I first bought Alibaba back in 2021 at $200, rode it all the way down to $60 in 2022 and sold it at $150 in 2025. Crazy journey.

My thesis back then was pretty simple: Alibaba had one of the strongest vertically integrated e-commerce ecosystems in China. Ant Financial was an incredibly valuable asset and Alibaba Cloud had the potential to become a major growth engine. My variant perception at the time was that the market had overreacted to China’s regulatory crackdown as it was temporary. Alibaba was priced like a permanently impaired business when the underlying business was stronger than that.

Looking back, I was right about some of them. Alibaba’s core commerce business didn’t disappear. Cloud did become increasingly important. Today, Alibaba is still one of China’s largest and most valuable technology companies.

I was also wrong a lot. I underestimated how much regulation and geopolitics could change the multiple investors were willing to pay for Chinese companies. I also underestimated how brutally competitive Chinese e-commerce businesses would become, with the emergence of PDD, JD, Douyin and Meituan. Ant Financial was the biggest lesson. I was wrong to assume that value would accrue back to Alibaba shareholders in the way I expected.

My biggest lesson is that being right about the business is not enough. You have to understand where the government wants the country to go and whether your investment is aligned with that direction.

That lesson is a big part of how I think about Alibaba today. My thesis today is different.

My thesis is: Alibaba is poised to become the biggest AI winner in China.

Alibaba trades at roughly $114 today. My bull case is that it can be worth $415/share by 2030 (roughly 3.6x or 35% IRR) from here.

My core assumptions are:

• Total Revenue will reach $300B (16% CAGR)

• AI Cloud & Compute Revenue will reach $120B (37% CAGR)

• AI Labs & Apps Revenue will reach $30B (90% CAGR)

• AI Labs & Apps grow from roughly 1% of revenue today to almost 10% in 2030

• AI Labs & Apps will be EBIT positive in 2030

• EBIT will be $60B (20% EBIT margin)

There are three parts to my thesis:

1. Global AI usage continues to compound and cheaper models continue to take share

2. Alibaba owns almost the full AI stack

3. Alibaba is long the broader China AI ecosystem

My variant perception is that the street is far too pessimistic on AI Labs & Applications profitability. The street models it as a loss-making segment all the way through FY38.

I think they’re wrong.

I think AI Labs and Applications reaches profitability by FY31 (CY2030) and grows toward 10% of Alibaba’s total revenue from just 1.2% as of Q1 ‘27.

1. Global AI usage continues to compound and cheaper models continue to take share

The world is going to consume a lot more AI. I don’t think that is a surprise to anyone. If we were to take a step back and look at the amount of people that still haven’t had access to AI, you realise how early we still are.

An estimated 5.75 billion people around the world have never used AI before. Only a tiny fraction of the world pays for AI (1.2% – 100 million people), an even smaller group uses AI to write code and an even smaller fraction runs agents at scale.

Chinese models went from basically irrelevant on OpenRouter to around 60% of routed token consumption in less than two years.

Why? I believe price is a big reason why. Chinese models are increasingly more competitive because most AI workloads don’t require the best model.

I think Qwen can become a huge top of funnel for Alibaba. Alibaba is trying to get developers to build on Qwen, businesses to deploy Qwen and users to interact with Qwen. Alibaba will figure out a way to monetize everything downstream of it (Cloud, ads, commerce, enterprise applications and apps).

2. Alibaba owns the almost the full AI stack

Most AI companies today only own one or two layers of the stack. Alibaba owns almost all of them (applications, APIs, models, data infrastructure, cloud and even the chips).

What makes this even more interesting to me is that Alibaba also owns ecommerce and payments. This creates an interesting flywheel between AI usage, distribution and transactions.

I believe this matters because platform wars are usually won by whoever owns the ecosystem.

Alibaba already has multiple places to monetize through its existing ecosystem (Cloud, Taobao, Alipay, Amap, Fliggy and enterprise applications).

Alibaba Cloud is already the clear market leader in China with roughly 40% market share in the cloud market. It has become the AWS of China.

There is another reason why I think owning the full stack matters in China.

China’s latest Five-Year Plan runs from 2026 to 2030. One of the biggest priorities in the plan is technology self-reliance (AI, chips, cloud infrastructure and getting AI deployed across the economy).

That lines up with what Alibaba is building.

Alibaba already has:

• Models: Qwen

• Chip design: T-Head

• Compute: Alibaba Cloud

• Applications: Qwen + Enterprise products

• Distribution: Taobao, Tmall, Alipay, Amap, Fliggy and DingTalk

3. Alibaba is long the whole China AI ecosystem

The part I think is even more interesting is that Alibaba owns more than just Qwen and its own AI stack. It also owns stakes in several major Chinese AI companies (Kimi, Zhipu, MiniMax, CXMT, Unitree).

I don’t think the market fully credits these stakes today. As these companies get bigger and some eventually IPO, their valuation will be much harder to ignore. So even if Qwen doesn’t end up being the #1 model in China, Alibaba still has exposure to many other potential winners.

Risks

China remains China.

The biggest risk to my thesis is still China.

Regulation, geopolitics and semiconductor restrictions can all affect Alibaba’s business and the multiples investors’ are willing to pay for Chinese companies. I learnt this the hard way in 2021.

Mitigant: China risk has not disappeared. What has changed is the alignment of incentives. In 2021, Alibaba’s growing influence in areas like fintech increasingly ran against Beijing’s priorities. Today, Alibaba is spending aggressively on AI (models, chips, cloud and compute infrastructure). These are areas China has explicitly made strategic priorities for in 2026 till 2030.

Why is the market mispricing this?

I think there are 3 reasons.

1. The costs are showing up before the revenue

Alibaba is spending aggressively on AI infrastructure today. Capex is rising, free cash flow is getting hit and AI Labs is still deep in loss-making.

The market has yet to price in what the infrastructure will earn several years from now.

2. AI monetization has yet to show up inside AI Labs

If Qwen drives more model usage, some of the economics show up in Alibaba Cloud.

Alibaba is already monetizing AI successfully but it hasn’t directly impacted the AI Labs line item.

3. AI Labs is still new and barely modelled

Consensus models AI Labs as loss-making all the way through FY38. Even by FY31, the mean estimate still has the segment losing roughly RMB 21 billion. I think that is far too pessimistic.

Street FY31E AI Labs EBIT: -RMB 21B

My FY31E AI Labs EBIT: positive

Conclusion

I think right now the market values Alibaba as an e-commerce company and a way to get exposure to China.

I think that misses the bigger opportunity.

Alibaba owns the models, cloud, applications, distribution and large parts of the underlying infrastructure.

Based on my FY31 revenue and EBITDA estimates and Alibaba’s own historical valuation range, I believe Alibaba can be worth $415/share by 2030.

Five years ago, I thought Alibaba was a misunderstood e-commerce business with valuable assets. Today, I think it has the opportunity to become the company that captures more of China’s AI value than anyone else.

Let’s see if I’m less wrong this time.

Every month, we highlight a leading analyst who takes a long-term view with investing.

This month we highlight David Kroger at T. RowePrice Digital Assets.

David Kroger is the Director of Digital Assets Research at T. Rowe Price, where he brings a systematic, data-driven approach to navigating the rapidly evolving digital asset space.

With a diverse background spanning venture capital, equity research, and data science, David specializes in bridging the gap between traditional fundamental analysis and the unique, high-velocity dynamics of crypto markets. His work focuses on integrating deep research, on-chain analytics, and active risk management to build institutional-grade investment strategies, such as the T. Rowe Price Active Crypto ETF (TKNZ).

Read on to learn more about David’s research philosophy, how he combines systematic data with discretionary judgment, and why he believes digital assets are becoming a critical layer of the next generation of financial infrastructure.

What is your story? What was your journey into investing and digital assets?

I started in venture capital right out of school, focused on AI, robotics, augmented reality, and virtual reality. That experience introduced me to alternative ways of analyzing companies and thinking about addressable markets. From there, I joined Piper Sandler as an equity research data scientist, where we used alternative data to analyze public companies.

I initially looked at digital assets as a way to practice and improve my data science techniques. The market was producing an enormous amount of observable data, along with more than enough volatility to test different ideas. A senior colleague and I built an algorithm that identified trending names based on their internet presence. We called it the Media-Derived Value algorithm, or MDV, and it surfaced names such as Bitcoin and GameStop. After that, I went headfirst into digital assets, first by writing research on the space and eventually moving into it full-time in 2021, when I led the research deck on digital assets at TD Cowen, then StoneX. I joined T. Rowe Price in 2025.

You’ve been at T. Rowe Price for over a year now as Director of Digital Assets Research. What was the path that led you to join T. Rowe Price, and why did you join the firm?

Over my years leading digital asset research on the sell side, I got to know a lot of great and very smart people across the industry. One of those people was Blue Macellari, the Head of Digital Assets at T. Rowe Price. She gave me the opportunity to join the firm and help build out its digital asset research and modeling capabilities, with the goal of developing what would eventually become our first exchanged-traded in this space, T. Rowe Price Active Crypto ETF (TKNZ).

What attracted me was the opportunity to combine T. Rowe Price’s long history of fundamental investment research, portfolio construction, and risk management with an asset class that is still developing in real time. Digital assets create a unique research challenge because you need to understand the technology, the economics, the data, and the market structure. T. Rowe Price had both the resources and the investment culture to approach that challenge seriously.

You were recently named a portfolio manager for the active crypto ETF that you helped launch. Congratulations. Can you tell us a little about TKNZ and why it is important to T. Rowe Price, to you personally, and to the industry?

Thank you. TKNZ is designed to be a one-stop shop for investors who want digital asset exposure without having to learn the mechanics of every asset, worry about security and custody, or continuously maintain and rebalance a portfolio as the market moves. It brings research, portfolio construction, execution, and risk management together in a familiar investment vehicle.

It is personally important to me because of the team behind it. I have the opportunity to work with Blue, as well as our co-portfolio managers Stefan Hubrich, Sean McWilliams, and Dante Pearson, all of whom bring deep and complementary expertise to the team’s modeling, portfolio construction, and fundamental research capabilities. For T. Rowe Price, the product extends the firm’s research-driven investment approach into a new asset class. For the industry, I think it represents another step toward moving digital asset investing beyond single-token exposure and toward a more diversified, institutionally managed framework.

It seems like you have always straddled the line between systematic and discretionary investing. What matters most for digital assets: factors, fundamentals, sentiment, or momentum?

We think about digital assets through three main pillars. The first is technology and economics, which is probably the closest equivalent to traditional fundamentals. That includes understanding how the token works, the security model, the supply and emissions schedule, the token’s utility, and how value may accrue to holders.

The second pillar is drivers of growth. That means understanding the sector and ecosystem, how easy the technology is to use, the strength of the team and developer community, and what improvement proposals or product changes are in the works that could materially affect the fundamentals. We then view those first two pillars through a thematic and momentum lens.

No single input matters most in every environment. There are assets with strong fundamentals that never attract an incremental buyer, and the opportunity cost of not participating in a popular narrative can be expensive - AI in equities is a good analogy. Our process is systematic in how we collect, organize, and compare the evidence, but discretionary in how we interpret that evidence and ultimately size the risk.

What is an investment thesis or core market belief that you hold, and what are the main drivers behind it?

One of my core beliefs is that digital asset markets tend to price different things in different market environments. Fundamentals and revenue adoption are often the fallback conversation in a bear market. Investors focus on revenue, usage, token economics, value accrual, and whether a project can survive a difficult environment.

In a bull market, total addressable market, user adoption, ecosystem growth, and the broader narrative tend to become more discussed. Investors are more willing to underwrite what something could become rather than only what it is producing today. The mistake is assuming that one framework works in every market. Our team can analyze digital assets on both fronts and use those insights to position accordingly. We want to identify assets with durable technology and economics, while also recognizing that adoption, liquidity, and narrative often determine when that value is realized.

Why digital assets right now in 2026? Given the growing demand for AI exposure, what could make investors more bullish on crypto and digital assets more broadly?

In my view, one reason to look at digital assets in 2026 is that the asset class increasingly overlaps with some of the largest technology and financial themes in the market, including AI, tokenization, prediction markets, and the movement toward 24/7 financial markets. AI agents may eventually need to hold assets, make payments, and transact with one another. Blockchains provide a natural, programmable settlement layer for that type of activity.

What could make investors more bullish is evidence that these use cases are turning into real economic activity. That could include more users, more on-chain transactions, greater adoption of tokenized assets, and stronger revenue generation. Digital assets are fighting for investment dollars that are also heading toward areas such as AI and prediction markets, but digital assets participate in both of those segments and may ultimately help enable them. We are already seeing the asset class influence traditional finance through areas such as tokenization and the expectation that more markets will eventually trade around the clock.

Investment research tools have been drastically affected by advancements in technology, particularly AI. What does your current investment research process look like, and how has it changed over the years?

I think one of the biggest changes in digital asset research has come through products such as Artemis, which streamline data processing and make the information easier to understand. I used to have to crawl Discord servers and read through multiple drafts of white papers just to find the most recent emissions schedule. We still do that from time to time when it is necessary, but now I can use API calls, or have an LLM write SQL code, to pull the relevant KPIs much more quickly.

I also have Excel plugins that update our models, relative analysis, and other inputs alongside larger macro variables. Even on the nonquantitative side, the research process has changed. Many protocols now have business development or investor-relations teams that can help educate investors and answer questions about proposed upgrades, governance changes, or announced partnerships.

Technology has not replaced the underlying diligence. What it has done is compress the time between asking a question and being able to analyze it. That allows us to spend less time gathering and cleaning data and more time determining whether the investment thesis is actually sound.

What problem does active management solve in digital assets, and how does TKNZ’s multi-token, risk-managed approach differ from passive spot Bitcoin or index products?

Digital assets are not a homogeneous market. Bitcoin remains an important part of the opportunity set, but the ecosystem has expanded to include networks and applications with very different technologies, use cases, economic models, and return drivers. TKNZ gives investors access to that broader opportunity through a single, institutionally managed vehicle. Our team uses fundamental research, on-chain data, quantitative models, and direct engagement with project teams to evaluate the durability of each investment thesis, rather than simply allocating based on market capitalization or recent price performance.

The active structure matters just as much on the risk side. We can adjust position sizes as liquidity, volatility, correlations, or fundamentals change. We can manage concentrations relative to Bitcoin and hold USDC when we believe a more defensive posture makes sense. TKNZ also provides professional oversight of custody, trading, execution, liquidity, and counterparty risk - what we sometimes refer to as operational alpha.

The result is not simply broader crypto exposure. It is a research-driven portfolio that combines asset selection with active risk management and applies the same type of institutional discipline that T. Rowe Price brings to other asset classes.

You spent the early part of your career as a data intelligence analyst at Piper Sandler and then as a data scientist at Cowen before moving into investing. What lessons from your data science background have influenced your investment style and approach?

At Piper Sandler, I worked with a ton of alternative data across different market sectors. To write differentiated research, the teams needed interesting and unique data that the rest of the Street either did not have or had not considered. I spent hours looking for an edge in the data (before LLMs could generate web scrapers in minutes) and I would sometimes gather daily data for years before being able to test whether there was a meaningful relationship with earnings or other KPIs.

That background taught me to be skeptical in a useful way. Data can tell you what happened and, in some cases, provide insight into what may happen next. But you have to understand how the data was created, what biases may be embedded in it, and whether the relationship makes economic sense.

That is particularly relevant in digital assets because so much of the data is public and observable on-chain. The challenge is not necessarily getting access to data; it is separating signal from noise. That discipline shapes how I build models, challenge popular narratives, and determine when a quantitative result is strong enough to influence a portfolio position.

What are some of your favorite tools for creating digital asset investment strategies for T. Rowe Price’s product suite? What do you use every day, and why?

Artemis Analyst is easily one of my favorite tools. The ability to consult an LLM that is connected to a large digital asset database and have it provide near-real-time analytics is extremely time-efficient. Something as involved as building a model from historical on-chain data can now be generated, evaluated, and refined much more quickly than it could have been in the past.

From there, the sky is the limit. I have used it for time-series analysis, factor analysis, relative analysis, momentum indicators, and cross-sectional comparisons - really anything that can help us gain additional insight into an investment. It also makes it easier to test a question before committing the time and resources required to build a larger model.

The tool itself is not the strategy, though. The real value comes from knowing which questions to ask, validating the output, and connecting a quantitative signal back to the technology, token economics, and real-world adoption. The tools make us faster, but the investment judgment still has to come from the team.

This is a bold thesis from Raagulan Pathy, the Founder and CEO of KAST. He doesn’t want KAST to be labeled a crypto company, rather KAST one day being in the same realm as Nubank and Revolut, or even bigger.

Due to the regional nature of banking, there hasn’t been a truly global fintech. Companies like Nubank and Revolut have billion-plus market caps, but 90% of their revenue tends to be concentrated in a handful of countries or a single continent. The exception to this rule is to build your own network. Visa and Mastercard have exceeded the $100bn+ range by building global payment networks and taking a small cut on each transaction. In order to achieve the multiple hundred billion market cap status, one needs an entirely new rail.

Enter Stablecoins

While much has been written about stablecoins, it’s hard to discount their growth. In less than 10 years stablecoins have gone from $3bn in market cap to over $300bn. Over the past few years, major companies like Stripe and Mastercard have made acquisitions to enter this space and governments around the world have developed legislation to allow this technology to flourish.

Stablecoins at a turning point

Stablecoins found their original use case in trading, but several years later have found real-world utility. In 2023 stablecoins found another use case of everyday payments. Stablecoin-linked payments have had explosive growth over the past two years. Card spend was almost non-existent at the start of November with about $3mn in spend. May 2026 was the first month that stablecoin spending surpassed the $600mn mark. Different cards found different audiences. RedotPay took market share in emerging markets, EtherFi took share in DeFi and KAST found a unique “Digital Nomad” customer.

KAST started with an interesting customer type of people who were working remotely and needed to spend money. Earlier this year, Stripe reported data showing significant growth in the solopreneur audience. Data from Carta shows that About 36% of startups founded on Carta in 2025 were led by solo founders, a jump from 31% in 2024.

Data supplied to Artemis from KAST confirms the real users spending money. KAST consumers are using their card for everyday purchases like retail, food, and travel. Spending patterns have also remained consistent, indicating customers routinely rely on their KAST card for purchases.

What’s the next leg of growth for KAST?

Cards are just the first step for KAST and the company expects 50% of its revenue to be from products other than cards in the next year. The diverse set of product offerings that KAST has planned validates the thesis of stablecoins being a global payment ramp. KAST has a team working on business accounts for solopreneurs, but also stablecoin-linked applications that resemble WhatsApp for African countries. Below are examples of the KAST suite of products:

KAST for Business: This is a full-blown B2B product designed to be a global competitor to corporate spend platforms like Brex and Ramp. B2B payments are disproportionately cross-border, and stablecoins give KAST a single payment network that reaches markets traditional rails don’t. It is currently in beta with 3,000 companies on the waitlist and through a partnership with Deel, it will also include features like payroll and employer of record services.

Kastro: This is a “lighter” social money app tailored specifically for emerging markets across Africa, Latin America, and parts of Asia. It’s designed to feel like a WhatsApp for money, instead of a traditional banking app.

Expanded Money Movement Infrastructure: KAST is heavily investing in the backend tech to allow instant money movement in and out of 180+countries.

KAST Reserve: Allows customers to earn rewards daily representing between 6 and 12% growth on USD balances and spend around the world.

In Conclusion

In some ways, KAST feels like the Amazon Web Services of money. AWS started working with many different startups and as the startups scaled, they needed more compute, Amazon was there to help them service their needs. As many digital nomads and global entrepreneurs start businesses and eventually take on more employees, they could see an interesting rise in using KAST’s card and products.

As one of the first companies to universally embrace stablecoins, KAST is expanding quickly into all sorts of money movement. Very few companies can target business-to-business audiences that were competitors to Brex and Ramp, underbanked customers in Africa and global pay-in and payouts across Europe. The 24/7 nature and instant settlement of stablecoins could allow KAST to achieve Pathy’s ultimate thesis.

Join us on X/Twitter every Monday, Wednesday and Friday at 1:00pm EDT for Alpha with Artemis where we will drop all sorts of alpha from our AI analyst. Can’t make the recording? No worries, we’ll send out summaries each week and the charts that matter to make sure you stay up to date. With that, let’s see some highlights from episode #1 and episode #2.

Spark/Sky Near 1Yr Low Market Cap/Revenue

Andrew showed crypto protocols trading within 20% of their one-year market cap-to-revenue lows and highlighted SKY and Spark as an opportunity (Andrew owns SPK token). Both tokens were trading near 1yr lows of valuation, despite both of them increasing revenue Q/Q. Over the past 7 days, Spark rallied over 40% due to a resurgence of interest in crypto and is now up 1.7% YTD while other crypto lending protocols are still down between 20 - 60%.

SaaS is Bifurcating

There is an opportunity emerging of SaaS companies who have deep moats, network effects and can monetize AI. SaaS companies with “shallow” moats have seen revenue drop 50%+ Y/Y. However, others like Salesforce, Monday.com and Toast have seen continued revenue growth, despite having mixed stock performance. For example, Toast benefits from having a strong network effect of restaurants where its machine learning models can help all restaurants with pricing, reservations and takeout. Plus, people will (hopefully) keep eating out regardless of AGI. Toast revenue growth accelerated in the previous quarter from 22% → 23%.

Another example, Monday.com, is trading at ~$1.4B ARR/ ~$3.8B market cap (~2–3x ARR, ~20x earnings), and is repositioning as an AI-agent platform serving large enterprises. While net dollar retention is trending down, Jon’s take is the market is stuck on the “Kanban board / Jira competitor” framing from early 2026 and not what it could be.

Ultimate SaaS Thesis: (credited to Chris Zeoli, ex-Insight/SignalFire)

If inference costs fall sharply, labs lose pricing power and value accrues to the application layer whose COGS collapse — Cursor, Harvey, Sierra, Figma, Monday. This year’s “cooked” enterprise software names may be the net winners.

Episode #2

Hyperliquid + Lighter at Record Multiples

Both are at all-time-high prices and all-time-high multiples (40–42x annualized fees for Hyperliquid). Hyperliquid's monthly revenue actually halved from $84M to $40M into July before a recent daily spike ($5.3M/day). While the tokens are rallying because of a possible green light in the United States, regulated exchanges (CME, NYSE) trade at a far less multiple. Both protocols will need a presidential endorsement to hopefully grow TAM and a surge in equity trading, which could be fueled by a rumored Anthropic IPO in Q4.

Analyst Uncertainty Creates Alpha

One avenue we have been looking into for generating alpha is by tracking analyst uncertainty. The idea is if analysts are uncertain or have widely varying views of a stock, that could be an opportunity for large outperformance. The company with the largest difference in analyst estimates is Revolution Medicines, which is actually up 165% this year. Biotech (UTHR, MRNA, AXSM) and AI (LI, MBLY, SPCX) also have widely different views of how revenue will unfold for these companies.

Catch us next week for more insights!

Thesis: The market thinks Circle’s moats are weak, stablecoins are a commodity, and a consortium like Open Standard will take a majority share. We believe Circle’s moats are deeper and more defensible than they get credit for, and the first-mover advantage is being under-appreciated by the market.

Core drivers:

1. Stablecoins will CAGR at 40% and reach >$1T by 2030.

2. The market is winner-takes-most on liquidity and network effects so a consortium like OUSD will struggle to take share.

3. Circle is being priced as a stablecoin issuer instead of a full-stack money platform.

Why now?

Circle’s second-worst trading day ever came on the announcement of Open Standard, a consortium stablecoin backed by 140+ companies including the likes of Stripe, Visa, Mastercard, and Google, to name a few. The stock dropped 17% on the news. The market’s reaction was telling: Stripe is going to get the band together and take down the Circle/Tether duopoly, distributing stablecoin revenue proportionally among consortium members. The news drove CRCL price near its lowest point ever.

Stablecoin growth

We believe many investors don’t believe stablecoins will reach $1T by 2030. They’ll point to the stagnating growth of stablecoins. But for the first time ever, stablecoin supply has decoupled from crypto prices. While crypto is down 50-70% from recent highs, stablecoin supply has remained flat, indicating that it is now a category of its own. If stablecoin supply continues to grow at its trailing 3 year growth rate, global supply will reach above $1T by 2030E.

The stablecoin market is winner-takes-most and liquidity/network effects are paramount

Over the past several years, tens of issuers have tried to dismantle the Circle/Tether duopoly. In spite of the hundreds of stablecoins now being issued, the two giants maintain over 80% share of supply. The first mover advantages gained by these players are really hard to overcome. Liquidity across chains, apps and exchanges is very hard to bootstrap, and Circle is already light years ahead of challengers.

On OUSD specifically: the market clearly thinks OUSD is a major threat to Circle’s business. Yet, history shows that consortia rarely succeed. A successful consortium needs:

• Aligned incentives among members — OUSD has this to an extent via interest revenue distribution

• Clear governance — Open Standard seems to be weak here, with several announcement “partners” revealing they were not consulted and remain non-committal

• Existential pressure — I would argue most institutions don’t see stablecoins as existential quite yet, though Stripe maybe does

So Open Standard has ~1/3 requirements based on today’s information.

Circle is being priced as a stablecoin issuer instead of a full-stack money platform

The market views Circle as just the issuer of USDC. It discounts Circle’s revenue since it is almost entirely interest revenue at the whims of the Fed.

In reality, Circle is building a full stack money product for the future of the internet. It is a technology company at its core.

Circle valuation vs Payments peers. There is clearly a gap between the credit card networks and everyone else. If Circle builds out the next generation, full stack payments system, its market cap and valuation will more closely mirror the card networks, taking bps on turnover rather than float.

Envisioning a $50B Circle

Today, Circle run rates ~$2.8B in revenue and trades at an $18B val for a P/S of 6.7x, well below the payment networks (14x) and high growth fintech like HOOD (17x). It trades at nearly the exact same multiple as COIN, a stock viewed primarily as a crypto exchange.

The market views Circle as a business that is pro-cyclical with crypto and whose revenue is rate-sensitive. Circle will outgrow both the label and the precarious revenues, allowing it to command a greater multiple — 10x is conservative and fair.

If predictions are correct, and liquidity/network effects are strong moats as we assume, in a 2030 with stablecoin supply at $1T, USDC share at 20% and rates at 2%, CRCL is poised to print $4B in interest revenue.

On revenue diversification, Circle is beginning to see traction in key growth products like Circle Payments Network. In spite of crypto prices remaining suppressed and stablecoin supply flat, volume on Circle Payments Network has grown at an explosive rate, $23B annualized as of the latest disclosure in late July 2026 — 6.8x growth YoY (off an admittedly small base), and 70% MoM. If growth remains high at a CAGR of 60-65%, volume would reach about $200B by 2030. At a 20bps take rate, that would generate an incremental $400M in revenue.

Looking to Arc chain, in the case it reaches the same scale as Tron (another stablecoin focused chain), Arc will generate $0.5B in fees.

The estimates above put CRCL near $5B in revenue, with 20% coming from growing payments/settlement related business lines, a combination which investors could understandably value at a higher multiple. With revenue growth and multiple expansion above we get $5B * 10 = $50B. The case for a $50B CRCL is not far fetched.

Every month, we highlight a leading analyst who takes a long-term view with investing.

This month we highlight Dr. Crossroads.

Dr. Crossroads (Roy) is an investor and creator known for deep-dive research across AI, fintech, and emerging technology on YouTube and X.

Roy takes a long-term, concentrated approach to investing, focusing on generational companies, strong leadership teams, and moments when improving fundamentals are still under appreciated by the market. That framework has shaped his investments in names like Palantir, Robinhood, Aehr Test Systems, and Figure.

Read on to learn more about his investing philosophy, how he builds conviction, and why he remains bullish on AI infrastructure and the next generation of financial platforms.

What is your story? How did you get into investing?

When I was 13, my Dad took time to show me his investments and the power of compounded gains over time. These were mostly in broad index funds, and he highlighted the importance of building for the future and that the best advantage I had was time. He talked to me about retirement and how to plan for that (such as the rule of 4%).

I later found out that my great grandfather was an avid investor until the Great Depression, when his gold mine claims amounted to nothing, his cattle stock was struggling, and his stocks had crashed. My grandfather (his son) ended up not investing as a result of seeing his Dad lose almost everything from investing, and as a result he struggled financially with a small pension in retirement.

How would you describe your investment philosophy?

I am primarily a long-term investor, preferring to pay for generational companies at fair prices instead of mediocre ones at cheap prices. If possible, I also like to identify a business whose narrative (which drives short-term price action more than fundamentals) is about to meaningfully shift. While I also do some swing trading, I find the long-term mindset works best when I’ve picked companies that are going to perform well no matter what the macro throws their way. This doesn’t mean the stock always goes up, but that the company continues to execute and grow in challenging conditions.

A significant part of my investment thesis is leadership. While some investors are happiest buying a company that a paperweight could effectively run, we’ve seen how “safe” companies that appear to have a strong moat are being disrupted in unprecedented ways due to shifting technology. As just one recent example, we have had the so-called SaaSpocalypse this year, as companies that were once deemed safe with deep moats are now viewed in a more skeptical lens due to the advent of AI. It’s here that leadership is essential.

A paperweight wouldn’t help pivot a company successfully. An elite leadership team would.

There’s a range of criteria I use to evaluate leadership, from whether or not they were able to keep past promises, to their tendency to chase shiny objects instead of focusing on the business core. I also look at whether leadership is proactive vs reactive in their business decisions, their past successes, if they’re overly focused on the stock vs what they are building, and how they treat shareholder capital.

What has been one of your best investments? How did you build conviction in deciding to invest?

While I invested in index funds earlier, I began investing in individual equities at the start of 2021. My first monetary success was via the AMC squeeze event, where I did roughly a 4x in a few months and took profit accordingly. In terms of best actual investments, those include Palantir, Robinhood, and Aehr Test Systems.

I began buying PLTR in January 2021. I had observed at my workplace the disfunctional, disorganized state of enterprise software, and Palantir seemed to have what would be very disruptive in the space with Foundry on top of their government segment. I ended up being early. Palantir, like most high-flying stocks, retraced immensely in 2022. Prior to and during 2022, I realized Palantir had what it took to be a generational company, and as such I rotated 90% of my individual investments into this one stock, averaging down to $12 per share and buying a number of LEAP options.

While I still hold Palantir as a top 5 position, I took many gains along the way. Some of these gains fueled my next 10x investment: Robinhood, which was trading at a 1x EV at the time, with a large cash position and no debt, much like Palantir. I looked at the narrative and realized a “golden confluence” moment was about to occur in this cyclical business, with high interest rates, the start of a bull market, and the crypto bull market. At the time, I thought of Robinhood as an excellent swing trade. Over time, as I saw Vlad and his team execute relentlessly, I realized I had stumbled on another generational opportunity and it became a long-term position for me.

I’ve had quite a few other successes along the way, with AEHR being a major success for me this year. Again, the setup was clear. Aehr was shifting from 90% SiC revenue for their wafer and package level burn in testing equipment offerings to something more durable with the advent of AI and the significantly higher cost of failure of GPUs (and other components), along with the need for increased throughput via WLBI testing. The company had no debt, was in the process of a major accretive acquisition, and the CEO was incredibly knowledgeable in the space, with only one major misstep (related to EVs and renewables). The narrative shifted first, followed by comments on bookings, then bookings and finally the beginning of the revenue acceleration. I believe this will be my next 10x, with an average in the low $20’s.

With each of these, the investment thesis was initially basic, but by following announcements, quarterly earnings, and trends in their space my conviction increased even as their execution improved. It made each easy to hold through volatility as they relentlessly execute.

You recently wrote about Hyperscaler Earnings and the signal on AI. What is your thesis moving forward on the AI investment bottleneck stack?

We’re still dramatically underestimating the demand for data inference related to the AI trade, and the debt for these CapEx buildouts is (mostly) not scary when balanced against take-or-pay contract structures. Simply put, there’s a lot of negativity priced into data centers at the moment, yet there is insatiable demand which looks to be durable at least through 2030, if not beyond.

That said, there are dozens of ways to play this AI trade, and I think most are still in the early innings, from storage, to memory, to data centers, to automated testing equipment, power, cooling, enterprise software, cybersecurity, and more. If someone wants exposure, it’s important to find just a couple of themes investors can understand and are interested in, and find the winners in that space.

There are some parts of the AI trade that have such immense returns that we’ll see many new entries into the market to sate demand. We saw that with hyperscalers developing their ASICs, with many companies pivoting to data center businesses, and most recently within storage and memory. The challenge is that many of these projects take years and significant investment to get off the ground. As such, I prefer to concentrate on the leaders in the space unless it’s clear that disruption is warranted or if the leader isn’t executing well. Even as new entries come online, the TAM is larger than most realize, and companies who are led well might at worst see a smaller market share of a much larger pie.

You work full time, hold a PhD and are a finance creator and influencer! How did you get to where you are today?

I figure that the time will pass regardless, so it might as well pass while building things I love and working towards a direction I’d like to be in the future. Aside from one area in my life, I’ve had more success than I imagined, but I don’t let success go to my head or stop me from growing.

Some of these opportunities came even when I wasn’t expecting them, but what made many of them possible is being consistent: creating content, providing value, doing research, examining the other side of a thesis, and finding enough flexibility to pursue new opportunities.

My faith has been a benefit in that I don’t view success by a number on the screen (whether followers, income, or net worth), but by faithfulness. As such, I’m not willing to compromise character or shortchange an important relationship for financial gain or personal benefit. Over the short-term, that has cost me some opportunities. Over the long-term, I think it’s benefitted me in both qualitative and quantitative ways.

What advice do you have for investors who are beginning to invest in 2026?

2026 has been another rollercoaster market. That’s likely to continue through the rest of Trump’s term. During the downturns, don’t give up on the market. Instead, use that as a barometer for how much volatility you’re really okay with. Concentrate in just a few high quality names and learn everything you can about them. If the thesis weakens, don’t be afraid to pivot, but be careful about chasing green candles and shiny new objects.

Lastly, “boring” investing can be very successful. If the volatility is too much, especially if you are young, there’s nothing wrong with just investing in an index fund and focusing your time and energy into earning more. That’s largely what I did until 2021.

What’s your thesis for ?

The simple thesis for Figure Technologies: there are few companies who are equally adept at TradFi and cryptocurrency. They’ve built something special with HELOCS, but this was just the first step in their vision. They’re able to leverage incredible efficiencies that almost nobody can compete with, and do so as a platform which invites participation from competition.

Although the “year of first lien” is largely being ignored by the market due to the moribund state of real estate, and this likely worsens for 3-6 months following the election, this is the perfect time to build and prepare for future volume that will come. Current conditions make it tough for competition, and Figure is incredibly well-positioned for an eventual thaw in this market, but even without the thaw, they’re growing 100% Y/Y and many of their offerings benefit from higher rates.

What’s your most contrarian view on the markets right now?

Back in 2023, I said we’re likely to be in a durable bull market through 2029 or 2030, due to how much of an impact AI (with robotics behind it) will have on our economy. That’s no longer contrarian.

Over the last year, I was one of a vocal minority saying that we were not in a bubble (except with private equity). That no longer appears to be contrarian.

Today, the most contrarian take I have is one that should be obvious: CapEx buildouts, even when fueled by debt, should be seen as bullish when done by credible companies and leadership teams. These are being done by some of the most successful companies and leadership teams, and even by some which are incredibly conservative (such as TSMC).

How have you been able to use Artemis in your investment research process?

I’ve really enjoyed using the Analyst feature, especially as this has evolved, to create charts and pull quotes from recent interviews and earnings calls. It’s also been great at quantifying some trends that I’ve sensed, but wasn’t able to put concrete numbers on. Additionally, I like being able to track elements of my investing portfolio on Artemis, including the Robinhood Chain, Rothera / Predictions markets, Figure Technologies blockchain metrics, and stablecoin activities for several investments.

How to get in contact with Roy?

You can find Dr. Crossroads on X and YouTube.

Figure spent six years building a home equity lending platform designed to shorten origination times and reduce costs. Today, a growing portion of its business consists of marketplace activity involving loans that Figure neither sources nor funds, a model that we believe may generate attractive incremental margins. In Q1 2026, partners originate 78% of the volume crossing the platform, Connect volume has run from $8M in Q4 2024 to $1.6B, and fee lines carrying relatively high incremental margins have grown from 5% of revenue to 28%. We expect Figure to become a capital-light toll road on loan flow. The market is still underwriting the balance sheet.

The question that matters, then, is how much loan volume can move through the platform without Figure sourcing the borrower or funding the loan itself.

Summary

This memo is a joint perspective from Artemis and North Island Ventures (NIV). Artemis is a digital finance research firm focused on blockchain and equity assets. NIV is an NYC-based investment firm, focused at the intersection of blockchain, fintech, and AI.

Figure Connect, launched in June 2024, is a marketplace that sits between loan originators and institutional buyers. Partners originate the loans, institutional capital funds them, and Figure earns a fee for underwriting and distribution. Figure reports all of this activity as Consumer Loan Marketplace (CLM) volume: every loan originated on its platform, plus third-party loans traded on Connect.

The model is already scaling. Partner-originated loans are 78% of volume, Connect reached 56% of CLM volume within seven quarters of launch, loan inventory has fallen from 31 days to 16, and ecosystem and technology fees have grown from 5% of revenue to 28%.

Additional loan supply may become available from two sources. The pending Kiavi acquisition adds a $7B annual origination engine portions of which Figure should be able route through the same infrastructure. Separately, available industry data indicates that second-lien lending has begun to recover after more than a decade of contraction, expanding Figure’s core market.

Meanwhile, the Street appears to be pricing in take-rate compression as Figure expands into first liens, a structurally lower-take-rate product than its core second-lien business. Yet Figure’s disclosed net take rate has risen even as first-lien mix has nearly doubled. We think that disconnect is one piece of a broader misunderstanding of the business, and it helps explain why consensus estimates remain too low.

Background

Figure was founded in 2018 by Mike Cagney and June Ou, respectively the co-founder and the CTO of SoFi. Figure and its roughly 390 partners originated $8.4B of loans in 2025, up 63% year over year, and the platform is currently running at ~$17B annualized, up 130%. That is about 5% of all U.S. loans backed by residential real estate.

We view Figure’s original advantage as speed and cost. The company built two things in tandem: a fully automated home equity origination platform, and Provenance, a purpose-built blockchain on which every Figure loan is originated, registered, and ultimately securitized. Together, they compressed home equity lending from roughly six weeks and $11,000 of cost to under seven days and $1,000. That advantage, in our opinion, built Figure’s position in second-lien HELOCs, where it is now the largest nonbank originator.

A second lien sits behind the borrower’s existing mortgage: the homeowner leaves a low-rate first mortgage untouched and borrows against the equity above it. A first lien holds the primary claim on the home, whether that is the original mortgage or a new loan to an owner who carries none.

Connect may represent an important shift in the economics of the business.

Thesis

We see two potential catalysts playing out through 2026 and beyond: the continued scaling of Figure Connect, and the pending Kiavi acquisition, which should double Figure’s first-lien business when it closes in Q4. The third section below lays out why neither is in the price.

1. Connect is converting Figure from a lender into a marketplace

Through 2024, Figure was, in our opinion, a strong loan originator inside a structurally mediocre business model. It sourced borrowers through a sales force, funded loans with warehouse facilities, held them on balance sheet for roughly a month, and monetized them on sale.

We view those economics as typical of a specialty lender: capital-intensive, dependent on external funding, exposed to credit and volume cycles, and therefore valued at the 5–7x EBITDA multiples the sector typically commands.

Connect changes how portions of Figure’s growth are funded. Partners source the borrower and originate against presold commitments from institutional buyers, while Figure provides the underwriting and distribution infrastructure and collects a marketplace fee of roughly 3%. Connect volume can therefore scale without a corresponding increase in customer acquisition spend, warehouse funding, or credit exposure.

Connect volume went from $8M in Q4’24 to $1.6B in Q1’26. Partner-originated volume is now roughly 78% of total flow, which means most of the loans reaching Figure arrive without Figure bearing the customer acquisition cost (CAC).

Figure’s reported balance-sheet metrics also reflect the shift. Loan inventory has fallen from roughly 31 days to 16 even as volume has nearly tripled. Warehouse borrowings are just $15 million against $1.9 billion of committed capacity, less than 1% drawn. And against $6.8 billion of securitized collateral, Figure’s maximum disclosed exposure is only $378 million.

The revenue mix is shifting toward the highest-quality line on the income statement: ecosystem and technology fees. These are platform and marketplace fees that carry near-100% incremental margins, and they have grown from 5% of revenue in FY’23 to 28% in Q1’26.

From FY’23 to FY’25, revenue grew 142% against 18% cost growth. Incremental EBITDA margins ran 82% in 2024 and 91% in 2025, lifting reported adjusted EBITDA margin from negative 4% to 49%. Continued volume growth on the current cost base should support further expansion toward management’s medium-term target of 60% by 2028.

2. More loan supply may be coming

Connect gets more valuable as more volume runs through it, and Figure has two large sources of new supply potentially arriving at once.

Start with second liens. Industry data indicate that balances have recently increased following approximately 13 years of contraction after the global financial crisis. Bank HELOC balances reached $287B in 2026 and are growing again, having fallen from a 2009 peak of $600B. The industry opened 1.2M HELOC lines in 2025, the most since 2022 but still only about half the mid-2000s peak.

Homeowners now hold 71.6% of their real estate value as equity, the highest level in roughly 35 years. Much of that equity sits behind mortgages locked in at 3–4%, which makes a full refinancing uneconomic at today’s ~7% rates. For a homeowner who wants that equity without giving up a low-rate first mortgage, a second lien is one rational way to get it.

Figure enters this recovery as the largest nonbank HELOC originator, and into what we believe is a relatively thin competitive field: capital requirements, compliance costs, and the absence of GSE support have pushed many lenders out of the category.

Kiavi, meanwhile, gives Figure a second front — first liens, where the acquisition roughly doubles the existing business. First liens serve borrowers that second liens cannot: homeowners who own outright, roughly 40% of U.S. homes, and those for whom refinancing the existing balance still makes sense. Balances typically run $200K to $300K, and annual origination flow is roughly $2T, about ten times the size of the second-lien opportunity.

Kiavi is the largest originator of residential transition loans (RTL), short-term bridge loans of roughly 12 months that fund fix-and-flip purchases and renovations. RTL is 85% of its volume. Its RTL market share grew from 2.1% in 2020 to 9.7% in 2025, through three years of frozen housing turnover, while subscale competitors lost their warehouse lines and shut down. In 2025 it did over $7B of volume and more than $250M of revenue, growing 30%, at roughly 40% EBITDA margins.

The structure of the deal matters too in our view. Figure and Sixth Street are jointly paying $717M in cash, $538M from Figure and $179M from Sixth Street. Figure keeps the operating platform, while Sixth Street places Kiavi’s loan assets into a JV backed by more than $3B of forward purchase commitments, pre-seeding Connect demand.

3. The market is likely misreading the transition

We see three primary reasons the mispricing exists.

1. The Street may be underestimating the durability of Figure’s take rate. First liens structurally carry lower take rates than Figure’s core second-lien business, so as they become a larger share of volume, mix should mechanically pressure the company-wide take rate.

So far, it hasn’t. Figure’s disclosed net take rate rose from 3.4% in Q4’24 to 3.8% in Q1’26 even as first-lien mix nearly doubled, while second-lien pricing held at roughly 4.5–5.0%. Underlying pricing appears healthy enough to absorb the mix shift.

The distinction matters for valuation. Price erosion reduces the economics of existing volume; mix shift into first liens adds incremental volume at a lower take rate. Connect, separately, lets more of that volume scale with less capital, little to no CAC, and higher margins. Consensus appears to be underwriting take-rate compression despite little evidence of deterioration in product-level pricing.

2. Thin coverage. Only four analysts cover FIGR today. Since Q1’26 earnings, Figure has published weekly CLM volumes on its own website, giving investors a near-real-time proxy for revenue. Yet only one sell-side analyst explicitly models CLM volume, and even that analyst’s estimates sit below the volumes Figure is already self-reporting. Figure is handing the market a live read on its revenue line, and the market is largely ignoring it.

3. The stock trades in the wrong bucket. Our assessment: FIGR initially traded as a crypto proxy and now behaves more like a momentum name, even as the business comes to resemble a financial marketplace such as Tradeweb or ICE.

The valuation has yet to reflect that transition we think. FIGR trades at roughly 12x EBITDA, its lowest multiple since going public: above the 5–7x typically assigned to specialty lenders, but well below what scaled financial marketplaces command.

Figure does retain crypto-adjacent businesses, including the YLDS stablecoin and its tokenized securities marketplace, but together they contribute less than 2% of revenue. Management appears focused on correcting the perception. Figure’s Investor Education Presentation steers investors away from the crypto narrative and toward the two variables that increasingly determine the earnings power of the business: CLM volume and take rate.

Counter Thesis

The thesis breaks if Figure’s competitive advantage proves temporary, housing volumes deteriorate materially, or the economics of its core products weaken.

• AI commoditizes underwriting. A former Figure product manager argued that AI could make five-day closings table stakes, eroding Figure’s speed advantage and forcing competition on price. We believe the more durable moat is the marketplace, but industry-wide improvements in underwriting speed and cost remain a key risk.

• Housing volumes weaken. Connect reduces credit exposure but not volume risk. HELOC delinquencies are rising from record lows, and a housing downturn would pressure HELOC demand and hit fix-and-flip lending particularly hard.

• Core pricing deteriorates. The thesis depends on product-level pricing staying healthy as lower-take-rate products become a larger share of volume. A sustained decline in second-lien take rates, or a net take rate below ~3.5%, would suggest that competitive pricing pressure is starting to overwhelm the benefit of incremental volume.

• Governance risk crystallizes. Cagney and Ou control ~71% of the vote through a dual-class structure, leaving shareholders with limited recourse. The counterweight is alignment: Cagney’s substantial FIGR holdings and nearly decade spent building Figure give him significant financial and professional exposure to the company’s long-term outcome.

Conclusion

Our assessment is that the Street remains focused on two things: the risk that lower-take-rate products dilute Figure’s economics, and the loans Figure still originates itself. We think that misses the transition. Connect is letting substantially more volume move through the platform with less capital, less CAC, and higher incremental margins. From our perspective, the more lending volume that moves across those rails, the less Figure resembles the lender the market still underwrites.

Important Disclosures

This article and all of the information presented herein is being provided by Artemis Analytics Inc. and North Island Ventures, LLC (“NIV”) (each an “Author” and together, the “Authors”) solely for informational purposes to provide educational content and general market commentary.

No statements included herein relate to or contain any offer of NIV’s investment advisory services, nor does any content herein, including but not limited to any referenced companies, constitute an offer to sell or the solicitation of any offer to buy any security or shares or interest in any of NIV’s current or future private investment funds (each a “Fund”), which may only be made at the time a qualified offeree receives the applicable confidential private placement memorandum, subscription documents and governing documents (collectively, the “Operative Documents”).

The information herein is not intended to provide, and should not be relied upon for, accounting, legal or tax advice or investment recommendations. NIV is not acting, and does not purport to be or act as, an adviser or fiduciary, in any capacity, with respect to any prospective investor in any Fund. A recipient of this article should consult his or her tax, legal, accounting or other advisors about the matters discussed herein.

While the information presented in this article is believed to be accurate, neither Author nor their respective affiliates or representatives (i) makes any express or implied warranty as to the completeness or accuracy of the information presented herein, (ii) shall be liable for errors appearing in this article or (iii) has any duty to update the information set forth herein. Certain information contained herein constitutes “forward-looking statements,” which can be identified by the use of terms such as “may”, “will”, “should”, “expect”, “anticipate”, “project”, “estimate”, “intend”, “continue“, “target” or “believe” (or the negatives thereof) or other variations thereon or comparable terminology. Due to various risks and uncertainties, actual events or results or actual performance of investments may differ materially from those reflected or contemplated in such forward-looking statements. No representation or warranty is made as to such forward-looking statements.

Any analyses and opinions contained herein (including market commentary, statements or forecasts) may contain elements of subjectivity (including certain assumptions) or be based on incomplete or incorrect information. The Authors do not undertake to update the information contained herein.

This document displays certain trademarks and logos owned by the companies discussed herein. All such trademarks remain property of their respective owners, and are used only to directly identify or describe the companies being discussed herein. Their use is not intended to indicate and in no way indicates any sponsorship, endorsement or affiliation between the Authors or their affiliates, on the one hand, and the owners of such trademarks, on the other hand.

Accounts advised by NIV are currently invested in Figure and may make additional investments or exit positions at NIV’s discretion at any time for a variety of reasons and without further notice. No claim of performance (predecessor or otherwise) or prior track record is made in this document. Company metrics, including but not limited to, revenues, earnings, EBITDA, net income, and similar metrics, are not a proxy for the performance of a potential investment and should not be understood as such. There is no guarantee that achievement of any financial metrics, including changes in share price, will result in positive investment performance. Unforeseen or unaccounted for events could materially impact Figure’s performance and/or valuation—in such an event, projections could be materially altered. An investment’s performance is based on and must take into account a number of factors that the materials do not reflect, such as trading activity, terms of investment (e.g., share class), use of leverage, trading costs and other factors. Readers of this article should not interpret any presentation of share price as a proxy for any fund managed by NIV or investment performance. NIV makes no guarantee or assurance that any Fund will make similar investments. All investments are subject to the risk of total loss.

The discussions of Figure reference publicly-available information and certain estimates or projections have been provided by Figure or are primarily sourced from Figure. The Authors make no representation as to (i) the accuracy or reliability of any estimates or projections, (ii) the reasonableness of applicable assumptions and comprehensiveness of available inputs or (iii) the effectiveness or reliability of calculation methodologies. The Authors make no guarantee with respect to the prices at which any securities can be bought or sold. Numerous factors affect the performance of any investment.

• Revenue was $370.1 million (up 48% yoy)

• Net Dollar Retention Rate was 136%

• GAAP gross profit was $309.6 million; GAAP gross margin was 84%

• GAAP net loss was $(112.2) million, driven by SBC: $147.6 million

• 15,964 Paid Customers with more than $10k in ARR (up 34% yoy)

• 1,635 Paid Customers with more than $100k in ARR (up 46% yoy)

The numbers were strong yet, Figma fell 17% after earnings. I got curious about what happened and went down a deep rabbit hole.

I started thinking and realised there is a much bigger possibility for Figma:

I believe Figma will become the design layer for ALL creators and agents building software.

Before AI, the workflow was pretty straightforward:

Designer → Figma → Engineer → Code → Software

Now, the workflow starts to look more like:

Designer / Engineer / PM / Product Engineer / Agent
↓
Figma / Design System
↓
Software

Here are three things which support my thesis.

1. The rise of Product Engineers

The term “product engineers” was popularized through people such as Sherif Mansour, Jean-Michel Lemieux and Gergely Orosz, who helped shape how we think about the role today.

So, what is a product engineer?

A product engineer is someone who writes code, understands customer pain points, and helps shape the product.

AI makes software easier to build. As the cost of writing code falls, engineers are taking on more work across product and design. These boundaries are starting to blur.

I think this matters for Figma because engineering teams have historically been much larger than design teams. One of Figma’s largest customers now holds more seats for engineers than designers, and this will become more common in the future.

I got curious about on whether people are searching for “product engineer”. Google trends tells me what I have imagined. This term has been picked up meaningfully over the past few years.

If this trend continues, I believe product engineers will be one of the key growth drivers for Figma.

2. Every Employee will have an Agent (Figma Agent)

In 2026, everyone is talking about agents, agents, agents. The next expansion for Figma will be from humans to agents.

Traditional SaaS monetization is simple:

Cost Per Seat ✕ No. of Seats

I believe in an agentic world, every employee will have one or multiple agents working on their behalf. These agents will help to iterate (edits a component, turn a mockup into code) on products. These actions will naturally consume inference and compute.

And this is how SaaS business model will be in the future:

(Cost Per Seats ✕ No. of Seats) + (Credits Consumed ✕ Cost Per Credit)

I believe that Figma will continue to monetize people (seats) using their products and also the work being done through them (the agents).

3. Margins Expansion

AI is expensive. Compute and inference are cost to every business that wants to enable AI.

As Figma launched Figma Make, gross margins fell from 90% to 80% as inference became part of the cost of revenue. In Q2 ‘26, non-GAAP gross margin recovered to 85% as Figma began monetizing AI usage through credits.

What I (and everyone on the street) thought will be a cost to Figma, becomes an additional revenue stream.

I truly believe that as Figma continues to monetize their suite of AI products (Figma Make, Figma Agent, Figma Weave and Figma MCP), their margins will continue to expand.

Risks

AI Tools such as Claude/OpenAI will be able to replace Figma

The biggest risk to the thesis is that these AI tools, got so good that it will collapse the entire design-to-code workflow. If a user is able to describe in natural language how the look and feel of a product is and ship it end-to-end, Figma is probably cooked.

Mitigant: I think that is still far from the truth today (or anytime soon). Figma already owns the customers. Customers are familiar with the current tooling and have their workflows around it. Furthermore, many years of design systems are already embedded inside Figma which is a huge moat.

Why is the market mispricing this?

Figma stock price has been down 80% since the first closed ($115.50 on 7/31/2025).

In my view, three things suggest why Figma is mispriced:

1. Revenue in Q2 ’26 continues to grow 40% yoy (means people are still using Figma even with the implosion of Claude tools)

2. NRR (Net dollar Retention Rate) is at 136% (seats expansion still remains very high and market is very pessimistic about software stocks)

3. EV/ARR is trading near all time lows at 7.3x today (it is not cheap but given Figma’s growth rate, I feel it is reasonably priced)

Conclusion

The market still values Figma as a software company for designers.

I think that misses the bigger opportunity.

As the lines between roles continue to blur (the immersion of new roles: product engineers) and agents help in building software. Figma has THE opportunity to become the de facto design layer connecting all of them.

The bull case is no longer just more designers paying for more seats.

It is:

1. More creators using Figma,

2. More agents operating through Figma,

3. and more AI usage being monetized on top of it.

When that happens, Figma’s TAM will become significantly larger than what the market is pricing in today.

We told our team and half of them left. We were terrified. But raising the scale of our ambition to democratize investing for all requires balls.

Here is that internal memo.

Investing For All.

Today, we’re pivoting from crypto data to building AI analysts for all.

As part of that, we’re now covering ALL datasets that matter for investing -- blockchain, equities, and alternative data.

When Anthony and I started Artemis in 2022, we wanted to make crypto investing accessible to everyone.

This was inspired by wanting to serve people like my Yeh Yeh (grandfather), who worked 2 jobs, had to raise a family of 8, but invested in HK stocks in the 1970s and made just enough to send my dad off to college in the U.S. This was inspired by Anthony’s desire to help people like his father who tried to break into Wall Street in the 1970s and inspired Anthony’s love for investing.

We wanted crypto tokens to be more transparent and accessible so people like our Yeh Yeh and dad could make money in crypto in the 2020s.

However in 2026, that vision is too narrow. The same liquid token funds we serve are now primarily invested in crypto equities, fintech and AI stocks.

Our new mission: make investing accessible to all and allow anyone to harness the power of LLMs to make money in ALL assets — equities, tokens, private markets.

The thesis was always for Artemis to build the operating system for investing and build the #1 investment platform in the world: the place where investors figure out what to invest in and why.

Our strategy was to 1) build proprietary data around stablecoins, prediction markets, financial and qualitative signals that moves markets 2) create the investing network that connects retail investors, institutions, and public / private companies in a single place, where people can pitch each other their investment theses.

The first step towards our vision, is to make Analyst publicly available to all investors today : https://www.artemis.ai/analyst

The problem

Investing is overwhelming in 2026. There are X posts, podcasts, Reddit posts in /wallstreetbets, earnings, financials, 13Fs, sell-side research firms price targets, etc.

Investors are turning to independent analysts like Citrini, Semi Analysis, Amit the Investor, and others to cut through the noise and build conviction.

In a world of AI, hedge funds and retail are building out skills on Claude and hacking together Hermes Agents and OpenClaw but are fed noise not signal. Also it’s expensive to buy all the datasets and cobble them together via connectors -- and the fear is Claude / Chat hallucinates with just web search.

THE TRUTH IS, NOBODY HAS FIGURED OUT HOW TO USE AI FOR INVESTING.

Until today.

Artemis Analyst

We spent the last few months building the investing harness of my dreams.

We had to as part of our work publishing investment theses on Robinhood, Hyperliquid, Coinbase, Circle, Figure and more.

Our AI Analyst is so good that Andrew and I use it up to dozens of times...just for fun to diligence investment opportunities:

We spend $2M year on data and compute (long $SNOW) and have a world class team of data engineers t build an autonomous data system to get the signals we need.... this the dream team if I were to start a hedge fund in 2026.

Today, we get to launch Artemis Analyst with 20+ proprietary datasets, including amazing partnerships with Sacra, Fiscal AI, Serafis and MORE that are NOT available in frontier models.

Andrew who was previously from Ribbit Capital sums it up:

“After decades of investing, I finally have my dream product. I used to spend countless hours sifting through filings, financial data, documents, and podcasts — now it’s all at my fingertips with Artemis Analyst”

What we’re launching today

Today we’re unveiling a new Artemis for investors:

• The Artemis Feed: the new daily homepage for investors. An insights feed that surfaces the signals that matter across public companies, tokens, and private companies powered by Artemis Analyst: https://www.artemis.ai/home

• Unique Datasets: Consensus Estimates, ALL Prediction Markets in the U.S, Private Company Financials, Podcast Data, Stablecoin Payment Volume and MORE now only available on Artemis Analyst.

• Artemis Analyst: we continue to double down on building the best investing harness with proprietary data: https://www.artemis.ai/analyst

One of the largest asset managers told us:

“I have your Analyst next to my actual ChatGPT and I actually default to your Artemis chat more than my ChatGPT.”

Starting in August, we’re opening the same analyst up to ALL investors.

This is our first step toward building what we call the thesis investment platform. More to come in September.

Mobile app launching very soon so you have an Analyst on the go :)

💜 Team Artemis

The week ended mixed, with MSFT up 22% of Azure revenue growth, fintech/crypto stocks COIN and HOOD ended down an average of 8.2%, and memory stocks down double digits.

This week also saw the month of July come to a close. The chart below tells a meaningful story for the month: memory valuations corrected down, while the inverse thesis — “software is dead” — defied expectations ($FIG up 33% M/M).

Lastly, on the macro front, the Fed met and decided not to change short term interest rates as , though some members of the committee did vote to raise rates 25bps. This second bit would likely indicate future rate hikes, especially if the war in Iran does not see a resolution and inflation does not fall as quickly as hoped.

Some Artemis specific news/announcements:

• We posted our Coinbase Q2 earnings estimates, and were included in Coinbase’s own distribution!

• We released Episodes 3 and 4 of our Thesis Podcast

  • Thesis 3: Hyperliquid $230 by 2030, with Artemis’ own ZJ

  • Thesis 4: DeFi Decade with Ishmael Asad (Bitwise) and why $AAVE is a Winner

• We attended Robinhood‘s Q2 Earnings call live, and Jon got to ask Vlad a question

• We shipped a lot on product like a refreshed look, dark mode, privacy controls, and much more

  • See the full Weekly Ship Log here

Today We Highlight:

• Major DeFi protocols continue to mature

• Major Earnings Week Recap

DeFi continues to mature

Last week we released the 4th episode of our Thesis podcast: DeFi Decade with Ishamel Asad. And it couldn’t have come at a better time.

Decentralized finance (DeFi) has long been the wild west, insular and rife with dodgy characters, malicious code, and scams. It is often said that onchain finance is relearning all of the lessons that traditional finance has already learned — at many times the speed. The maturation and growth of onchain finance has been hindered by a couple of big blockers, chief among them: an inability to embrace compliance, and a lack of fixed rate credit products. Recently, there were two major developments that aim to address these issues:

1. Uniswap, the world’s largest decentralized spot exchange, introduced Permissioned Pools, allowing Uniswap pool deployers to enforce compliance directly onchain. Uniswap’s Permissioned Pools were built in collaboration with firms looking to bring regulated assets like stocks, funds, and more onchain. Launch partners include Superstate, Securitize and Dowgo.

2. Morpho, the world’s second largest decentralized lending protocol, launched Morpho Midnight, their implementation of onchain, non-custodial fixed-rate lending markets.

So what?

For public blockchains and their applications to be beneficiaries of real world asset tokenization, liquidity on chain must continue to improve and regulated assets must be able to remain compliant. Uniswap and Morpho are giants in the public, permissionless onchain financial system. Uniswap, the world’s largest decentralized spot exchange, who counts BlackRock among its recent investors, released permissioned pools, directly addressing the fear of lack of permissions and compliance for assets. At the same time Morpho announced Midnight, their solution to the lack of fixed-rate debt onchain. While everyone is looking at AI, semis etc, the best teams in DeFi are continuing to ship new major protocol versions and upgrades, and continue to march forward toward the vision of an internet-native financial system. We remain extremely excited and optimistic about developments in democratizing and improving finance.

Major Earnings Week Recap

This past week was the most consequential of this earnings season. About $25 trillion worth of companies announced earnings, including the hyperscalers/CapEx spend leaders like AMZN, META, and MSFT.

This quarter ended an H1 of debt issuance where hyperscalers issued more debt than in any prior full year. And we’re only halfway through. Hyperscalers have started running out of FCF to invest in CapEx and will likely continue reaching to credit markets for funding.

Per Apollo’s Rob Bittencourt, partner and head of thematic investing(Tech Disruptors, Jun 11):

“If you look back at capex amongst those five companies in 2019, I think it was below $100 billion. We’re now talking about over $750 billion of expected capex across those five companies… Those five companies collectively generate about $800 billion of EBITDA… But the capex numbers have gotten so large that now they’re in a position where they need to look toward alternative sources of capital.”

“By our estimates, about 50% of net IG issuance year to date is AI related. The hyperscalers issued close to $100 billion last year. I think they’re at $120 or $130 billion this year. That number’s only going to go up.”

AI CapEx does not look like it will slow down any time soon as hyperscalers continued to guide robust spending. Credit markets are becoming a key player and could act as a moderator.

Also noteworthy, crypto related stocks were down through earnings, COIN and HOOD ended the week down 7.6% and 8.8%, respectively, on weaker crypto revenue. This week we are watching out for CRCL earnings, and FIGR and SECZ next week.

Charts of the week

• Robinhood chain developer activity soars past Ethereum

• 62% of HIP-3 volume happens when US markets are closed

• Hyperscaler CapEx has reached 35% of revenue

Other News

• X Money rolling out to Premium users

• New York wants to shut down Kalshi

• Hugging Face posted their post mortem of the OpenAI rogue agent attack

• Odds for a Senate vote on the CLARITY Act before the August recess fall to all time lows

Building this index required data that did not exist in standardized form. Over the past year, our team has worked closely with the Pantera and S&P teams and spoke with countless protocol teams, investors and community members to better refine the metrics and datasets. We read through countless docs, Discord/Telegram channels, blockchain explorers, and more over the past year to understand each protocol individually. This results in two things we think will move the industry forward:

1. Outstanding Supply: a new supply metric that gives a more accurate picture of a token’s economic size, typically a value sitting between the token’s fully diluted value and its float.

   See our thoughts on supply metrics here

   .

2. Standardized revenue methodologies: for each token in the index we’ve standardized what revenue means so that a dollar of revenue is comparable across protocols no matter if they are a lending market, an exchange, or a neobank.

   See our thoughts on crypto revenue here

   .

At Artemis, we are genuinely excited for the next phase of growth of blockchains and digital finance. The world is undergoing some of the most radical societal transformations humanity has ever seen, and we believe that blockchains will be a critical part of the future state of the world.

We invite protocol teams and involved community members to reach out to us if you have any questions or would like to connect to learn more about the index. Our intention as the data providers for this index is to have the utmost integrity in the data that powers it. To that end, if you are a protocol team that we have not been in contact with yet, please reach out to support@artemisanalytics.xyz or @x0wes on TG.

See our docs for answers to frequently asked questions.

Risk assets went nowhere. The composition underneath changed a lot. Bitcoin finished at $64,444, down 0.7%. Ethereum was up 0.6%.

The six-week ETF bleed finally stopped:

• Spot bitcoin ETFs: +$75.5M (Jul 13-17), +$33.9M (Jul 20-24)

  • Bitcoin ETF AUM: $78.9B

• Ethereum ETFs: roughly +$104M in each of the two recent weeks

The real repricing happened in rates. Fed hike odds for July 28-29 went from 10.7% on July 15 to nearly 35% by July 22, almost entirely because Brent pushed toward $100 on the Iran conflict.

Unlike last week, semis recovered.

DELL (+10.4%) and MU (+8.5%) led the basket, a sharp reversal for memory after Micron’s 28% slide from its June highs. NVDA (+2.0%) beat the index it anchors (QQQ -1.5%). DeFi extended its run with UNI (+2.9%) and AAVE (+2.3%).

Crypto equities were mixed. CRCL (+3.2%) was the only one with a real gain. COIN (+0.75%) round-tripped its entire 9.6% CLARITY pop from July 21.

The damage sat in retail fintech: SKY (-6.2%), HOOD (-5.0%) and SOFI (-4.8%), with HYPE (-3.5%) close behind. BTC (-0.7%) and the broad tape barely moved (SPY -0.5%, DIA -0.3%).

The basket averaged +0.41% on the week, median +0.18%.

Today We Highlight:

• The S&P Pantera Digital Asset Index, powered by Artemis data, tracks revenue-generating tokens and weights them by adjusted market cap

• Robinhood Chain: What is actually happening under the hood

Introducing the S&P Pantera Digital Asset Index powered by Artemis

S&P Dow Jones Indices and Pantera Capital launched the S&P Pantera Digital Asset Index (SPPDA) on July 20.

It is an 18-token benchmark drawn from the S&P Cryptocurrency Broad Digital Asset Index that screens on protocol revenue, then applies capped market-cap weightings.

Artemis supplies the revenue classification, the circulating supply and outstanding supply data. Lukka supplies pricing.

The five largest constituents by weight are ETH, BNB, SOL, TRX and HYPE.

Bitcoin and XRP both fail the revenue test outright. For scale: XRP’s entire 14-year burn history is roughly $16M, against the more than $3B generated annually by the 18 tokens that qualify.

So, what?

S&P is careful about what this measures. The methodology calls protocol revenue "an objective, rules-based indicator of economic activity" and explicitly says it implies nothing about investor returns or cash flows.

Put plainly, revenue decides which tokens get in, and market cap decides how much of each you own. So if you saw the headlines and assumed this is a way to own crypto's biggest earners, it is not.

Moreover, a token needs at least 30% of its supply actually circulating to qualify, which quietly disqualifies projects whose headline valuations rest on coins nobody can sell yet.

This rule runs on Artemis’ Outstanding Supply data, and nobody has written about it yet. The first change to the lineup is decided on the third Friday of August and takes effect September 18.

Robinhood Chain: What is actually happening under the hood

Start with what Robinhood Chain is, because most people skips this.

Robinhood Chain is an Arbitrum Orbit L2, which means Robinhood rents Ethereum’s security and runs its own cheap, fast transaction layer on top of it. It went live July 1.

The pitch was tokenized real-world assets: put stocks, ETFs and private equity onchain. Robinhood’s customers can trade them around the clock, including in places where opening a US brokerage account is not an option. However, what actually launched was a memecoin casino and what is happening now is that the casino is slowly turning into the thing it was supposed to be (tokenization of RWAs).

Just last week, we showed Robinhood Chain as the leading chain by DEX volume, on a peak day. Over the full week of July 18-25 it ranks #4 at $4.7B, behind BSC, Ethereum and Base and ahead of Solana.

Activity on Robinhood Chain:

• DEX Volume averages around ~$500M daily

• DAUs averages around 250K users daily

• Monthly Active Users hit an ATH of 2.1 million users

• Roughly two thirds of the daily base is returning

• Transactions averages around 6M daily

• Fees averages around $200K daily

Most of the money on Robinhood’s chain is not Robinhood’s.

TVL sits at $595.1M as of July 25, up from essentially zero in mid-June. The composition is the story:

• Morpho: $236.2M

• Ethena: $177.0M

• Uniswap: $48.3M

• Maple: $48.1M

• Lighter: $25.0M

• Robinhood Stock Tokens: $24.7M

Morpho and Ethena are 70% of the chain between them. Robinhood Stock Tokens are 4.2%.

Uniswap owns the trading layer outright.

Of $397.4M in DEX volume on July 25:

• Uniswap V3: $251.9M

• Uniswap V4: $80.1M

• Uniswap V2: $58.4M

• Everything else combined: under $7M

That is 98% Uniswap. Arcus, PancakeSwap and every fork on the chain share the rest.

Memecoins still made up majority of all the DEX Volume:

• Memecoin Pairs: $297.0M (74.7%)

• Ethereum: $64.2M (16.1%)

• Robinhood Stock Token Pairs: $36M (9.1%)

• Protocol Token Pairs: $219.6K (0.1%)

One launchpad is most of the chain.

On July 25, pons.family did:

• $99.7M of $155.8M in launchpad volume (64%)

• 1.3M of 1.6M launchpad transactions (81%)

• 11.2K of 17.6K tokens deployed (64%)

The next largest, Bankr, did $13.1M.

Over seven days pons.family generated $7.86M in gross fees on $826M of volume. Gas fees for the entire chain over the same period were roughly $1.51M.

The app layer earns about five times what the L2 does.

The thing it was built for: It is small, growing fast, and broader than stocks.

Total tokenized market cap on the chain is $24.8M:

• Stocks: $19.1M

• Private equities: $1.9M

• ETFs: $1.5M

• Commodities: $1.2M

• US Treasuries: $800.1K

That is up from roughly $5.5M on June 29, about 4.5x in under a month. Tokenized-equity holders went from 29,407 to 48,470 in five days. Stock-token pairs are now 9.1% of DEX volume at $36M a day, against memecoins at 74.7%.

So what?

Robinhood built the distribution. Third parties captured the economics.

Morpho and Ethena hold the deposits. Uniswap clears the trades. Pons.family earns the fees. Robinhood collects base fees annualizing to roughly $78M.

That is the fat-app, thin-chain thesis with a public-company income statement attached.

Charts of the Week

• Hyperliquid OI hits ATH of $11.4B on July 24, 2026

• Roughly $104M of net ethereum inflows of the last two weeks while bitcoin flows are choppy

Other Notable News

• The London Stock Exchange announced LSE 24 on July 21, a 24/5 venue with agent-native access and onchain settlement, and the SEC announced a September 17 roundtable on 24-hour US equity trading two days later.

• Strategy sold $263.5M of MSTR shares between July 13 and 19 and bought no bitcoin for a fourth straight week, lifting its USD reserve to $3.225B against roughly $1.76B of annual preferred and interest obligations.

• Nine firms including Strategy, BlackRock, Coinbase and Galaxy pledged $15M over three years to bitcoin post-quantum research, with more than 7 million BTC sitting in outputs with exposed public keys.

• Regulators blew past the GENIUS Act rulemaking deadline on July 18 with zero final rules across six agencies, pushing full effect to January 18, 2027, while stablecoin supply grew 18.6% to $308.1B anyway.

• PayPal’s board formally rejected the $60.50 per share Stripe and Advent bid, hired Goldman Sachs and Evercore, and is holding out for closer to $70 ahead of its July 28 print.

• CLARITY still has no cloture motion filed, with the August 7 recess as the gate.

Thanks for reading! Go deeper on any story above with Artemis Analyst, our AI analyst built on Artemis’s institutional data. Ask it anything (”chart Robinhood Chain stock-token volume by day,” “rank SPPDA constituents by protocol revenue”) and get data-backed answers, comps and charts in seconds, right in the Terminal. Or pull live numbers straight into your models with =ART() in Excel and Google Sheets.

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 theme of the week was indices falling, led by an unwinding in risk assets amid a reescalation of the US-Iran war, and a global selloff in memory stocks. The Nasdaq100 closed the week down more than 4.1 percentage points, and the S&P500 down 1.5. There were some positive signs as PayPal rose 22% following an acquisition offer from private competitor Stripe, while Figma rose 13% as part of its month-long rebound that has been defying the “software is dead” narrative.

Memory stocks were the talk of the town in H1’26, but since their June highs, major names have seen valuations correct: Micron down 28%, SanDisk 38%. The meltdown continued into last week but there was a bright spot in the AI industry as Kimi launched their latest open weight model (soon-to-be open, rather) called K3. K3 performed well in benchmarks, ranking behind only Anthropic’s Fable 5 and OpenAI’s GPT-5.6 Sol on Artificial Analysis’ intelligence index. This reinvigorated the conversation around open vs proprietary weight models which has so many implications across the full AI stack.

Today We Highlight:

• Kalshi Introduces Compute Forward Curves

• Stripe tries to buy payments giant PayPal for $53B

Kalshi Introduces Compute Forward Curves

Pioneering a new class of market, Kalshi has done over $120B in volume so far this year. Prediction markets have captured the zeitgeist as a new type of market that enables betting, hedging or speculating on any event. Their early traction was led by politics, and more recently, has been driven primarily by sports (+75-90%). Their popularity has put them under the microscope. They’ve come under serious scrutiny from critics and regulators, who have criticized the dominance of sports-related volume, commenting that the activity is not all that different from the activity seen on regulated sportsbooks like DraftKings and Fanduel. But the promise and potential of prediction markets extends far beyond mere sports betting. Kalshi’s May raise at a $22B valuation (and search to raise again at 40B) indicates that investors are willing to underwrite that case.

This week, Kalshi took a step towards that future by introducing a first of its kind market-implied compute forward curve. This is not sports betting and it is not political speculation, its a real signal for one of the most consequential markets in decades, maybe ever. The benchmark is derived from Kalshi’s own event contracts for individual chip models, which it uses to calculate an implied future market price. For instance, the chart below shows the historical and future implied prices for NVIDIA B200’s which indicate that the market is in backwardation (future prices lower than current spot prices).

Although Kalshi is the first to build a forward curve derived from prediction market data, they are not the first to this space. Just a few weeks ago, Ornn announced the launch of their own forward curves, though you have to talk to sales before being able to view it. Silicon Data is another platform building in this space, though their data is also gated behind sales. Interestingly, Kalshi’s seems to be the only one freely accessible for anyone. Kalshi also uses Ornn compute price indices to settle many compute-related markets. On the crypto side, Venice is an interesting case study in pricing compute as its VVV/DIEM token system confers holders $1/day in compute API credits.

So what?

The compute market has been catapulted to the center of global attention by the explosive growth of artificial intelligence. Analysts project annual AI-related capex to reach $5-10 trillion by 2030 while experts are noticing that compute markets have begun to resemble traditional commodity markets like oil and natural gas. But the compute market is still relatively crude: there is no pricing benchmark or hedging mechanism, which is a problem for companies holding billions of dollars of compute on their balance sheet. These markets are missing the financial instruments used to hedge and track in standard commodities markets.

Kalshi and other prediction markets have a lot of work to do to remove the perception that they are just an alternative to traditional sports books. Search “Kalshi” on Google and you’ll see it is labeled as a “Sports betting company”. While the data at the moment supports that, prediction markets offer a vast canvas for markets and financial products. Kalshi’s compute forward curves show that they intend to continue innovating and become a core part of financial markets, especially at the cutting edge.

Stripe tries to buy payments giant PayPal for $53B

Last week, Stripe and PE firm Advent International, offered to buy PayPal at a $53B valuation, an 85% valuation drop from its ~$340B peak in the summer of 2021. PayPal’s board rejected it.

Deal terms at a glance

• Price: $60.50/share, a 28% premium to PayPal’s Tuesday (July 14) close of $47.37.

• Equity value: ~$53.4B (all cash).

• Financing: ~$50B committed bank debt + $17B equity from Stripe, Advent and Block.

• Structure: Stripe and Advent each hold 50%, with PayPal kept intact rather than broken up.

• Buyer context: Stripe was valued at $159B in a February tender offer, has raised ~$10.4B since 2010, and completed 21 known acquisitions - the largest previously being stablecoin platform Bridge for $1.1B. A PayPal deal would dwarf all of them.

PayPal has been a falling giant: in FY25 it processed $1.8T in payments volume, eclipsed for the first time by Stripe’s $1.9T. Stripe raised capital earlier this year at a $159B valuation, valuing the private company about 4x what the public market had been valuing PayPal. This might be surprising given both platforms processed a similar volume of payments in 2025, but Stripe’s outstanding growth (34% Y/Y vs PayPal’s 7%) has earned it its premium.

PayPal stock reached 9-year lows despite having nearly half a billion users and billions in free cash flow (recent quarter FCF yields have been 10-15%). But this acquisition attempt does not come as a surprise given the recent consolidation in the payments industry: WorldPay was acquired by Global Payments for $24.25B in 2025, FIS acquired Issuer Solutions in the same transaction, and Brex was acquired by Capital One.

So what?

PayPal stock jumped 16% on the news, despite its board rejecting the offer. Stripe’s attempt to take over PayPal shows their ambition to increase the GDP of the internet by having a fully integrated product suite that spans the full stack. Also, this acquisition would give Stripe full stack ownership over the stablecoin stack: Tempo as the settlement layer, Bridge for issuing and orchestration, Privvy as the wallet and management layer and, at last, PayPal’s PYUSD ($3B circulating) as the asset with existing distribution.

Charts of the week

• Robinhood Chain becomes the leading chain by DEX Volume

• USDC transfer volume on HyperEVM hits an all-time-high

  

• Kalshi’s daily active perpetuals traders surges to 60% that of Hyperliquid

• Kimi K3 ranks 3rd on Artificial Analysis’ Intelligence index

• SK Hynix ADRs listed on the NASDAQ, and the shares traded at a BIG premium to their Korean counterparts

Other Notable News

• Introducing GOFR: the Galaxy Onchain Financing Rate

• JTX enters beta

• SpaceX trades below its IPO price

• Trump’s social media platform, Truth Social, to sell priority access to posts

Thanks for reading! Go deeper on any story above with Artemis Analyst — our new AI analyst built on Artemis’s institutional data. Ask it anything (”compare Kalshi vs. Polymarket volume,” “chart spot BTC ETF flows”) and get data-backed answers, comps, and charts in seconds – right in the Terminal.

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.

Robinhood Chain is off to a fast start to democratize finance for all in <1 month.

• >300K Daily Active Addresses

• >$1B Daily Spot DEX Volume (#3 among chains)

• >$300m of stablecoin supply

• +$40m Annualized Daily Fees on Robinhood Chain

• >$300M+ in TVL driven by Morpho, Ethena, Uniswap, etc

  

I invested in Robinhood back in 2019 at Whale Rock (L/S Hedgefund) in a pre-IPO round and covered the Coinbase IPO roadshow back in 2020. I started Artemis to focus people on investing in real enduring assets in crypto and equities vs speculate on memecoins.

I was horrified and sad when I went into my Robinhood Wallet last week and saw that I could really only trade meme coins and in just 3 days since buying a nominal amount of $CASCHAT, I’ve been air dropped a ton of random and meaningless tokens (including a token called “Pointless Coin”)

Yes, Robinhood is #3 in spot DEX volume today.

BUT, majority of DEX volume is still primarily meme coins (animal memes, vlad and robinhood memes, or general memes). PLEASE, Robinhood DO NOT build a meme coin chain.

There’s a lot of learnings for Robinhood Crypto from Coinbase’s Base as it’s still a much larger chain today.

Brian Armstrong even doubled down on the quote of steering folks to real and enduring use cases in his reply.

I get the temptation as meme coins are a great way to attract early users, partners that drive liquidity and volumes (come for the memes, stay for real use cases). Large independent projects are able to flourish like Aerodrome which dominate volumes on Base and has built a real business model)

But memecoins lose people money and destroy trust.

Look no further than the Base meme coins that were launched in early 2024 that are down another 90% this year. (Down 99% from the peak in early 2024).

Meme coins AREN’T durable and hurt customers and further alienates retail from using blockchains.

Also memecoins on Robinhood Chain further the sentiment from Wall Street and Hedgefunds that Robinhood is the meme stock trading app embroidered in the GME / meme stock mania of early 2021.

Tweets like these don’t help:

Wall Street struggles to understand Robinhood as is. DO NOT double down on what hurt the brand in 2021.

Instead, focus liquidity and attention on the Robinhood Wallet and ecosystem on Arcus (former dydx team, OG / early leading perp DEX) and tokenized stocks. I love that Arcus today allows anyone in the world to trade spot tokenized stocks globally.

These stocks are up only in the last decade.

Please ACTUALLY focus on RWAs on Robinhood chain to expand financial access to more and more investors who can trade stocks and pre-IPO companies. Teams like RWA.xyz and Artemis would gladly help highlight these use cases.

Robinhood’s biggest bear case is its fully saturated the U.S market with 27m funded accounts and the street modeling to 31-32M funded accounts in FY2028 which is not explosive.

My bull case for Robinhood is access to 100M+ more international investors who can now invest in RWAs, prediction markets, stablecoins, public companies and pre-IPO stocks through Robinhood Chain and act as an early top of funnel for Robinhood App.

There’s real demand for 24/7/365 tokenized stocks -- look no further than Trade.xyz on Hyperliquid where people are trading primarily equities like SK Hynix which does $68B of actual revenue and is a real business that people want access to.

If through Robinhood Chain, Robinhood grows to say 100M MAUs in 2030 and ARPU is currently ~$171 and say international users have lower ARPU because they’re paying Chain Fees instead of trading through Robinhood App which has higher take rates, and ARPU is $100. That’s $10B of revenue on just consumer revenue alone in 2030 and beats the streets estimate of $8.78B by FY29.

Vlad and Johann, you have the opportunity to fulfill crypto’s promise of enabling anyone to access finance through tokenizing assets.

Please don’t f*ck it up by making Robinhood a memecoin chain.

Wall Street, your customers and the world will thank you for it.

Sincerely,

Jon Ma

Robinhood User since 2017.

Global risk assets closed the first full week of July on firm footing, but beneath a placid index tape the composition of leadership shifted hard. The Dow crossed 53,000 for the first time and set new record closes, capping a first half in which the Dow rose ~8.9%, the S&P 500 ~9.6%, the Nasdaq ~12.8%, and the Russell 2000 ~22% – its best first half since 1991. The macro backdrop turned friendlier: a soft June jobs report and new Fed Chair Kevin Warsh’s comment that inflation risks had come down revived hopes of a less-hawkish Fed, reversing months of “higher-for-longer” repricing. The counterweight was geopolitics: fresh attacks in the Strait of Hormuz and the revocation of Iranian sanction waivers pushed oil and the 30-year Treasury yield higher (back above 5%), lifting the discount-rate hurdle for long-duration growth just as Q2 earnings season opened. By later in the week that tension produced a genuine chip rout: Samsung’s record quarterly guidance on July 7 could not stop a semiconductor sell-off, and the Nasdaq wobbled even as the broader tape held.

The week’s scoreboard tells the rotation story in one picture. The biggest single-name winners were not the AI-hardware leaders but recovering high-beta and crypto-levered names: Rivian (+19.8%), Robinhood (+15.4%, riding its Robinhood Chain launch), Palantir (+14.6%), Coinbase (+11.4%) and Strategy (+8.7%), alongside a sharp crypto bounce (ETH +11.8%, BTC +6.5%). The benchmarks advanced only modestly (Dow +1.6%, S&P 500 +1.4%, Nasdaq-100 −0.2%). The damage was concentrated in the AI-capex complex: Corning (−23.8%) and SanDisk (−14.9%) were dumped even as the mega-cap AI names (NVDA +0.3%, AVGO +0.4%) went sideways — a textbook “sell the crowded winners, rotate into everything else” week.

Today We Highlight:

• The Great Rotation: Capex-payoff doubts pull money out of AI hardware and into "boring" value.

• Robinhood Chain: Robinhood ships its own Arbitrum-based L2 with 24/7 tokenized stocks and a day-one DeFi stack.

• Open USD: A ~140-partner consortium (Google, Visa, Stripe, BlackRock, Coinbase…) unveils a stablecoin for “global money movement.”

The Great Rotation: AI Collapses as “Boring” Stocks Rise

The most important market story of the week is that capital is rotating out AI hardware and into the value and defensive names that sat out the 2025 boom. The proximate cause is a growing investor demand for evidence that record AI capital spending will actually pay off. Big-tech AI capex is set to climb another ~50% in 2026 to well over $600bn, and the gap between investment and revenue is now wider than the 2001 telecom bubble as capex is growing roughly 46% faster than sales. With US tech/AI multiples near EV/EBITDA of ~25x, the market has begun to punish spending even when operational results beat, because the fear is margin compression as monetization lags.

Corning fell 23.8% and SanDisk 14.9% over the week even as Nvidia and Broadcom went flat and Samsung was sold despite a record quarter. That is the signature of a positioning unwind, not a demand collapse: breadth stayed positive, the equal-weighted index held up better than the cap-weighted one, and leadership migrated toward financials, insurers, healthcare, staples and midstream energy. The Dow made records while the Nasdaq-100 slipped.

The Fed and the resumed US-Iran war are accelerants. After a hawkish June (PCE at ~4.1%, Goldman pushing its next cut to 2027), the early-July dovish inflection re-rated rate-sensitive value higher; separately, Hormuz pushed oil and the long bond up, mechanically raising the hurdle for long-duration AI winners at the worst possible moment.

Robinhood Chain: From App to Infrastructure

On July 1, Robinhood launched the public mainnet of Robinhood Chain, an Ethereum Layer-2 built on Arbitrum’s stack, purpose-built for tokenized real-world assets that trade 24/7 and plug directly into DeFi. The headline product is Stock Tokens: on-chain exposure to equities like NVDA, AAPL and GOOG, structured as tokenized debt securities (economic exposure, not legal ownership), available via the Robinhood Wallet in 120+ countries, though not the US. What sets it apart from the usual corporate-chain announcement is that a working DeFi stack was live on day one: Uniswap (a dedicated AMM as primary liquidity), Lighter, 1inch, Rialto and Arcus (from the dYdX team), with Chainlink as the official oracle. Robinhood also rolled out Earn, a Morpho-powered lending product targeting ~7% APY on USDG.

The strategic read is that this is the culmination of Robinhood’s 2025 acquisition spree (Bitstamp, WonderFi, the European tokenized-equity pilots) composed into one vertically integrated, on-chain brokerage – assets tokenized on its own network, traded through its own wallet, financed through integrated lending, custodied on its own stack. The market liked it: HOOD was one of the week’s biggest winners, up 15.4%. The open question is regulatory; the line between a compliant synthetic instrument and an unregistered security varies sharply by jurisdiction, and multiple regulators (including an SEC that just added crypto rulemaking to its July agenda) are drawing those lines right now.

Open USD: the consortium stablecoin

The week’s biggest fintech headline was the unveiling of Open USD, a new stablecoin from Open Standard — an independent venture led by former Coinbase product lead Zach Abrams and backed by a roster of roughly 140 partners that reads like a who’s-who of tech and finance: Google, Samsung, IBM, Coinbase, Solana, BlackRock, Standard Chartered, US Bank, American Express, BBVA, Visa, Mastercard, BNY and Stripe. Pitched as “a new stablecoin for global money movement… built for the internet economy,” it is set to go live later this year.

The ambition is enormous; so is the incumbency it must overcome. Stablecoin supply remains extraordinarily concentrated: on Artemis data, USDT (~$185B) and USDC (~$76B) together account for the overwhelming majority of the ~$295B+ tracked, and even well-funded fintech entrants remain small — PYUSD sits near $2.8B, Ripple’s RLUSD near $1.6B, USD1 near $4.2B. Open USD would launch into the low end of that distribution. What makes it different is distribution: a consortium spanning card networks, banks and Big Tech could route real-world payment flow to a stablecoin from day one, which is exactly the “monetization model” fintechs have been circling. It is also the clearest sign yet that stablecoins have become table stakes — the same week, Standard Chartered opened direct USDC minting/redemption for institutions via Circle, and PayPal moved PYUSD to native issuance on Polygon.

Charts of the Week

Other Notable News

• DTCC began limited production trades of tokenized Russell 1000 equities, ETFs and US Treasuries in July, backed by 50+ firms (BlackRock, Goldman, JPMorgan, Circle, Ondo), with a full launch slated for October; unlike Robinhood’s synthetic model, DTCC tokenizes assets already in custody, so the tokens carry actual legal ownership

• MiCA’s transitional period closed on July 1, locking Binance out of the EU after it withdrew its Greek application on June 24; capital is flowing to licensed operators, underscored by SBI’s ~$289M acquisition of Japanese exchange Bitbank — a nine-figure purchase of “permissions”

Thanks for reading! Go deeper on any story above with Artemis Analyst — our new AI analyst built on Artemis's institutional data. Ask it anything ("compare Kalshi vs. Polymarket volume," "chart spot BTC ETF flows") and get data-backed answers, comps, and charts in seconds – right in the Terminal.

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.

Risk assets sold off hard this week. The mid-June bounce on US-Iran de-escalation had already faded. Bitcoin slid below $60,000 on Thursday, June 25, for the first time since 2024, opening near $60,983 and trading down to roughly $59,300 intraday. Ethereum falling in step from about $1,620 toward the mid-$1,500s. The selling was mechanical as much as macro: spot bitcoin ETFs bled for 6 consecutive weeks. The lone exception was DeFi: AAVE ripped higher on a fresh Standard Chartered initiation with a $3,500 by 2030 target, Aave V4 traction with deposits past $200 million, and USDT deposits flowing back into the protocol, even as the rest of the screen turned red.

DeFi stood alone: AAVE (+30.18%) and SOL (+3.49%) were the only names in the green, while SKY (-14.06%), HOOD (-8.75%) and COIN (-8.70%) lagged, with NVDA (-8.62%) and CRCL (-8.30%) close behind. Unlike last week, AI and semis offered no cover (NVDA -8.62%, QQQ -4.60%) and gold slipped (GLD -3.48%); only the Dow (DIA +0.43%) and Micron (MU -0.15%) finished near flat. The basket averaged -3.10% on the week, median -3.64%.

Today We Highlight:

• Bitcoin breaks $60K as spot ETFs bleed for 6th consecutive weeks

• Strategy’s treasury sinks roughly $13.3B underwater, hits a 23-month low, and the repair job drags on

• Robinhood closes an upsized $2.2B zero-coupon convertible, raising cheap capital into the teeth of the selloff

Bitcoin breaks $60K as spot ETFs bleed for 6th consecutive weeks

The headline number is simple: bitcoin traded below $60,000 on Thursday for the first time since 2024.

US crypto ETF flows have flipped from bid to bleed. Net outflows ran a sixth straight week, with roughly $227 million out in the latest week, and the damage runs deeper than the weekly tape: back-to-back record outflow streaks in May and June pulled an estimated $7.2 billion out, IBIT alone shedding about $3.3 billion, and tipped 2026 flows negative on the year for the first time. The 2024-25 institutional buyer has became a seller for now.

Strategy’s treasury goes underwater as the STRC repair drags on

Last week the story was STRC breaking its peg. This week it spilled over into the equity.

MSTR fell to its lowest level in nearly 23 months, with the stock testing the low-to-mid $90s intraday and printing into the $80s on the worst sessions. The reason is arithmetic: Strategy holds 847,363 BTC bought for about $64.10B at an average cost near $75,651 per coin.

With bitcoin around $60K, that treasury is now roughly $13.3B underwater, the first time the core trade has been meaningfully offside since the model was built.

The critics got louder this week, as STRC came under immense pressure (Cash Reserves to USD coverage roughly sits around 9.8 months at point of writing). Peter Schiff escalated his attack, framing the STRC structure as a Ponzi, while CryptoQuant's CEO argued Strategy should halt bitcoin buys outright and rebuild cash. The slide traces back to the company's roughly $1.5 billion convertible-note buyback in May. With a software business that covers only a fraction of more than $1.2 billion in annual preferred-dividend obligations, the funding gap is bridged almost entirely by capital markets (selling ATM shares).

Robinhood closes a $2.2B zero-coupon convertible into the selloff

The contrast of the week: while crypto-native balance sheets delever, Robinhood walked into the convertible market and walked out with $2.2B at a 0.00% coupon. exercised the $200M greenshoe, and closed the upsized deal June 25, with about $290M earmarked for buybacks and $123.2M for capped calls that lift the effective conversion price to $174.42. This effectively push the no-net-dilution threshold to roughly $237.85, or about $304 after the repurchase.

Shares dipped a reflexive 4% on dilution worries, but a zero-coupon convertible is essentially free optionality financing. Robinhood is using it opportunistically rather than out of need: Truist reiterated a Buy with a $100 target, citing record June volumes in equities, options, and prediction markets, with crypto the lone soft spot. Raising long-dated, zero-cost capital into a risk-off tape is a move only platforms with real balance-sheet credibility can pull off right now

Charts of the Week

• Prediction Markets continue to hit ATH in trading volume driven by World Cup.

• RWA perp volume continue to increase

Other Notable News

• The Senate passed the 21st Century Road to Housing Act on June 22, which bans the Federal Reserve from issuing a retail CBDC through 2030

• The DOJ seized infrastructure tied to Huione in a crypto money-laundering action

• Binance will suspend most EU services from July 1 after withdrawing its Greek MiCA application on June 24, and is now pursuing a license in France; funds remain withdrawable and ESMA has told unauthorized providers to stop onboarding EU clients

• BofA changed its base case to three quarter-point Fed hikes in 2026, with FedWatch pricing a 60.7% chance of a hike by October

Thanks for reading! Stay ahead this week by using the Artemis Terminal to pull the underlying data on any of the stories above (COIN segment revenue and pre-IPO perps volume, spot BTC ETF flows, HYPE fees and buybacks) or pull live numbers straight into your models with =ART() in Excel/Google Sheets.

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.

We recently hosted a Artemis’ x DormDAO Quant Blockchain Competition, where teams chose one of two research tracks: build a systematic crypto factor strategy or design a prediction market strategy.

It was to reward rigorous research, clear reasoning, and thoughtful strategy design.

Our winner’s from SMU (@smu_blockchain), @Xynersrn and @kangnuza, built a two-layer quant strategy: first, a BTC regime filter decides whether the market is worth taking risk in; then, a cross-sectional model ranks a basket of crypto assets and crypto-linked equities using momentum and funding signals.

The result is a strategy that spends most of its time in cash, but concentrates capital when both the market regime and asset-level signals line up.

Without further ado, here is the winning memo from our Quant competiton.

Abstract

Two layers, one decision per week. Layer 1 (a 5-factor BTC composite) decides whether to be in the market. Layer 2 (a 2-factor cross-sectional ranking on 7 Hyperliquid perps) decides what to hold. The strategy is flat 83% of the time and concentrates capital in the windows where both layers agree.

Over January 2022 through May 2026 this posts Sharpe 1.31, max drawdown -27.2%, and 5.63x final equity. Net of Hyperliquid fees and tiered slippage: Sharpe 1.27, 5.33x. The strategy beats BTC buy-and-hold (1.93x, -64.3% DD) and an equal-weight alt portfolio (4.41x, -66.6% DD) on every metric while spending only 17% of weeks deployed.

Net of transaction costs (Hyperliquid taker fees + tiered slippage modelled per-rebalance): Sharpe 1.27, 5.33x, 47.0% annual return. Cost drag is 3.1% of Sharpe and 5.3% of final equity.

Performance metric convention. Throughout this report, annualised return is geometric CAGR

((final/initial)^(52/N) − 1) and Sharpe is CAGR / (σ_weekly × √52). The textbook

arithmetic-mean Sharpe is slightly higher because variance drag pushes geometric below arithmetic for high-vol return streams; we report the CAGR convention, so all numbers tie back to the final equity figure.

1.2 Architecture in One Picture

T-1 lag on every signal. Weekly rebalance on Friday close. Both layers were optimised independently before being combined.

2. Why Two Layers

Altcoins are levered bets on Bitcoin. The correlation is structural: when BTC trends down, alts get crushed regardless of individual fundamentals. A pure cross-sectional ranking of alts would still hold positions through bear markets and suffer catastrophic drawdowns, our equal-weight alt control portfolio finishes at 4.41x but takes a -66.6% drawdown along the way.

The regime gate is not a market-timing signal. It does not call tops or bottoms. It is a blunt filter that asks: is the macro environment favourable for taking alt risk? When the answer is no, the strategy steps aside.

This produces two complementary edges:

• Regime timing: avoiding the -64% drawdowns that come from being long during bear markets.

• Cross-sectional selection: we pick the strongest alts within each BULL window.

Neither alone produces the full result. The regime gate applied to BTC alone (long BTC during BULL, cash otherwise, 2020 to 2026) returns 3.72x at -47% DD; better risk-adjusted than BTC buy-and-hold but well short of what the alt layer adds. The alt engine without the gate is the equal-weight alt portfolio (4.41x, -66.6% DD). Stacked together, they produce 5.63x at -27.2%. The gate routes capital into the top 3 ranked alts instead of BTC.

The 2022 stress test makes this concrete: BTC drew down 64%, the equal-weight alt portfolio drew down 66.6%, the strategy drew down 0%. The regime gate kept us flat for the entire year. Most of the strategy’s risk-adjusted edge over the benchmarks comes from this kind of avoided drawdown; most of the absolute return comes from the alt layer in BULL years (see §6.5 for the year-by-year split).

3. Layer 1, BTC Regime Gate

3.1 Signal Composition

The BTC composite is a weighted sum of five signals, each normalised to 0-100 via a rolling 2-year min-max:

The 2-year rolling normalisation is non-negotiable; full-sample normalisation would leak future highs and lows into historical scores. The trade-off is a 104-week warmup; the integrated backtest starts in 2022, safely past it.

3.2 T-1 Lag

The regime decision for the coming week uses only data available at the prior Friday’s close. This applies to every signal before classification, every price before entry, and every factor score before ranking. A strategy cannot trade Friday’s close on Friday’s data.

3.3 Classification

The NEUTRAL band prevents whipsaw at the boundaries. When the score hovers, the strategy waits for conviction rather than flipping.

3.4 Threshold Choice

The shipped BULL threshold is 60. Sensitivity testing across the full sample:

Threshold 55 produces the highest in-sample Sharpe (1.46), but at a wider drawdown (-35.1% vs -27.2%). We ship 60 deliberately, not because it is the Sharpe peak but because:

1. Drawdown control matters more than the extra 0.15 Sharpe for a strategy positioned around risk discipline.

2. Walk-forward conservatism. The training-window optimum was 50 (Sharpe 1.25); its OOS Sharpe of 2.07 came with -33.1% DD. Picking the in-sample optimum would be over-fitting.

3. The relationship is well-behaved across 50-60. Sharpe is comparable; only at 65 does the gate become too restrictive and Sharpe collapse to 0.83.

We are explicitly not running at the in-sample Sharpe peak.

3.5 Regime Distribution

Two windows matter, and they give slightly different distributions:

The integrated subset is more BEAR-heavy because it excludes the late-2020/2021 bull run and includes the 2022 bear plus the 2025-26 BEAR stretch. In both windows the strategy is flat during BEAR and NEUTRAL , 83% of the integrated backtest (51% BEAR + 32% NEUTRAL), and 82% of the full diagnostic window. BULL weeks (17-19%) are the only weeks capital is deployed.

3.6 BTC-Only Validation and Design Evolution

As an intermediate check, we tested the gate on BTC alone, long BTC during BULL, cash otherwise, over Jan 2020 → May 2026. Three regime variants and the buy-and-hold benchmark:

Option B matches BTC’s Sharpe (0.77 vs 0.77) at less than half the drawdown (-47% vs -77%). It does not match BTC’s absolute return (3.72x vs 11.24x); that gap is exactly what the alt layer closes in the integrated strategy.

The ETF Inflows signal was originally included at 20% weight but removed on May 5, Option B Sharpe rose from 0.64 to 0.77 on removal. The signal’s short history (Jan 2024+) prevented the 2-year rolling normalisation from working.

4. Layer 2 , Alt Factor Engine

4.1 The Two Active Factors

We began with five factors. Two remain , revenue growth and activity momentum were removed on May 7 after factor attribution showed each contributed +0.01 Sharpe. They had Artemis on-chain data for only 2 of 7 assets (SOL, HYPE) and added no cross-sectional differentiation. Open-interest confirmation was planned at 10% weight but Hyperliquid only exposes current OI snapshots, not historical series.

Both factors use cross-sectional ranking: score = 1.0 − (rank − 1)/(n − 1), where n counts only assets with real data. Assets without data on a factor receive neutral 50, neither benefit nor penalty. The model degrades gracefully: when funding is silent across the universe, ranking falls back to momentum-only.

Funding rate. Crypto-native, not in the Fundamentals 1 factor set. Negative funding means the marginal trader is short and paying a premium; positioning extremes tend to revert.

HL fundingHistory returns data from May 2023 onward for SOL/BNB/HYPE and from January 2026 onward for XMR. Bybit linear-perp funding fills each token’s pre-HL window via bybit_funding_pull.py, so SOL, BNB, and XMR carry real funding signal across the full backtest. HYPE is silent before its Dec 2024 TGE (handled by the phantom-pick gate, §5.4).

After splicing, every token in the universe registers nonzero funding in every week post-May-2023 except HYPE pre-TGE; the cross-sectional rank is well-populated whenever the strategy is active. Equity perps (COIN, HOOD, CRCL) carry HIP-3 funding from their respective HL listing dates forward; pre-listing equity weeks score neutral 50 on funding, since no cross-venue equity-perp proxy exists.

Price momentum (7-week). Selected through a grid search across {2, 3, 4, 5, 6, 7, 8, 10}-week windows. The 7-week window dominates by a clear margin:

Marginal Sharpe contribution: -0.35 (removing momentum drops Sharpe from 1.31 to 0.96). It is the dominant factor.

Factor interaction. When funding is silent (most weeks), momentum provides the only cross-sectional differentiation. When funding fires, the combined signal is stronger than either alone. The shipped model (Sharpe 1.31) clearly beats momentum-only (1.18) and funding-only (0.96).

4.2 Portfolio Construction

Every BULL week:

1. Rank all 7 assets by composite alt score (0.45 × funding + 0.55 × momentum)

2. Select top 3 (or fewer if fewer have valid scores)

3. Weight by signal strength: w_i = score_i / Σ top_3_scores

4. Cap any single asset at 40%, redistribute excess to the uncapped survivors

5. Apply leverage by BTC conviction:

6. BTC ≥ 70 → 2.5x

7. BTC 60-70 → 1.5x

Why these parameter choices. Each is a tuned design decision:

• Top 3 selection. Top-1 concentrates risk in a single name; top-5+ on a 7-asset universe selects almost the whole investable set and dilutes the cross-sectional signal. Top-3 keeps differentiation while still allowing diversification across the typical 5-7 available names.

• 40% single-asset cap. With top-3 and signal weighting, an extreme score distribution can produce raw weights as high as ~70% on the best name. The 40% cap floors the second and third positions at real weight, preserving diversification benefit. We tested removing the cap (concentration up to 70%) and Sharpe dropped. The cap is not just defensive; it is positive-EV.

• Leverage tiers (1.5x / 2.5x at the 60 / 70 boundary). Two tiers keep the model interpretable.

The 60 boundary is the BULL classification threshold; the 70 boundary marks high-conviction BULL where adding leverage is risk-justified. We tested continuous leverage scaling (BTC score → leverage as a smooth function) and discrete tiers performed comparably with far less complexity. Leverage above 2.5x produced larger drawdowns without proportional return improvement.

• Friday rebalance. HL funding settles on a 8-hour cadence and the week-end Friday close is the natural alignment point for weekly factors. Monte Carlo with Mon-Thu rebalance days (§6.3) confirms the strategy is not dependent on Friday; it works across all weekday rebalance points, with median Sharpe similar across days.

4.3 Removed and Why

• Revenue growth , +0.01 Sharpe delta. Data for 2/7 assets. Removed May 7.

• Activity momentum , +0.01 Sharpe delta. Data for 2/7 assets. Removed May 7.

• OI confirmation , never built. HL provides only current snapshots, not history.

• Short overlay (BEAR). Tested twice. Original (May 7) vs 1.13 baseline: Sharpe 0.48,

DD -75%. Re-tested May 10 vs current 1.31 baseline: bottom-2 short → Sharpe 0.10, DD -86.1%, 1.38x. Top 2 short → Sharpe 0.05, DD -81.7%, 1.18x. Both destroyed capital. The factor model identifies hated, oversold assets. In bear-market rallies, those are exactly the assets that bounce hardest. Ranking for longs does not invert to ranking for shorts. Permanently rejected.

5 Investable Universe

5.1 Selection Criteria

Two hard constraints from Artemis: (1) listed on Hyperliquid with active perp markets, (2) Artemis price data coverage. Beyond that, we wanted assets where the factor model could rank effectively; usable funding history, a clear value-driver story, and enough cross-sectional diversity that the top 3 selection isn’t picking among near-duplicates.

We deliberately kept the universe compact (7 assets) for three reasons. First, every asset must clear a real value-driver bar; we didn’t want to pad the count with minor listings the model can’t evaluate honestly. Second, a smaller universe lets each name carry real weight in the cross-sectional rank rather than diluting signal across many similar tokens. Third, with a 40% concentration cap and top 3 selection, there’s no operational benefit to a 15-asset universe; the extra asset is rarely picked.

The active count varies from 3 (early 2022: SOL, BNB, HOOD) to 7 by December 2025. During BULL weeks the strategy actually trades, 5-7 assets are available, the flat 2022 period predates most BULL weeks, so the strategy never trades on the smaller set.

5.2 Per-Asset Rationale

The 7 assets span six distinct value-driver categories. This matters: a cross-sectional ranking model only adds value when the inputs have meaningfully different return drivers. A universe of seven L1 alts would mostly produce correlated picks; a universe with structural variety lets the factor model differentiate during real regime shifts.

SOL, L1 protocol token. Leading alt L1 by on-chain throughput and ecosystem breadth. Artemis provides full coverage (fees + DAU) , one of only two assets with usable on-chain data. Price history spans the full backtest. Canonical “alt L1 beta” position.

HYPE, DEX token (Hyperliquid native). The leading perps DEX by volume , a deliberate venue bet.

HIP-3 equity perps are live and HIP-4 prediction markets are on the roadmap. Artemis covers fees and DAU. TGE’d 29 Nov 2024; gated before that date. We carve DEX out from L1s because HYPE’s value driver is venue fee capture, not L1 settlement throughput.

BNB, CEX token. Value drivers are Binance ecosystem flows (burns, launchpad, fee discounts), not on-chain BNB Chain activity. We tested including BNB Chain metrics and Sharpe collapsed (1.08 →

0.59). BNB competes on price + funding only; this is the empirically defensible specification.

XMR, Privacy coin (diversifier). Structurally different holder base and use case from other alts. No on-chain data is possible (shielded chain). HL only listed XMR perps on 2026-01-16, so funding is Bybit-spliced for the ~140 pre-HL weeks (the same splice methodology is applied to SOL and BNB for the ~70 pre-May-2023 weeks where HL has no fundingHistory data; see §5.4). The XMR splice is signal-critical: it lets XMR be picked in 9 of 38 BULL weeks (24%) instead of permanently scoring neutral-50 on funding.

COIN, CEX equity. Coinbase, largest US-domiciled crypto exchange. Quarterly revenue from Artemis (10+ quarters). Fixed-cost business provides revenue-leverage beta to crypto volume. HL HIP-3 from 25 Nov 2025; yfinance backfills pre-listing price.

HOOD, CEX equity (retail brokerage). Robinhood; picks up broader retail risk-on flow beyond pure crypto-native channels. Crypto revenue ~40-50% of total during bull cycles. HL HIP-3 from 26 Nov 2025.

CRCL, Stablecoin equity. Circle (USDC issuer). Revenue = float × interest rates , the only asset with a macro driver uncorrelated to “crypto goes up.” IPO’d 5 Jun 2025; gated. Quarterly history too short (2 quarters) for ranking , competes on price + funding for now.

The construction summary:

Diversity is measured, not asserted. The pairwise correlation matrix of weekly returns over the full backtest:

The off-diagonal average is 0.34. Within-category averages are higher as expected (token perps 0.35, equity perps 0.58). With the six-category split above, the only within-category pair is COIN-HOOD (0.64); excluding that, the strict cross-category average is 0.34, and the token × equity cross-block average is 0.28 , token perps and HIP-3 equity perps are moderately correlated in weekly returns. This is exactly what a cross-sectional ranking model wants: when the token sleeve and equity sleeve are picking among genuinely different return streams, top-3 selection is optimising over real cross-sectional information rather than near-duplicates.

Caveat, survivorship bias. Every asset in this universe survived 2022 to 2026, and most performed strongly: SOL roughly 20x’d from its 2022 low, HOOD and COIN rode the crypto-equity recovery, HYPE rallied sharply post-TGE. We did not select these names at random from HL’s full listing; they were chosen because they have recognisable tickers, liquid markets, and a clear value-driver story. A universe constructed with the benefit of hindsight will always make any ranking model look better than it would have in real time.

Past performance does not guarantee future results: the factor relationships that held over this window may not hold over the next cycle, and the assets that dominated this backtest may not dominate the next. This is the single largest source of uncertainty between backtest and live performance, and we treat it as the dominant risk in §8.

5.3 Universe Table

5.4 Special Cases

• Phantom vs venue-mismatch. Only HYPE (gated before 2024-12-06 TGE) and CRCL (gated before 2025-06-06 IPO) physically did not exist at some point during the backtest; picking them pre-existence would be a lookahead error. The other five assets traded on other venues before their HL listings; we treat those as venue-mismatch (defensible with backfilled prices), not phantom. An earlier blanket HL-listing gate was abandoned May 9 because it conflated the two cases.

• BNB on-chain exclusion. BNB Chain metrics are available under Artemis ID bnb, but including them in cross-sectional ranking degraded Sharpe from 1.08 to 0.59 on the earlier baseline. BNB is an exchange token; its value drivers are CEX burns, launchpad allocations, and fee discounts, not on-chain BNB Chain activity. The model treats BNB as competing on price + funding only.

• Bybit funding splice (SOL, BNB, XMR). HL fundingHistory returns data from May 2023 onward for SOL/BNB/HYPE and from January 2026 onward for XMR. Bybit linear-perp funding fills each token’s pre-HL window via bybit_funding_pull.py (SOL/BNB ~70 weeks, XMR ~140 weeks). The XMR splice is signal-critical (XMR is selected in 9 of 38 BULL weeks, all inside the Bybit-only window). The SOL/BNB splice is precautionary: the SOL/BNB Bybit window (Jan 2022 → Apr 2023) is entirely BEAR/NEUTRAL in this backtest, so it affects zero picks, but we keep it for methodological consistency (one principled rule: “HL where available, Bybit as cross-venue proxy where not”) and to avoid having to defend a cherry-picked splice scope. Within the overlap window (post-May-2023), HL is primary; Bybit is used only where HL has no data.

• Equity perps as crypto-native assets. COIN, HOOD, CRCL are crypto-native businesses whose quarterly revenue tracks industry cycles. They provide equity exposure within the same execution venue as token perps; no separate brokerage, no equity-market hours constraint, funding mechanism identical to tokens.

• BTC as a tradeable position, tested and rejected (May 10). Adding BTC as an 8th asset dropped Sharpe from 1.31 to 1.13 and equity from 5.63x to 4.41x. BTC was never selected in any of 38 BULL weeks; risk-on regimes favour higher-beta alts on both factors. Adding BTC still hurts performance because it shifts the cross-sectional normalisation denominator. BTC’s informational value is as a regime gate, not a portfolio position.

6. Validation

6.1 Walk-Forward (Out-of-Sample)

The BULL threshold was optimised on 2022 to 2023 and tested on 2024 to 2026, with all other parameters held at shipped values.

Train-window optimiser chose threshold 50 (looser than shipped 60). The test-window Sharpe of 2.07 is not strong OOS validation; the test window contains 65 of the backtest’s 104 lifetime BULL weeks, making it structurally the easier half. The walk-forward shows that a threshold fit on a low-BULL window generalises directionally (Sharpe stays strongly positive), but the magnitude of the test result is driven by the 2024 to 2025 bull market, not by the threshold choice. We ship 60, which is more conservative than the train-window optimum of 50.

6.2 Transaction Costs

Hyperliquid taker fees (0.035% per notional) plus tiered slippage (0.02% for <$50K, 0.05% for $50K-$200K, 0.10% above). Costs charged on rebalance turnover only, not on static holds.

Low turnover keeps drag manageable. Average weekly turnover is 9.6% of portfolio value (including flat weeks). On traded weeks only, turnover runs ~57%: a full entry or exit turns over 100%, while weeks where the top-3 ranking is unchanged turn over 0%. The 1.8pp annual return drag is far smaller than a naive full-round-trip-every-week model would estimate.

6.3 Monte Carlo Stress Testing

250 backtest runs with randomised perturbations:

• Rebalance day: Mon-Thu (shipped: Fri)

• BULL threshold: ±3 points (shipped: 60)

• 1-2 random asset dropouts per run (shipped: 7)

24.8% of runs exceed Sharpe 1.0; 82.0% exceed 0.5. The shipped Sharpe of 1.31 sits in the upper tail of this distribution. The gap between median (0.77) and shipped (1.31) indicates genuine specification sensitivity; the chosen configuration sits in the upper end of what was achievable. Dropping 1-2 of 7 assets at random is mechanically the heaviest perturbation (removing a high-contributing name like SOL or HOOD materially shrinks the cross-section available for ranking), though we have not formally decomposed the three perturbation sources. The strategy stays profitable across most configurations, but the shipped result is an upper-tail outcome, not the central expectation.

6.4 Parameter Optimisation Sweep

The momentum window and factor weights were selected through a 112-configuration grid search: 7 momentum windows × 4 thresholds × 4 weight combos {40/60, 45/55, 55/45, 60/40}. The full grid is at results/integrated/optimization_sweep.csv.

The sweep’s Sharpe-max configuration is 7-week momentum, 45/55 weights, threshold 55 (Sharpe

1.46, DD -35.1%). Shipped is the same momentum and weights but threshold 60 (Sharpe 1.31, DD -27.2%), chosen for drawdown control and walk-forward conservatism (see §3.4).

6.5 Sub-Period Breakdown

The full-sample Sharpe of 1.31 averages over a window with very different sub-periods. Year-by-year:

Year return (YoY) is the realised year-on-year equity change (End equity / prior year’s End equity − 1), so the yearly returns compound exactly to the headline 5.63x final equity. Per-year Sharpe is computed on that year’s weekly returns under the same convention as §1.1.

2024 is unambiguously the dominant year: 24 of 38 lifetime BULL weeks, +149.6% return, Sharpe 3.17, and the entire -27.2% peak drawdown sat within it. This concentration is a real concern and we do not hide it: a strategy that posts most of its excess return in one year is heavily exposed to that year being repeatable.

That said, excluding 2024 the strategy still posts Sharpe 1.18 and 2.26x equity over the remaining 175 weeks (combining 2022, 2023, 2025, 2026 YTD). The 2025 sub-period in particular , Sharpe 1.77 over a different sub-cycle than 2024 , provides a second positive observation rather than a single one. The strategy is concentrated, not single-window. Whether 2024’s magnitude generalises to future cycles is unknowable; that the framework produces positive risk-adjusted returns across multiple sub-periods is the most we can claim from in-sample data.

7. Factor Attribution

Each factor was zeroed individually and the backtest re-run; the remaining factor renormalised to 100%.

Momentum at 7 weeks is the dominant factor: removing it drops Sharpe by 0.35. Funding is secondary but still matters removing it widens drawdown from -27.2% to -35.4%. Both contribute materially.

The momentum-only model (Sharpe 1.18, -35.4% DD) shows that momentum alone carries most of the signal but loses funding’s diversifying effect. The funding-only model (Sharpe 0.96, -23.3% DD) shows that pure positioning ranking is positive risk-adjusted but lower equity (4.11x vs 5.63x).

Both factors matter in every specification tested, though their relative weight shifts with the momentum window; at shorter lookbacks funding dominates, at longer windows momentum leads. The qualitative conclusion is stable across baselines.

8. Critical Evaluation

8.1 Data Limitations

• On-chain coverage is narrow. Only SOL and HYPE have Artemis fees/DAU. Privacy coins (XMR), exchange tokens (BNB), and equity perps don’t have on-chain activity in the sense Artemis measures. The revenue/activity factors were removed precisely because of this gap. Structural to the crypto data environment, not a fixable pull issue.

• No historical OI from Hyperliquid. Planned 10% weight, blocked by platform. Factor is designed and ready if HL adds historical OI.

• Short backtest window. Jan 2022 → May 2026, ~4.25 years, one crypto cycle. Bounded by HL’s history (perp dex mid-2022, fundingHistory mid-2023). One-cycle validation carries uncertainty about structurally different regimes.

• Survivorship bias. Discussed at length in §5.2 , every asset in the universe survived 2022 to 2026 and most performed strongly. We do not model delistings or death events. The relationships that held over this single cycle may not hold over the next, and the gap between backtest and live performance is almost always negative. This is the dominant source of risk-of-being-wrong in the backtest, more than any specification choice.

8.2 Model Risk

• Specification sensitivity. The Monte Carlo distribution puts the shipped configuration in the upper tail (Sharpe 1.31 vs median 0.77). The walk-forward and full sensitivity suite partially mitigate this, but the gap is a real signal that specification matters. We have six tuned choices (BULL threshold, factor weights, momentum window, top-N, cap, leverage tiers) on a ~4-year backtest; within defensible range but worth flagging.

• BTC decorrelation. The architecture assumes alt returns are levered BTC returns during BULL and levered BTC losses during BEAR. This relationship has held across multiple cycles but it is empirical, not law.

• Funding mechanism change. The funding factor depends on Hyperliquid’s specific 8-hour cadence and 0.01% per-period cap. A change to the mechanism would alter signal properties with no historical precedent to recalibrate.

• Equity perp regulatory risk. COIN, HOOD, CRCL trade on HL’s HIP-3 dex. Regulatory action against equity perps would shrink the universe to 4 crypto-native assets and reduce cross-sectional selection.

• Static universe composition. The backtest uses a fixed 7-asset universe. In practice, perps delist, liquidity migrates, and new listings create opportunities. A static universe embeds an implicit survivorship benefit, assets that would have been delisted or become untradeable are never removed. Live operation would require periodic universe review, and the judgment calls involved in adding or removing assets are a source of divergence between backtest and live performance that the backtest does not capture.

8.3 Tested and Rejected

Every material design path explored and abandoned, with dates and metrics:

8.4 Forward-Looking Risks

Three structural changes could break this strategy:

1. HL funding mechanism change alters the signal with no historical precedent.

2. BTC-alt decorrelation, the gate loses its rationale; strategy is flat through alt rallies that occur outside BTC BULL.

3. Equity perp regulatory action, universe shrinks to 4 crypto-native assets.

None of these are priced into the backtest.

9. Differentiation vs Artemis Fundamentals 1

Fundamentals 1 is the natural reference: a long-short, vol-targeted, ~24-asset factor strategy with reported Sharpe 1.73. Our 1.31 is lower in absolute terms, but achieved with deliberately different design choices that are worth being explicit about:

The trade is intentional: we accept a lower Sharpe for a simpler, long-only, smaller-universe strategy that is easier to operate, easier to interpret, and that avoids the implementation burden of long-short rebalancing across 24 assets. The funding-rate factor is the qualitative differentiator: it is crypto-native, picks up positioning information that price-only signals miss, and is not part of the Fundamentals 1 set. The equity-perp inclusion is the structural differentiator; it lets us trade crypto-correlated equity exposure on the same venue as the underlying tokens, without the equity-market-hours and brokerage friction TradFi factor strategies face.

10. Reproducibility

10. 1 Quick Start

pip install -r requirements.txt python strategy/alts/backtest.py
macOS users: substitute python3 for python and pip3 for pip — recent
Full source at https://github.com/ImNuza/artemis-juara

The integrated backtest reproduces all headline numbers. Reads from data/alts/ and data/btc/, applies the Layer 1 regime gate, runs the Layer 2 factor ranking, and outputs the equity curve and performance metrics.

10. 2 Analysis Suite

python strategy/alts/walkforward.py # §6.1 python strategy/alts/cost_model.py # §6.2 python strategy/alts/monte_carlo.py # §6.3 python strategy/alts/sensitivity.py # §3.4 python strategy/alts/factor_attribution.py # §7 python strategy/alts/momentum_sensitivity.py # §4.1 python strategy/alts/position_log.py # regen docs/position_log.md

10.3 Data Refresh

python strategy/data_pull/hl_data_pull.py # token perps + funding python strategy/data_pull/pull_hl_equities.py # HIP-3 equity perps python strategy/data_pull/pull_equities.py # yfinance pre-listing backfill python strategy/data_pull/artemis_data_pull.py # Artemis fees + DAU python -m strategy.btc_regime.run # regenerate BTC regime CSV

10.4 Code Structure

strategy/
├── btc_regime/ # ImNuza: BTC regime gate
├── alts/ # Xyners: integrated backtest + analysis
└── data_pull/ # data fetch scripts
data/
├── btc/ # BTC regime inputs
└── alts/ # alt prices, funding, fees, regime CSV
results/
├── btc_regime/ # BTC regime outputs
└── integrated/ # integrated backtest outputs
docs # documentations

A complete week-by-week position log , every rebalance, weights chosen, leverage applied, BTC score, weekly return, is at docs/position_log.md.

Price source hierarchy. Hyperliquid is primary for every asset, we trade on HL, so HL’s mark/close is the correct fill reference. Token perps use the main HL dex; equity perps use HIP-3 from each asset’s HL listing date forward. yfinance is used only to backfill equity history before HL listing.

Phantom-pick gate. HYPE excluded before 2024-12-06 (TGE 2024-11-29). CRCL excluded before 2025-06-06 (IPO 2025-06-05). As they did not exist. All other assets traded on other venues before their HL perp listings.

Appendix A: Key Files

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.

Risk assets round-tripped through the week. Bitcoin and ether had already been beaten down in early June as the US-Iran war and the threat to the Strait of Hormuz drove crypto into full risk-off, with bitcoin erasing all gains made since the start of the war and sliding below its pre-war levels around June 4. Sentiment then flipped: by June 16 both bitcoin and ether posted their strongest opening levels in about two weeks following news that the US and Iran were making meaningful strides toward a peace deal that would reopen the Strait of Hormuz, and as of June 17 bitcoin and ether were up 6.4% and 9.3% respectively versus a week earlier. The rally faded into Friday: the Fed under new Chair Kevin Warsh left rates unchanged on June 17 and signaled rate cuts are all but off the table for 2026, and a strengthening dollar pressured crypto, gold and silver lower, with bitcoin opening near $62,900 on June 19, down 2.4% on the day.

Brokers and DeFi led: HOOD (+17.3%), AAVE (+16.0%) and HYPE (+15.7%) topped the table, while MSTR (-6.3%), CRCL (-2.9%) and BTC (-1.0%) lagged. AI/semis were firm (TSM +9.7%, NVDA +2.8%, QQQ +3.3%) and gold was flat (GLD +0.2%) as the post-Fed dollar bid capped it.

Today We Highlight:

• SpaceX drives Pre-IPO trading; popular pre-IPO market Ventuals sunsets

• Canton announces $355 raise to grow its institutional-grade blockchain

• Strategy’s is in the midst of its worst depeg ever

SpaceX goes public, and Hyperliquid ran the unofficial IPO

The month’s marquee event: SpaceX sold roughly 555.6 million Class A shares at $135, raising about $75 billion at a ~$1.75 trillion valuation, marking the largest IPO in history. Its valuation is enough to make it roughly the seventh-largest US company, just above Tesla.

This year, a new class of market has been driving price discovery ahead of IPOs. Trade.xyz launched the SPCX-USDC pre-IPO perpetual on Hyperliquid on May 18 at a $150 reference price (a $1.78T implied valuation); it spiked to $216 within hours before settling at $202.89, up 12.7% on the day. By IPO week, SPCX accounted for roughly 30% of all HIP-3 trading volume on Hyperliquid, with aggregated open interest above $225M and more than $2.2B in cumulative volume. SPCX aggregated VWAP sat near $155, a 15% premium over the $135 IPO price, and printed near $172 on listing morning.

The pitch is that these synthetic markets price accurately: the earlier Cerebras pre-IPO perp closed within 3% of its Nasdaq open, versus ~35% spreads on traditional platforms, which is exactly the kind of performance that builds a reputation. Even NYSE/ICE chief Jeff Sprecher publicly called Hyperliquid “bigger than NASDAQ,” noting it’s run by just “11 people” (Coindesk).

The flip side of the pre-IPO boom is consolidation. Ventuals is the project behind the OpenAI and Anthropic perpetual markets on Hyperliquid, having facilitated over $650M in trading volume. On June 15 it is winding down and that its team will join another project building within the Hyperliquid ecosystem, and the team framed it as “one leg of the journey ends, and a new one begins”. Ventuals was one of the go-to markets for getting exposure to the equity of major private AI labs like OpenAI and Anthropic.

Ventuals went out having proven the demand: it generated more than $650 million in trading volume and raised over 500,000 HYPE through its vHYPE staking mechanism. The team also confirmed the points and referral programs are discontinued and that no Ventuals token will launch, since the wind-down removes any path to a value-accruing token. The bigger signal: TradeXYZ now accounts for nearly 97% of HIP-3 trading volume, making it the dominant operator in the category. For context on scale, Hyperliquid processed roughly $234 billion in perpetual futures volume over the past month.

Canton lands a $355M+ round as Wall Street picks an institutional chain

Digital Asset, the creator of the Canton Network, announced on June 11 a $355 million round led by a16z crypto, with participation from ABN Amro, the Abu Dhabi Investment Authority, Apollo, BNP Paribas, Citadel Securities, CME Ventures, Coinbase Ventures, HSBC, S&P Global, Tradeweb, Optiver and more than 20 other institutions. The round closed above an earlier $300 million target at a ~$2 billion valuation, and follows roughly $185 million raised over the prior 12 months (a $135M DRW/Tradeweb round in June 2025 and a $50M BNY/Nasdaq/S&P/iCapital extension in December 2025). a16z itself contributed about $100 million.

Canton is a public, permissionless Layer 1 with configurable privacy built for institutional finance, using Digital Asset’s Daml smart-contract language to enable tokenized-asset workflows while keeping transaction data confidential and the network now has more than 700 ecosystem participants. The raise lands in a heating institutional-chain race. It comes just after a16z closed a $2.2 billion fifth crypto fund, and alongside peers Stripe/Paradigm’s Tempo (reportedly $500M at a $5B valuation) and Circle’s Arc ($222M at a $3B valuation). Not everyone’s convinced: TD analyst Lance Vitanza has called Canton “a glorified database in the cloud” that isn’t consistent with bitcoin’s open, trustless, permissionless architecture.

Strategy’s STRC “Stretch” preferred breaks its peg

STRC, one of Strategy’s preferred equity instruments, traded this week near $82, 18% lower than its $100 par value. In our equities data, STRC sat pinned at a flat $100 through mid-April, last closed at par in mid-May, and has slid steadily since - ending June 18 at $88.59.

STRC is the centerpiece of Strategy’s funding machine. When it trades at or above $100, Strategy issues fresh STRC through an at-the-market (ATM) program and uses the proceeds to buy bitcoin; below par that channel becomes uneconomic, and the company has paused it. So the depeg directly throttles one of Saylor’s main levers for accumulating BTC.

The slide is partly self-inflicted: to fund STRC distributions, Strategy sold 32 bitcoin for about $2.5M in late May. This marked its first BTC sale since it began accumulating in 2022, rattling a market unused to Strategy ever being a seller. STRC then broke below $95 on June 3, which under Strategy’s February dividend framework triggers a recommended rate hike of at least 0.5% (~$53M/year of added cost). The dividend has been held at 11.5% for four straight months but is now widely expected to rise to at least 11.75% at the June 30 reset, and STRC is shifting to twice-monthly payments (first record date June 30, first payment July 15) in a bid to “stabilize price, dampen cyclicality, drive liquidity, and grow demand” (Strategy).

Strategy held 846,842 BTC as of June 14 - it bought 1,587 BTC (~$100M) that week, funded by $209M of common-stock ATM sales rather than STRC - but its legacy software business generates only ~$477M of annual revenue against more than $1.2B in preferred-dividend obligations, a gap bridged almost entirely by capital markets. STRC isn’t alone: the sibling $100-par STRF also sits below par (~$91), and rival Strive’s SATA preferred slipped under par too, prompting a move to daily dividends. Bears (Peter Schiff) are circling; bulls like Jesse Myers frame it as a leverage-driven liquidation cascade rather than a Terra/Luna-style break, arguing the fundamentals are unchanged and a June 30 dividend hike stabilizes it. This is a sobering reminder that a 12%-yielding “par” instrument is not a stablecoin. A few options present themselves to Strategy: sell BTC, sell MSTR, or skip STRC dividend payments. All eyes remain on Bitcoin’s largest treasury vehicle through this period of market tumult.

Charts of the Week

• Prediction markets trading volume rockets past $10B driven by World Cup mania and the New York Knicks first NBA Championship win in 53 years

• Agentic payments have begun to inflect

• NEAR Fundamentals have held strong amid mediocre price performance

Other Notable News

• Zelle steps into the remittance game with its own stablecoin (American Banker)

• Charles Schwab is looking to break into the prediction market space (WSJ)

• The Ethereum Foundation lost another leader as co-executive director Hsiao-Wei Wang resigned, following co-executive director Tomasz Stanczak’s earlier exit (Coindesk)

Thanks for reading! Stay ahead this week by using the Artemis Terminal to pull the underlying data on any of the stories above (COIN segment revenue and pre-IPO perps volume, spot BTC ETF flows, HYPE fees and buybacks) or pull live numbers straight into your models with =ART() in Excel/Google Sheets.

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.

• Risk-off May: Five of seven factors were negative as BTC fell 6.5% and ETH fell 15.4%, reflecting broad crypto market weakness. Only Fundamental 1 (+8.7%) and Growth (+3.9%) held positive ground.

• Quality signals outperformed price signals: Both positive factors rely on on-chain activity metrics rather than price momentum. Momentum (-7.9%), SMB (-12.8%), and Stablecoin 1 (-10.3%), all price- or flow-based, underperformed meaningfully.

• XLM was the month’s breakout: Stellar Lumens surged ~52% in May, appearing as a top contributor across Growth, Fundamental 1, and as a painful short-side loss in Value and Momentum, making it the single most consequential asset across the factor suite.

• Large caps squeezed small-cap shorts: HYPE (+69.5%) and NEAR (+66.9%) surged, delivering SMB’s worst month of 2026. The factor’s short (large-cap) book bore the brunt, underscoring the concentration risk in thin crypto markets.

• Long-run edge remains intact: Fundamental 1 (1.63 Sharpe), Growth (1.54), and Momentum (1.29) continue to show strong risk-adjusted performance since inception, reinforcing the case for systematic factor exposure across crypto cycles.

Market Context

May 2026 proved a challenging month for systematic factor strategies, with five of seven factors posting negative returns amid a broad crypto drawdown. BTC fell 6.5% and ETH lost 15.4%, dragging market-cap-weighted exposure lower. The two standout performers were Fundamental 1 (+8.70%) and Growth (+3.89%), both of which isolate on-chain activity signals over price momentum, suggesting that assets with genuine protocol usage held up and in some cases rallied sharply versus the broader market. SMB was the worst performer (-12.76%): large-cap names like HYPE and NEAR surged unexpectedly, creating losses for the small-long/large-short construction. Momentum (-7.93%) and Value (-7.78%) gave back recent gains as the risk-off rotation compressed dispersion across the board. Stablecoin 1 (-10.25%) was hurt by adverse chain-flow reversals.

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

Performance Summary

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 (75.9%), ETH (12.2%), BNB (4.9%), XRP (4.3%), HYPE (0.8%), ZEC (0.4%), XLM (0.4%), CC (0.4%), ADA (0.4%), LINK (0.3%)

Performance Metrics

The Market Risk factor returned -6.53% in May, broadly in line with BTC’s decline. With BTC comprising ~76% of the portfolio, the strategy is effectively a leveraged bet on Bitcoin sentiment. ETH’s steeper -15.4% drop weighed on the remaining exposure, while small-weight holdings like HYPE (+69.5%) and ZEC (+59.8%) were too tiny to offset losses. The factor’s -30.85% YTD drawdown reflects the broader 2026 crypto bear phase, though the long-run annualized return of +10% since inception continues to reflect the market’s structural upward drift. Max drawdown of -52.4% since Jan 2022 is consistent with crypto’s historic peak-to-trough behavior.

Top 5 Winners & Losers for May

HYPE (+69.5%) and ZEC (+59.8%) were the portfolio’s top contributors in May, though their combined weight below 1% meant the impact on total return was minimal. BNB +11.5% at a 4.2% weight provided modest positive drag. The real story was the concentration risk: BTC (-6.5%) and ETH (-15.4%) together represent ~88% of the portfolio and both declined, while BCH fell 33.6%, an outsized move for a relatively illiquid holding. The winners/losers table highlights how, even in a down month, high-beta small-weight names can diverge dramatically from the overall portfolio trajectory.

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 delivered -12.76% in May, its worst month in 2026, as large-cap assets dramatically outperformed small caps. The factor shorts large caps and longs small caps, so HYPE (+69.5%), NEAR (+66.9%), and ZEC (+59.8%), all in the short (large-cap) leg, created severe losses that swamped gains on the long side. Structurally, this reflects a flight-to-quality dynamic common in bear markets: liquidity concentrates in the most well-known assets, punishing the small-cap long book. Despite May’s setback, the strategy’s annualized return of +8.7% and Sharpe of 0.47 since inception suggest the small-cap premium persists over time.

Top 5 Winners & Losers for May

INJ (+74.7%) and AKT (+61.1%) were the top contributors on the long (small-cap) side, reflecting strong token-specific catalysts. BCH shorting (+33.5%) also added value as that large-cap name sold off. The catastrophic losses came from the short book: HYPE (-69.5%) and NEAR (-66.9%) as shorts both surged, each carrying equal weight. The factor’s equal-weighting across 20+ names per leg means any outsized move in even one position has a material impact. The May episode underscores that SMB can experience large short-term reversals when market narratives sharply favor large-cap assets.

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 returned -7.78% in May as several high-value (low MC/fees) long positions underperformed while some low-value shorts rallied. The factor longs assets trading cheaply relative to fees generated and shorts expensive ones, a sensible long-run signal, but subject to momentum reversals. Notably, the YTD return is still positive at +5.72%, and the all-time cumulative return of +36.8% is the strongest of the price-based factors. The May drawdown reflects temporary mispricing as investors bid up high-MC/fees assets (e.g., XLM, TON) in anticipation of future fee growth rather than current fundamentals.

Top 5 Winners & Losers for May

HYPE (+69.5%), NEAR (+66.9%), and AKT (+61.1%) powered the long book, confirming that some low-MC/fees assets delivered exceptional returns. CHZ (+32.0%) and OP (+23.2%) shorts worked as expected. The main drag was the short book’s reversal: XLM (-51.8%) and TON (-43.6%) were shorted as expensive assets but rallied hard, a reminder that sentiment can override fundamentals in the short run. MON long (-39.1%) and INJ short (-37.2%) also hurt, the latter particularly frustrating given INJ’s strong momentum simultaneously benefiting the Momentum factor.

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 -7.93% in May as trend reversals hit both the long and short books. The strategy builds positions based on recent 3-week Sharpe-adjusted returns, making it vulnerable when markets abruptly change direction. INJ (+74.7%) was a standout long position that sustained its trend, but XLM reversed dramatically from the short book, surging 53% to become the factor’s biggest single detractor. At a YTD return of -12.23%, Momentum has given back gains from a strong 2025. Over the full period, however, it remains the second-best-performing factor at +50.2% annualized and Sharpe of 1.29.

Top 5 Winners & Losers for May

INJ’s +74.7% return as a sustained long position was the standout, carrying 5.8% average weight and driving meaningful positive contribution. HYPE (+57.6%) and NEAR (+54.4%) as short-lived momentum entries also added value. On the other side, XLM’s -53.0% surge from a short position was the single largest detractor: the asset had been flagged as a negative-momentum name but reversed sharply. ORCA (-26.3%), DEEP (-24.8%), and JUP (-24.1%) as long positions that faltered further compounded losses, suggesting momentum in mid-cap DeFi names broke down mid-month.

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 was one of only two positive factors in May, returning +3.89%. The strategy longs assets with accelerating fee and DAU growth and shorts decelerating ones, a signal that proved prescient in May’s selective environment. HYPE and XLM both featured as longs with genuine on-chain activity expansion, while the short book correctly identified slowing growth in ARB, BONK, and ZRO. The +2.35% YTD return and annualized +36.0% return with a Sortino of 3.16 make Growth one of the more risk-efficient factors in the suite.

Top 5 Winners & Losers for May

HYPE (+67.5%) and XLM (+66.4%) drove outsized gains on the long side, with both assets showing strong DAU and fee growth ahead of their price moves. ARB short (+35.4%), BONK short (+28.8%), and ZRO short (+26.6%) confirmed the factor’s identification of decelerating growth protocols. The losses came from JUP short (-40.9%), JUP’s activity growth re-accelerated unexpectedly, causing the short to lose significantly. ICP long (-27.4%), PENDLE (-25.9%), and STRK (-25.8%) also disappointed, with on-chain activity failing to translate into price outperformance in a broadly weak market.

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

Fundamental 1 was the top-performing factor in May at +8.70%, its best monthly return in 2026. The strategy’s composite signal, combining DAU growth, revenue quality, stability, and MC/fees mean-reversion, correctly identified assets with improving on-chain health. XLM’s positioning as a long based on MC/fees mean-reversion generated exceptional returns. The factor’s small universe (as few as 3 assets per leg in May) creates high concentration risk, which in this case worked in its favor. The annualized return of +54.4% and Sharpe of 1.63 remain the best in the factor suite since inception.

Top 5 Winners & Losers for May

XLM’s +73.6% return over two weeks as the top long position was the dominant driver; the asset scored highly on MC/fees mean-reversion as its ratio had compressed relative to its historical average. MON (+21.8%) also contributed positively as a one-week long. On the short side, NEAR (+11.8%), TAO (+10.0%), and ICP (+6.5%) all declined as expected. The main detractors were ADA long (-29.9%), where the signal misfired despite strong metrics, and SUI short (-25.4%) which rallied against the factor’s positioning. The small universe means each position carries 20% weight, amplifying both gains and losses.

Stablecoins 1

Construction

Equal-weighted long-short quintile portfolio: composite score combining rank z-scores of stablecoins flow. Long top quintile (~3-4 chains), short bottom quintile (~3-4 chains).

Performance Metrics

Stablecoin 1 returned -10.25% in May, its second consecutive negative month, bringing YTD losses to -11.96%. The factor tracks stablecoin inflows and outflows across chains, longing chains with net inflows (bullish signal) and shorting those with outflows. In May, the signal misfired as chains with expected outflows (SEI, APT) instead saw their native tokens rally. The annualized return of +12.6% and Sharpe of 0.37 reflect that this factor captures a real signal but with higher noise than the on-chain fundamental strategies. It also has the shortest track record in the suite at ~2 years.

Top 5 Winners & Losers for May

OP long (+32.7%) and MNT short (+23.7%) were the top contributors; both chains showed clear stablecoin flow dynamics that the signal correctly anticipated. TON short (+22.6%) and NEAR short (+11.8%) also contributed positively. The severe losses came from SEI short (-43.6%) and APT long (-41.9%): SEI experienced a native token rally despite stablecoin outflows, and APT’s stablecoin inflows failed to support its token price. ARB short (-22.7%) and SUI long (-13.3%) further compounded the negative return, suggesting stablecoin flow as a standalone signal had low predictive power for native token returns in May’s choppy environment.

Key Findings & Conclusion

May 2026 reinforced two durable lessons about systematic factor investing in crypto. First, on-chain activity metrics are more resilient than price-based signals in a risk-off environment. Growth and Fundamental 1 were the only two positive factors in May precisely because they measure what protocols are actually doing, not where their tokens have been trading. The remaining five factors, which rely on price, volume, or capital flows, all declined in a month when sentiment shifted faster than fundamentals. Second, single-name concentration risk is amplified in crypto’s thin market structure. XLM’s roughly 52% surge in May simultaneously drove the best returns in Fundamental 1 and Growth while punishing Value and Momentum shorts, illustrating how one asset can swing the outcome of multiple strategies at once. Looking across the full track record, the three highest-Sharpe factors (Fundamental 1 at 1.63, Growth at 1.54, and Momentum at 1.29) have all weathered periods like this before and continued compounding. The case for running a diversified multi-factor portfolio is exactly that no single month, and no single narrative, moves all seven strategies in the same direction.

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

Summary: We believe that most investors view Robinhood (HOOD) as a cyclical retail brokerage that trades with Bitcoin. This narrow view misses that Robinhood is becoming a financial super-app for Gen Z and Millennials, the demographic we anticipate being on the receiving end of one of the largest intergenerational wealth transfers in U.S. history. Robinhood Gold is, in our view, poised to become the Amazon Prime of retail finance, turning subscription revenue into a flywheel that can make Robinhood higher quality, more recurring, and less cyclical over time. Prediction markets are an emerging new retail asset class — and Robinhood already has more than 20x the monthly active users of the next-largest distribution-native competitors like Polymarket and Kalshi. We believe asset per customers will keep growing due to Robinhood’s strong product velocity and foothold with Gen Z and Millennials, a demographic we expect to benefit from an impending intergenerational wealth transfer, positioning Robinhood as a winner of retail finance.

Intro

This memo is a joint perspective from Artemis and North Island Ventures (NIV). Artemis is a digital finance research firm focused on blockchain and equity assets. NIV is an NYC-based investment firm, focused at the intersection of blockchain, fintech, and AI.

Our shared view: Robinhood is misunderstood as a highly correlated bet on crypto and investors are missing that Robinhood is becoming a financial super-app benefiting from a generational wealth transfer and that HOOD should benefit from 2 catalysts in 2026: prediction markets and the mega-cap IPOs of SpaceX and potentially Anthropic and OpenAI.

We believe that most investors still file Robinhood under “retail brokerage” and as an asset that heavily correlates with Bitcoin.

We want to demystify this assumption: Robinhood is finally becoming uncorrelated with Bitcoin with a correlation of 0.41 most recently after spending most of 2026 >0.70. The street prices the equity as if it is structurally tied to options and crypto trading cycles. We think that is wrong, and we think the mispricing closes as the super-app thesis becomes increasingly relevant.

1. The intergenerational wealth transfer and super-app. The median Robinhood customer is 35 years old, a demographic poised to benefit from Boomer populations’ wealth transfer to Gen Z and Millennials in the coming two decades, which we expect to be one of the largest intergenerational wealth transfers in U.S. history. While Boomer balances at legacy incumbents are more than 20x larger than Robinhood’s median customer’s average balance of roughly $12K, we view Robinhood’s relatively young customer base as positioning Robinhood to have little in terms of a legacy book to defend as compared to incumbents. The “why now” is the super-app. Robinhood finally has the product suite across investing, retirement, savings, spending, and speculating products to absorb the capital potentially coming from the intergenerational wealth transfer.

2. Prediction markets as a new retail asset class. Event contracts traded on Robinhood went from 300M in Q1’25 to 8.8B in Q1’26 — a 30× jump in four quarters. We estimate that prediction market revenue scaled from ~$3M to ~$104M over the same period. This is one of the fastest-growing product lines Robinhood has ever launched, with major catalysts still ahead: the World Cup, US mid-term elections, a full NFL season, and Rothera, Robinhood’s CFTC-regulated vertically integrated exchange and clearinghouse, going live.

We can already see a massive spike in prediction market volume on Kalshi and Polymarket that we expect will likely show up in Robinhood’s Q2 and Q3 ‘26 reported numbers as well.

Both tailwinds are early. Neither is reflected in the current forward multiple.

3. The mega IPOs could drive retail engagement. SpaceX is scheduled to IPO in early June and is planning to allocate a record 30% of shares to retail. Typical IPOs allocate 5-10% to retail. Anthropic and OpenAI may IPO in the coming months. The fundraising demand of these companies is unprecedented, and the retail interest appears strong. These IPOs could serve to increase engagement in Robinhood’s platform.

Why HOOD wins from the wealth transfer: Gold as the Amazon Prime flywheel

Gold subscribers grew from 1.14M at the end of 2022 to 4.34M in Q1’26 — 3.8× in roughly three years. The product costs $5/month (or $50/year). Our estimates put gross value to a member at ~$880/year (from higher cash yields, IRA match, free margin, cheaper options, Robinhood Gold Card, Morningstar research, Level II data, Robinhood Strategies).

What matters is the mix. Gold subscribers are ~16% of funded accounts but ~79% of assets on the platform. They are the asset compounders. Robinhood’s expanding product suite positions it well to convert more of their existing 27M funded accounts, into Gold subscribers. As Gold penetration rises, every metric that compounds — ARPU, net deposits, retention, gross margin — moves with it.

We believe Prime turned Amazon from a retailer into a consumer habit. The Prime subscription model increased engagement and ultimately spending on Amazon creating a stickier higher value customer. We believe Gold could do something similar for Robinhood and potentially make Robinhood the default financial home for the next generation.

We believe that the asset base is converging. Average AUC per funded customer grew from $2,700 at the end of 2022 to $11,900 at the end of 2025. Even at that pace, HOOD assets per customer remain ~5% of Schwab, Fidelity, and Interactive Brokers — all of whom carry far older books.

The margin profile shows the operating leverage has already arrived. Adjusted EBITDA margin went from -7% in 2022 to 29% in 2023, 48% in 2024, and 56% in 2025. Costs are largely fixed — technology, operations, and G&A together account for ~70% of the cost base — so incremental revenue lands at >70% incremental margins.

Why HOOD could win prediction markets

We believe this is the most underestimated part of the story.

In our view, three things make prediction markets a winner-take-most market: distribution, product UX, and integration with the user’s broader financial graph. We believe Robinhood could win on all three.

• Distribution. 27M funded accounts and ~13M MAUs. Kalshi has an estimated ~5M MAUs; Polymarket has ~0.5M. The acquisition gap to the next-largest distribution-native competitor is more than 20×.

• Product surface. Event contracts can sit alongside equities, options, and crypto in the same app, with the same wallet, statements, and risk system. Standalone prediction market apps cannot replicate that without becoming a brokerage.

• Vertical integration. Rothera, the company’s CFTC-regulated exchange and clearinghouse, lets Robinhood control take rate, add markets quickly, and avoid paying a third-party exchange. Even at flat monetization, vertical integration typically expands contribution by 25–30%.

We treat prediction markets as a new asset class — like spot crypto in 2017 or zero-commission equity trading in 2014. The pattern we see in both prior cases: the broker that owns retail distribution captures the category and re-rates.

How HOOD monetizes the super-app

The shape of Robinhood’s revenue is changing.

The transaction line is no longer the whole business. Net interest, Gold subscription fees, prediction markets, Robinhood’s high-yield checking, the Gold Card, retirement (TradePMR added ~$40B in advisory AUM in 2025), Robinhood Strategies, and an early estate-planning push are all independent monetization lines. Each touches a different lifecycle moment and stickifies the relationship. Together, we believe they are converting Robinhood from a transaction-revenue business levered to speculation into a recurring-revenue platform whose customers monetize across their financial lives.

Why the market is mispricing this

HOOD trades at ~34x 2027 consensus EPS, which is roughly in line with Interactive Brokers and Coinbase. In our view, three things suggest HOOD is mispriced:

1. Gold mix is still ~16% of accounts. As it climbs, ARPU is structurally positioned to converge toward the $621–$1,410 range that incumbents earn per account, against HOOD’s $167 today. That is not a forecast — that is the structural setup of the business.

2. Prediction markets are already one of the fastest-scaling product lines in U.S. retail finance, and Robinhood owns the distribution.

3. The cost base is largely fixed. Adjusted EBITDA margin already hit 56% in 2025 with the platform still in investment mode.

Coverage hasn’t caught up because Robinhood is between buckets. Brokerage analysts under-cover it. Fintech analysts treat it as crypto-exposed. Crypto analysts treat it as a brokerage. That is exactly the kind of pricing gap where we expect the asset to re-rate after the fact, not before.

Conclusion

The market is viewing Robinhood as a cyclical retail brokerage. The business it is becoming — a vertically integrated financial super-app for the demographic poised to benefit from the intergenerational wealth transfer from Boomers to Gen Z and Millennials, with a fast-growing new asset class of the cycle running through its app — is, in our view, a different kind of company.

We believe HOOD should be benchmarked against the platform comp set, not the brokerage comp set. The flywheel of (1) demographic ownership × (2) Gold-driven asset compounding × (3) prediction market category leadership compounds. Each new Gold sub raises ARPU. Each ARPU dollar raises the margin. Each new product locks the customer in further. In our view, the longer this compounds, the more durable the franchise becomes.

We don’t think this is a 2026 story. We think this is the beginning of a decade-long platform shift in how the next generation saves, trades, borrows, bets, and builds wealth.

Important Disclosures

This article and all of the information presented herein is being provided by Artemis Analytics Inc. and North Island Ventures, LLC (“NIV”) (each an “Author” and together, the “Authors”) solely for informational purposes to provide educational content and general market commentary.

No statements included herein relate to or contain any offer of NIV’s investment advisory services, nor does any content herein, including but not limited to any referenced companies, constitute an offer to sell or the solicitation of any offer to buy any security or shares or interest in any of NIV’s current or future private investment funds (each a “Fund”), which may only be made at the time a qualified offeree receives the applicable confidential private placement memorandum, subscription documents and governing documents (collectively, the “Operative Documents”).

The information herein is not intended to provide, and should not be relied upon for, accounting, legal or tax advice or investment recommendations. NIV is not acting, and does not purport to be or act as, an adviser or fiduciary, in any capacity, with respect to any prospective investor in any Fund. A recipient of this article should consult his or her tax, legal, accounting or other advisors about the matters discussed herein.

While the information presented in this article is believed to be accurate, neither Author nor their respective affiliates or representatives (i) makes any express or implied warranty as to the completeness or accuracy of the information presented herein, (ii) shall be liable for errors appearing in this article or (iii) has any duty to update the information set forth herein. Certain information contained herein constitutes “forward-looking statements,” which can be identified by the use of terms such as “may”, “will”, “should”, “expect”, “anticipate”, “project”, “estimate”, “intend”, “continue“, “target” or “believe” (or the negatives thereof) or other variations thereon or comparable terminology. Due to various risks and uncertainties, actual events or results or actual performance of investments may differ materially from those reflected or contemplated in such forward-looking statements. No representation or warranty is made as to such forward-looking statements.

Any analyses and opinions contained herein (including market commentary, statements or forecasts) may contain elements of subjectivity (including certain assumptions) or be based on incomplete or incorrect information. The Authors do not undertake to update the information contained herein.

This document displays certain trademarks and logos owned by the companies discussed herein. All such trademarks remain property of their respective owners, and are used only to directly identify or describe the companies being discussed herein. Their use is not intended to indicate and in no way indicates any sponsorship, endorsement or affiliation between the Authors or their affiliates, on the one hand, and the owners of such trademarks, on the other hand.

Accounts advised by NIV are currently invested in Robinhood and may make additional investments or exit positions at NIV’s discretion at any time for a variety of reasons and without further notice. No claim of performance (predecessor or otherwise) or prior track record is made in this document. Company metrics, including but not limited to, revenues, earnings, EBITDA, net income, and similar metrics, are not a proxy for the performance of a potential investment and should not be understood as such. There is no guarantee that achievement of any financial metrics, including changes in share price, will result in positive investment performance. Unforeseen or unaccounted for events could materially impact Robinhood’s performance and/or valuation—in such an event, projections could be materially altered. An investment’s performance is based on and must take into account a number of factors that the materials do not reflect, such as trading activity, terms of investment (e.g., share class), use of leverage, trading costs and other factors. Readers of this article should not interpret any presentation of share price as a proxy for any fund managed by NIV or investment performance. NIV makes no guarantee or assurance that any Fund will make similar investments. All investments are subject to the risk of total loss.

The discussions of Robinhood reference publicly-available information and certain estimates or projections have been provided by Robinhood or are primarily sourced from Robinhood. The Authors make no representation as to (i) the accuracy or reliability of any estimates or projections, (ii) the reasonableness of applicable assumptions and comprehensiveness of available inputs or (iii) the effectiveness or reliability of calculation methodologies. The Authors make no guarantee with respect to the prices at which any securities can be bought or sold. Numerous factors affect the performance of any investment.