional demand is rebuilding. In the week ending March 8, net inflows totaled $609.9 million, led by Bitcoin ETFs (+$568.5 million), while Ethereum (+$23.5 million) and Solana (+$22.0 million) also saw healthy demand. Aside from a small -$4.1 million outflow from Ripple products, flows were broadly constructive.

The bigger takeaway is that this no longer looks like a one-week bounce. After weeks of pressure, back-to-back positive prints point to a market where allocators are gradually stepping back into crypto exposure.

The macro case for why those flows could sustain runs directly through the Iran conflict.

(Hayes’ iOS Warfare is the most detailed articulation.)

The macro backdrop may also remain supportive. Geopolitical stress, higher oil prices, and shifting expectations around Fed easing are pushing investors to think more seriously about hard assets and alternative stores of value. If that macro uncertainty persists, ETF demand could remain firm as allocators continue rebuilding crypto exposure.

3. Tokenized Gold Supply Hits 1.2M oz onchain

Tokenized gold supply continued its breakout this week, reaching roughly 1.2 million ounces onchain, equivalent to about $6.1 billion in supply. The move marks a sharp acceleration from levels below $2 billion less than a year ago, underscoring how quickly investor demand for blockchain-based hard asset exposure is scaling.

In a market increasingly shaped by macro uncertainty, tokenized gold appears to be benefiting from both sides: the traditional safe-haven bid into gold itself, and growing comfort with onchain wrappers as a distribution rail.

The bigger takeaway is that tokenization is no longer just a payments or stablecoin story — it is now extending meaningfully into store-of-value assets, with gold emerging as one of the clearest real-world winners.

Charts of the Week

Rapid-fire visual data points.

• Prediction markets continued to gain traction this week, with total open interest reaching roughly $1.3 billion, led by Kalshi and Polymarket. Traders increasingly used event markets to express views on politics, macro, and geopolitical volatility in real time. The setup is notable because it shows demand expanding beyond pure crypto price speculation and into a broader market for real-world information and probabilities.

Other Notable News

• Mastercard launches a broad crypto partner program. On March 11, Mastercard rolled out a new crypto partner initiative spanning more than 85 firms, including Binance, Circle, Ripple, PayPal, Gemini, and Paxos. The program is designed to support cross-border payments, B2B transfers, and global payouts using stablecoins and digital assets across networks like Solana, Polygon, Avalanche, and Aptos. It marks another step toward plugging blockchain-based payment rails directly into mainstream financial infrastructure.

• Coinbase explores a potential investment partnership with Bybit. Reports on March 14 indicated that Coinbase is in talks with Bybit on a possible investment or strategic partnership. If completed, the relationship could help Bybit move closer to the compliant U.S. market while giving Coinbase exposure to one of the largest offshore exchanges. More broadly, the discussions reinforce a growing trend: major global exchanges are increasingly aligning with regulated counterparts as market structure matures.

• BitGo was selected by SoFi to power infrastructure and institutional distribution for SoFiUSD. The partnership makes BitGo a core provider for issuance, custody, and distribution of SoFi’s bank-issued dollar stablecoin, giving institutional clients direct access to SoFiUSD through BitGo’s platform. It is another sign that regulated banks are moving from stablecoin experimentation toward live deployment, with crypto-native infrastructure providers supplying the backend rails.

• Kraken’s banking arm moves closer to direct U.S. payment rail access. Kraken Financial’s Fed master account win continued to stand out as one of the more important infrastructure developments in the space. Direct connectivity to Fedwire and related payment rails would lower fiat transfer frictions for institutions and represents another example of crypto firms gaining access to core financial plumbing rather than building entirely outside of it.

Thanks for reading! Stay ahead this week by using the Artemis Terminal or pulling live data with =ART() in Excel.

Disclaimer: The authors of this content, as well as affiliates of Artemis Analytics, may have financial interests in the protocols or tokens mentioned. This does not constitute investment advice or a recommendation to buy, sell, or hold any asset. The information provided is for educational purposes only and should not be relied upon for financial, legal, or tax decisions. Readers should assess their own circumstances before making any financial choices. Views expressed may change without notice, and Artemis Analytics is not liable for any losses resulting from the use of this content.

Welcome back to Artemis’ Weekly Digital Finance Fundamentals!

This week broke from the slow grind lower of recent weeks. The U.S. struck Iran over the weekend, BTC dipped to ~$63,000 on Saturday, then short-squeezed back to $73,000 by midweek before fading to ~$71,000 by Friday close, up +3% on the week. SOL (+3.6%) led the majors, ETH (+2.2%) and BNB (+2.3%) tracked closely. PI (+16.5%) ripped off its February all-time low as traders front-ran Pi Day (March 14) and a potential Kraken listing, though the structural overhang looms large: 187.5 million tokens unlock this month alone, with only 9.4B of a 100B total supply currently circulating.

Crypto-linked equities outperformed tokens this week. CRCL (+22.2%) extended its post-Q4 earnings rally ($770M revenue, +77% YoY) with Citi naming it a top pick. FIGR (+21%) bounced off its post-earnings ATL after releasing strong February operating metrics showing consumer loan marketplace volume up ~20% MoM and YLDS stablecoin traction. COIN (+11.1%) surged after Trump publicly sided with the crypto industry over banks on the CLARITY Act stablecoin yield dispute, with Armstrong visiting the White House ahead of the post.

HYPE processed $9.6B in 24h perp volume (+37% WoW) as the war-weekend proved the case for always-on onchain venues when TradFi was closed. RWA kept grinding: tokenized U.S. Treasuries crossed $11B, up 22% YTD. On the mining side, Core Scientific (CORZ) locked in a $500M–$1B Morgan Stanley facility to convert its mining sites to AI data centers and plans to liquidate its entire BTC portfolio to fund the pivot, joining Bitdeer and MARA in the ongoing miner exodus from BTC to compute.

Today We Highlight:

• Polymarket Pulls Record Volume as Iran Conflict Drives Prediction Market Activity

• Highest BTC ETF Inflows In 2026

• Strategy’s STRC saw explosive volume as shares traded above $100 par value.

1. Polymarket Pulls Record Volume as Iran Conflict Drives Prediction Market Activity

Within hours of U.S. and Israeli strikes landing in Iran on Saturday, Polymarket spun up over 200 active markets covering ceasefire timelines, regime change, and Strait of Hormuz closure. The flagship “US strikes Iran by...” contract pulled $529 million in total volume, making it one of the largest single markets in the platform’s history. A separate market on Khamenei leaving power by March 31 resolved at 100% after Iranian state TV confirmed his death, drawing $45 million in volume and the top trader cleared $757,000.

But the scrutiny has centered on activity before the strikes: onchain analysts flagged six wallets that netted $1.2 million by betting on a U.S. strike on February 28, the exact day it occurred, with most wallets funded within 24 hours of the attack. A separate account made $553,000 betting on Khamenei’s removal just before the Israeli strike that killed him. Under U.S. commodity trading law, profiting from bets tied to death and violence is illegal, but most of the activity occurred on Polymarket’s offshore platform, outside the reach of U.S. regulators.

The regulatory picture is layered. The same week this played out, CFTC Chairman Selig said prediction market guidance is coming “in the very near future.” State gambling regulators are simultaneously pressing their own authority. The Trump administration dropped two Biden-era federal investigations into Polymarket, while Donald Trump Jr. serves as an adviser to the platform and his fund 1789 Capital is an investor. Polymarket settles in USDC which means the prediction market boom is quietly one of the larger demand drivers for onchain stablecoin infrastructure as well.

2. BTC Records Highest Net Flows

Going into the week, spot Bitcoin ETFs had bled $4.5 billion YTD across six consecutive weeks of outflows. Then the tape flipped: $1.7 billion in inflows since February 24, with March 5 marking the strongest single day since the outflow streak began (~$500 million, 10 of 11 funds positive). The reversal lasted one day, with $227.9 million in net outflows returned by end of session, but the inflows appear to be outright bullish bets rather than basis trades.

The macro case for why those flows could sustain runs directly through the Iran conflict.

The pattern has precedent: the 1990 Gulf War, post-9/11, and the 2009 Afghanistan surge all coincided with Fed easing cycles. Brent is already up ~24% in a month to $85/bbl on Strait of Hormuz disruption, and the ISM manufacturing PMI came in at 52.4 for February, signaling expansion that complicates the near-term rate picture. Friday’s jobs miss (−92K, unemployment to 4.4%) shifted expectations forward, with June emerging as the earliest likely cut date and Polymarket odds for a June cut increasing.

The oil-to-bond-volatility channel hasn’t triggered yet — the MOVE Index sits at ~75, well below the 130 level that has historically preceded Fed intervention. But Friday’s jobs miss opened a second path to the same destination: if the labor market is weakening, the Fed has cover to cut on growth grounds rather than waiting for a bond market dislocation. Both channels end with more liquidity — the difference is timing and the state of the economy when it arrives. The market is now pricing both possibilities simultaneously: oil-driven inflation risk that delays cuts, and labor market deterioration that accelerates them.

3. Strategy’s STRC saw explosive volume as shares traded above $100 par value.

Stretch, Strategy’s preferred stock recently hit $100 which allowed Strategy to use its at-the-market program to sell shares above the $100 par value to fund Bitcoin purchases. STRC saw strong activity in trading volumes reaching $778M in cumulative volume this week.

According to STRC tracking website strc.live, it is estimated that roughly 3,730 Bitcoin will be bought using purely STRC.

The success of STRC, which now has $3.47B notional value, displays the new playbook that DATs like Microstrategy are adopting. The shift is gradually away from convertible debt or purely Common stock ATMs, but instead through a combination of the preferreds and common stock. The Bitcoin accumulation machine seems to be unstoppable, with many analysts on X predicting a massive buy from Strategy next week.

Charts of the Week

Rapid-fire visual data points.

• Arbitrum bridge volume surge — Arbitrum experienced a dramatic bridge volume spike, exploding from $15.1M on February 26 to $98.8M on March 5, a staggering +554% increase. The surge accelerated dramatically in the final 48 hours, jumping from $21M to nearly $99M, suggesting higher Hyperliquid interest

Other Notable News

• CFTC to bring perps onshore. On March 3, CFTC Chairman Selig said the agency is working to bring “true perpetual futures” to the U.S. within the next month. Perps represent over 90% of global crypto derivatives volume but have existed almost entirely offshore.

• SEC submits crypto securities framework to the White House. On March 3, the SEC sent Commission-level guidance on how federal securities laws apply to different types of crypto assets. More enforceable than staff memos and harder to reverse.

• Fed, FDIC, OCC clarify tokenized securities treatment. Banking regulators confirmed that tokenized securities receive the same capital treatment as their traditional equivalents if they confer identical legal rights. Removes a key blocker for banks considering tokenization.

• Solv Protocol exploited for ~$2.7M. A reentrancy attack drained 38 SolvBTC from a Bitcoin Reserve Offering vault via a double-minting flaw exploited 22 times. Solv offered a 10% bounty and said it will cover affected users.

Thanks for reading! Stay ahead this week by using the Artemis Terminal or pulling live data with =ART() in Excel.

Disclaimer: The authors of this content, as well as affiliates of Artemis Analytics, may have financial interests in the protocols or tokens mentioned. This does not constitute investment advice or a recommendation to buy, sell, or hold any asset. The information provided is for educational purposes only and should not be relied upon for financial, legal, or tax decisions. Readers should assess their own circumstances before making any financial choices. Views expressed may change without notice, and Artemis Analytics is not liable for any losses resulting from the use of this content.