Alibaba, my dear alibaba…
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:
Global AI usage continues to compound and cheaper models continue to take share
Alibaba owns almost the full AI stack
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.