Alibaba Group’s domestic commerce engine lost most of its profit last fiscal year, with the Taobao and Tmall Group’s adjusted EBITA dropping 84% as CEO Eddie Wu funneled billions into artificial intelligence infrastructure that has yet to pay for itself. The results, disclosed in Alibaba’s annual 20-F filing with the U.S. Securities and Exchange Commission for the fiscal year ended March 31, 2025, put hard numbers on a gamble the company has been telegraphing for more than a year: sacrifice today’s margins to build the AI platform it believes will define the next decade of Chinese tech.
The Cloud Intelligence Group, Alibaba’s vehicle for AI and cloud services, posted revenue growth of 38% for the fiscal year, according to the Associated Press. That is a striking acceleration for a division that had been growing in the low double digits just two years earlier. But the cost of chasing that growth swallowed nearly all of the profit Alibaba’s core retail business once generated reliably.
Where the money is going
Alibaba’s 20-F filing describes the company’s overarching strategy as “user first, AI-driven,” language that signals AI has moved from a research initiative to the organizing principle for the entire conglomerate. The filing details sharply higher depreciation and amortization charges tied to new data centers, GPU clusters for model training, and supporting infrastructure. Research and development spending also climbed as the company expanded its Qwen family of large language models and integrated AI tools across its commerce, logistics, and cloud platforms.
In early 2025, Alibaba publicly committed to spending more than $50 billion on AI and cloud infrastructure over three years, a figure that dwarfs its previous capital plans and puts it in the same spending tier as U.S. hyperscalers like Microsoft and Amazon. That commitment, reiterated during the company’s most recent earnings call, is the clearest evidence of what the headline means by “doubling down.” Wu and his leadership team are not trimming around the edges; they are redirecting cash flow from a profitable but slowing e-commerce business into a capital-intensive buildout with uncertain returns.
The 20-F also documents an operating loss in Alibaba’s domestic retail operations, a line item that would have been almost unthinkable a few years ago when Taobao and Tmall were printing money. Rising technology costs, subsidies to attract merchants to AI-powered tools, and heavier investment in logistics infrastructure all contributed. Management acknowledged in the filing that the multi-year spending plan will continue to weigh on near-term profitability.
Growth is real, but so is the burn rate
The 38% jump in cloud and AI revenue is not a vanity metric. Alibaba’s Cloud Intelligence Group has been winning enterprise contracts for its Qwen-based AI services and expanding its public cloud footprint across Asia. The AP’s reporting, which drew on the company’s earnings call, noted that management framed the spending as essential to long-term competitiveness, a message aimed squarely at shareholders watching the profit line crater.
Still, 38% revenue growth looks very different depending on the denominator. Cloud and AI remain a fraction of Alibaba’s total revenue, which is still dominated by e-commerce transaction fees and advertising. Growing a smaller business fast is not the same as replacing the profits of a larger one, and the gap between what AI is earning and what it is costing remains wide. The company is effectively asking investors to fund a transition whose payoff timeline it has not publicly specified.
The competitive pressure Wu cannot ignore
Alibaba is not spending in a vacuum. Tencent, Baidu, ByteDance, and Huawei are all racing to build competing AI platforms, and each has disclosed significant infrastructure investments of its own. Baidu has staked its future on its Ernie family of models. ByteDance is deploying AI across its global content empire. Tencent is weaving large language models into WeChat and its enterprise cloud offerings. None of these rivals have published apples-to-apples spending comparisons, but the direction is clear: China’s biggest tech companies are all burning capital on AI at a pace that makes restraint look like surrender.
Alibaba’s 20-F lists competitive intensity as a material risk factor but does not quantify how its spending stacks up against peers. That omission makes it difficult for outside analysts to judge whether Wu is overspending or simply matching the table stakes of an industry-wide arms race. What is visible is that Alibaba’s cloud division, once the clear market leader in China, has seen its share erode as rivals closed the gap. The current investment blitz is partly defensive: an attempt to reclaim technological leadership before the window closes.
Regulatory and geopolitical wildcards
Two external forces could reshape the math on Alibaba’s AI bet. Inside China, evolving AI governance rules, including interim measures on generative AI services and data security requirements, add compliance costs and could limit how quickly new products reach the market. The 20-F flags policy uncertainty as a headwind but does not estimate its financial impact.
Outside China, U.S. export controls on advanced semiconductors continue to restrict Alibaba’s access to the most powerful AI chips. The company has been stockpiling Nvidia GPUs where possible and investing in domestic chip alternatives, but neither workaround fully replaces unrestricted access to cutting-edge hardware. If controls tighten further, the cost of training and running large models could rise, stretching the investment timeline even longer.
What investors still cannot see
For all the detail in the 20-F, several critical questions remain unanswered. Alibaba does not publish project-level return-on-investment data, so there is no way to tell which AI initiatives are closest to profitability and which are speculative bets. The filing includes no specific revenue or margin targets for the Cloud Intelligence Group in coming fiscal years, only general language about pursuing growth and improving efficiency. And the exact words Wu and other executives used on the earnings call are available only through paraphrased press accounts; Alibaba has not released a full public transcript as of late May 2026.
Analysts have offered their own estimates for when AI spending might inflect toward profitability, but those projections rest on assumptions about customer adoption rates, pricing power, and regulatory stability that are not grounded in company guidance. Anyone relying on a specific breakeven date should treat it as a guess, not a forecast.
The bet Alibaba is forcing shareholders to make
Alibaba’s fiscal 2025 results are not a story of failure. A company that grows a major business line by 38% in a single year is clearly gaining traction. But they are a story of extraordinary cost. An 84% decline in core commerce profit is the price tag for a strategic pivot that Wu has chosen to execute at full speed rather than in cautious increments.
For shareholders, the practical next step is straightforward: read the 20-F directly rather than relying on headline summaries. The filing’s segment reporting, capital allocation disclosures, and risk factors paint a far more detailed picture than any single earnings story can convey. Pay close attention to how much of Alibaba’s free cash flow is being redirected from legacy businesses into infrastructure that has not yet proven self-sustaining, and compare that burn rate against the cloud division’s revenue trajectory quarter by quarter.
The broader signal from these results is that China’s AI race has entered the phase where ambition meets arithmetic. Alibaba is betting that the pain is temporary and the platform it is building will become indispensable. Based on the evidence available through late May 2026, that outcome is plausible but far from guaranteed, and the bills are arriving faster than the revenue.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
