By
Edited by Wendy Wen
On May 14th, Alibaba released its first quantifiable AI valuation business report.
The report showed that Alibaba’s annual revenue exceeded 1 trillion yuan for the first time in the 2026 fiscal year, reaching 1.02367 trillion yuan. However, the net profit for the same period decreased by 19% to 102.1 billion yuan. As of March 31st, the quarterly free cash flow turned negative, with a net outflow of 17.3 billion yuan, compared to a net inflow of 37.43 billion yuan in the same period last year.
Due to the pressure on profits and cash flow, Alibaba’s US stock price fell before the market opened but rebounded 8% during the earnings call. Alibaba shared its AI MaaS and application annual recurring revenue (ARR) scale and targets for the first time, indicating that the capital market has initially recognized the value of Alibaba’s AI narrative, but there are still concerns.
The report showed that Alibaba’s AI-related products accounted for more than 30% of Alibaba Cloud’s external revenue for the first time, reaching 89.71 billion yuan, with annualized revenue exceeding 358 billion yuan. Alibaba Group CEO Wu Xunming repeatedly emphasized during the earnings call that AI has crossed the initial investment stage and entered the commercialization return period.
This judgment is supported by the rapid growth of AI MaaS (Model as a Service) business. Wu Xunming revealed that the ARR of AI model and application services, including the Hundred Refining MaaS platform, will break through 10 billion yuan in the June quarter and reach 30 billion yuan by the end of the year. Notably, the API Token demand data from May to June this year grew more than 10 times compared to November to December last year.
Alibaba’s AI MaaS business revenue is mainly composed of two parts: API service revenue from the Hundred Refining MaaS platform and subscription revenue from AI native software. Currently, most of the revenue comes from the former, which is the Token fee paid by enterprise customers for calling models. Commercialization mainly relies on the To B (enterprise) route, with domestic and foreign markets being basically the same. Why are enterprise customers willing to pay for it?
Wu Xunming mentioned during the earnings call that as AI transforms from conversational chat tools to intelligent body mode, helping customers complete complex reasoning tasks, customer demand continues to soar, even with the price increase of Tokens, the acceptance rate remains high, and it’s even in short supply. “Based on current demand, our supply cannot fully meet the demand, and there are still many customers in line.”
Alibaba has made a clear distinction between its investment pace in AI To B and AI To C. Enterprise customers have a stronger willingness to pay, and most reasoning resources are invested in the commercialization field. C-end business, including user acquisition of the Qianwen App, still requires a certain investment cycle.
Different from the past problem of low SaaS payment willingness of Chinese enterprises, Alibaba believes that this situation will change in the era of large models. Both Chinese and American enterprises are willing to pay for AI as long as the value of Tokens exceeds the cost, and enterprise demand is endless. This is also the core source of Alibaba’s continued large-scale investment in AI data centers.
On the same day, other AI players also released signals.
Tencent Holdings’ financial report included AI-related revenue in the cloud service segment for the first time, but did not separately disclose the AI revenue scale. Tencent mentioned that “AI infrastructure investment continues to increase,” but the commercialization pace is relatively cautious.
ByteDance’s commercialization pace is faster, and it has also significantly increased its AI capital expenditures, raising its 2026 AI capital expenditure plan to over 200 billion yuan. Everyone has reached a consensus that “this money must be spent.” It is understood that in the hot market of car companies, the competition between Huawei Cloud and Alibaba Cloud is intense.
Although the AI narrative has received a market response, there are several risk points that cannot be ignored. First, competition, Alibaba still needs to maintain the competitiveness of its models, and respond to potential changes in profit margins brought about by future model price reductions. Whether the revenue growth curve can meet expectations, the long-term return on investment still needs to be observed.
In terms of profit margins, Wu Xunming emphasized the value of self-researched chips. Currently, the deployment rate of Pingtouge self-researched chips in Alibaba Cloud is low. Except for TPU, CPU, storage network chips are all self-researched, and it is expected to achieve large-scale deployment of full-stack self-researched chips in the future. Foreign mainstream AI chips have a gross margin of 60-80%, so Alibaba believes that even if domestic chips have some room for improvement in performance and power consumption, there is still a huge space for cost-effectiveness optimization.
The second is capital expenditure. Domestic and foreign giants are all heavily investing in AI, and under the current pressure on free cash flow, Alibaba’s AI infrastructure investment will far exceed 380 billion yuan in the next two years. During the earnings call, Wu Xunming compared AI to the manufacturing industry, saying that to obtain more revenue, it is necessary to build two factories for AI training and reasoning, which requires a large amount of data center construction.
Although he believes that the return on investment is very certain, this indeed means that for a long time, Alibaba will still need to find a balance between strong capital expenditures and healthy cash flow.
In the instant retail battlefield, which was previously a money-burning war, Alibaba has also given a profit timeline. Alibaba China E-commerce Business Group CEO Jiang Fan revealed that in the first quarter, the overall order scale of instant retail reached 2.7 times that of the same period last year, with non-food retail orders growing three times.
After April, Alibaba maintained the order scale while improving the unit economic efficiency through logistics and delivery efficiency and optimizing the order structure, especially the further increase in the average order value. Management expressed confidence in achieving a positive unit economic efficiency (UE) for this business by the end of the 2027 fiscal year.
Of course, the overall e-commerce market still faces pressure. Alibaba’s e-commerce business revenue in the quarter was 96.292 billion yuan, down 1% year-over-year; during the same period, customer management revenue grew 1% year-over-year. If the impact of the new marketing development plan is excluded, customer management revenue grew 8% year-over-year.
It seems that the market has given Alibaba’s AI narrative logic an initial vote of confidence. But the real test is yet to come.
