Alibaba has entered the harvest period, and there are no major ripples in the food delivery war.
Text | Peng Qian
Editor | Qiao Qian
After Alibaba's massive investment of hundreds of billions, the returns are being realized one by one.
The financial report shows that Alibaba's revenue in the fourth fiscal quarter of 2026 (i.e., Q1 of the natural year) was 243.38 billion yuan, a year-on-year increase of 3%. Excluding the revenue of sold - off businesses such as Sun Art Retail and Intime, the revenue on a comparable basis increased by 11% year - on - year.
The double - digit year - on - year revenue growth was mainly driven by the cloud business (including AI cloud and traditional cloud). In Q1, Alibaba Cloud still performed brilliantly, maintaining a high - growth trend. The overall revenue increased by 38% year - on - year, in line with market expectations. Among them, the revenue from external customers increased by more than 40% year - on - year, accelerating from the 35% year - on - year growth in the previous quarter. The revenue from AI - related products has achieved three - digit growth for 11 consecutive quarters, and the continuous growth of the public cloud business has jointly promoted the high growth of the cloud business. The contribution of the e - commerce business to revenue was relatively stable, with domestic and overseas e - commerce both growing by 6% year - on - year.
As Alibaba is still in a high - investment transformation period, several of its profit indicators performed poorly: the operating profit was in a loss state; the adjusted EBITA decreased by 84% year - on - year, only reaching 5.2 billion yuan, compared with 32.6 billion yuan in the same period last year; the non - standard net profit also decreased by 100% year - on - year. However, thanks to the soaring stock prices of several AI companies such as Minimax and Zhipu that Alibaba invested in, Alibaba received substantial returns. The net investment income in fiscal year 2026 reached 87 billion yuan, a year - on - year increase of 320%, the highest since 2022, reversing the previous situation where most investments suffered losses due to the sluggish stock market.
Due to the less - than - ideal profit performance, Alibaba's U.S. stocks fell nearly 3% after the release of the financial report.
Since this year, with the advent of the Agent era, the explosion of computing power demand has triggered a surge in global AI - themed stocks. In the past two days, with the positive factor of Trump's visit to China with executives of several U.S. AI giants, the stock prices of Minimax and Zhipu continued to soar. The former once rose more than 30% in a single day, and the latter even soared 35% during trading. However, the stock prices of Alibaba and Tencent, which have made huge investments in AI, have not been so outstanding. Since the beginning of this year, the former has fallen nearly 7% in the Hong Kong stock market, and the latter has even dropped 23% - because the market no longer just rewards the AI ambitions of technology giants but wants to see real revenue - generating paths.
To stabilize market sentiment, after setting the annual revenue target of "AI and cloud (traditional cloud + AI cloud) to reach $100 billion in the next five years", Wu Yongming gave a clearer answer this time. He said at the financial report meeting that based on the soaring global demand for MaaS services, the ARR (i.e., the annualized revenue of AI models and application services) of the AI cloud business, namely "Bailian", will exceed 10 billion yuan by June this year, and will exceed 30 billion yuan by the end of this year; in addition, it is expected that in the next year, the proportion of Alibaba's AI - related revenue will exceed 50%, becoming the main engine for the revenue growth of Alibaba Cloud. Alibaba Cloud is now a business with an annual revenue of over 100 billion yuan, and a 50% proportion means an incremental revenue of tens of billions.
After significant profit losses due to the huge entry fee for transformation, Alibaba's management finally reassured the market, which led to a more than 8% surge in Alibaba's U.S. stocks in after - hours trading after the financial report was released.
The commercialization progress of the AI business is becoming the key factor controlling Alibaba's stock price.
Desperate to Generate Revenue, Alibaba's AI Also Wants to "Support the Family"
The market is taking a more rational approach to the increasingly heated AI competition, starting a new round of pricing assessment to examine whether large companies can realize returns after making huge investments in AI.
The top players across the ocean were the first to present their answers. Recently, the "Magnificent Seven" (referring to several U.S. technology companies that have performed the best in the AI wave, including NVIDIA, Microsoft, Amazon, Google's parent company Alphabet, Meta, Apple, and Tesla) released their financial reports. Their revenues all increased significantly, and capital expenditures continued to rise, but their stock prices were polar opposite. The key variable controlling the stock prices was the results of AI commercialization. Alphabet and Amazon were the top students in this assessment, and their stock prices rose accordingly.
As one of the domestic large companies that proposed to go all - in on AI earlier, after two years of continuous high - investment, Alibaba has also been talking more about AI commercialization revenue in the past two quarters, rather than simply emphasizing investment. At the beginning of the financial report meeting, Wu Yongming for the first time disclosed the specific data of AI revenue: the quarterly revenue of Alibaba Cloud's AI - related products was 8.971 billion yuan, and the annualized revenue exceeded 35.8 billion yuan, accounting for 30% of Alibaba Cloud's revenue. Three quarters ago, this figure was 20%.
With the overall expenditure of 380 billion yuan remaining unchanged for three years, compared with Tencent and ByteDance, Alibaba's current capital expenditure level is more stable. In Q1, it was 26.8 billion yuan, basically the same as 24.8 billion yuan in the same period last year. Tencent said it would double its capital expenditure this year. Its capital expenditure in Q1 was 32 billion yuan, a significant increase from 19.6 billion yuan in the same period last year. ByteDance also recently raised its capital expenditure to 200 billion yuan, more than doubling that of last year.
This gap is mainly due to the different stages and development paths of these companies. Compared with Google, Alibaba has taken the lead in achieving full - stack investment and layout. Tencent significantly increased its AI investment, which obviously has a remedial meaning. At the general meeting of shareholders held on the day when Tencent released its Q1 financial report, Ma Huateng admitted that Tencent's basic capabilities in the AI field were not outstanding in the early stage. He said bluntly: "A year ago, we thought we were on the boat, but later we found the boat was leaking. Now we feel we are on it, but we can't sit still. We still hope the boat can go faster." Tencent is increasing investment in large models and WeChat Agents, embedding AI functions in the existing WeChat ecosystem, rather than building a fully integrated platform from scratch like Alibaba.
ByteDance's capital expenditure is significantly higher than that of other manufacturers. This is not only because ByteDance is also moving towards a full - stack layout but also because the future high - speed growth of Volcengine will highly depend on ByteDance's investment in video models, which requires much more computing power than large - language models, and the cost naturally increases. According to the South China Morning Post, the 200 - billion - yuan cost of ByteDance mainly consists of two parts: 85 billion yuan for AI chip procurement and R & D, and 115 billion yuan for building computing power centers.
The huge capital expenditures of these giants must ultimately have corresponding commercial returns. However, objectively speaking, compared with their U.S. counterparts, the monetization efficiency of the entire domestic AI industry at present cannot support long - term investment at the scale of hundreds of billions. Although the gross profit margin of AI cloud (60 - 80%) is much higher than that of traditional cloud (15% - 20%), due to the ongoing price war in the domestic AI cloud business in the short term, and the V4 model released by the leading player DeepSeek even offered a 75% limited - time discount, manufacturers have to face the reality that the net profit margin is difficult to improve significantly for the time being. The goal of increasing the net profit margin of the cloud business to 20% (doubling) mentioned by Wu Yongming still requires time.
Wu Yongming also said bluntly at the financial report meeting that Alibaba's AI business is entering the cycle of commercial returns, and increasing AI revenue is Alibaba's current top priority. Due to various real - world factors in the industry, the profit margin can only be ranked second. The AI revenue of Alibaba Cloud mainly comes from self - developed models, mostly from API services provided based on the Qwen model and MaaS, and a small part comes from the subscription revenue of AI - native software such as DingTalk.
If the traditional cloud computing business is the foundation of Alibaba Cloud, the AI cloud is the battlefield that Alibaba must win. This determines the upper limit of Alibaba Cloud's future development and whether Wu Yongming's goal of "AI and cloud annual revenue (AI cloud plus traditional cloud) reaching $100 billion" can be successfully achieved. Citigroup also predicted this week that in the next five years, the revenue of Alibaba's AI and cloud will soar, with a compound annual growth rate of 90%, and the growth rate of MaaS will exceed 200%. This optimistic prediction is based on a new business flywheel: some merchants on the business side use Wukong, which helps merchants improve operational efficiency. All Taotian merchants are potential users of Wukong, bringing considerable revenue to MaaS.
Wu Yongming said that Alibaba Cloud is shifting from traditional cloud computing to model computing power and Agent services. Since the advent of the AGI era, the traditional "selling storage, servers, and broadband" is giving way to "selling model calls, computing power, and intelligent services", which means that Alibaba Cloud needs to break its past business inertia. Alibaba Cloud has dominated the traditional cloud market for many years and has also faced strong opponents such as Huawei Cloud head - on. However, it still faces considerable pressure when facing ByteDance, a fierce competitor in the AI cloud field.
In the AI - to - B field, the two sides are launching a new round of competition in the AI cloud field. With the main goal of selling as many Tokens as possible, Alibaba is making Bailian the core carrier of its Maas business. According to 36Kr, Bailian's revenue was still limited last year, and Wu Yongming's prediction of 3 billion yuan is a huge leap. Bailian currently integrates more than 200 mainstream third - party large models, including Qianwen, DeepSeek, Kimi, and Minimax. More than one million enterprises and individuals are currently using it, creating more than 800,000 AI agents. In Q1, the number of Bailian's customers also increased eight - fold.
ByteDance is also sprinting on the AI cloud path: on the one hand, it helps Volcengine generate revenue by promoting video models such as Seedance and has set a more aggressive goal. According to 36Kr, Volcengine under ByteDance recently raised its MaaS revenue target for this year from 10 billion yuan at the beginning of the year to 15 billion yuan; on the other hand, ByteDance's AI also needs to find more commercialization paths. Recently, Doubao launched a charging plan for professional users (between the B - end and C - end), while direct commercialization for the C - end is still relatively difficult at present.
To compete for a larger market share in the fierce market competition, Alibaba needs to continuously develop different types of models for different fields, such as the coding direction, multi - modal video generation direction, and future - oriented world models. This still requires continuous investment in talent introduction and technology. A source close to Alibaba said: "Model iteration is too fast. Maybe the 'top - stream' model will change in half a year. Only by continuously developing new models can we possibly catch new trends and hotspots."
36Kr learned that to sell more Tokens, multi - modality will also be one of the key development directions for Alibaba in the model field this year. Recently, Alibaba spent a sum of money to acquire a research institute specializing in multi - modal technology. The newly launched Happy Horse is the first AI video - generation application quickly launched by the Future Innovation Laboratory after integrating into Alibaba's ATH Business Group. Currently, it is trying to attract users from Seedance and Keling by leveraging advantages such as cost - effectiveness (consuming fewer Tokens to generate videos of the same length) and more efficient video - generation time (having a later commercialization progress and giving priority to ordinary users rather than video companies).
In addition to generating revenue, Alibaba is also trying to improve the cost structure of its AI business. The shortage of computing power is a common problem faced by large companies. The large - scale mass production of self - developed chips is an important way to optimize costs.
As a reference, the story of "replacing cost with self - development" has been successfully staged in the United States. The annualized revenue scale of Amazon's self - developed AI chips (Trainium and Inferentia) business has exceeded $20 billion and is growing at a three - digit rate. This directly proves that Amazon is trying to reduce its dependence on NVIDIA. At present, Alibaba's T-Head has also achieved large - scale mass production and wide - scale industrial application of self - developed GPU chips. More than 60% of the computing power serves external commercial customers, covering core scenarios such as the Internet, finance, and autonomous driving.
According to information obtained by 36Kr from institutional sources, Alibaba's capital expenditure in 2026 will rise to 150 - 170 billion yuan, which largely depends on how many H200 chips it can obtain, determined by the current situation. About half of the cost is for chip procurement, among which the quota for domestic chips is about 30 billion yuan, and most of the budget will be used for Alibaba's T - Head's Zhenwu 810E.
36Kr also learned that the rumored financing negotiation between DeepSeek and Alibaba actually took place last year. DeepSeek was the initiator, proposing to access scenarios such as Alipay. Alibaba also needed DeepSeek's infrastructure capabilities, but the negotiation between the two sides did not continue due to various reasons; the opportunity to restart cooperation this year was that T - Head and DeepSeek discussed matters related to chip procurement.
Wu Yongming said bluntly at the financial report meeting: "No card is idle." Almost all the AI computing power chips in Alibaba's servers are in a state of high - efficiency utilization, with no waste of resources, formally refuting the "theory of excessive computing power" and emphasizing that Alibaba's AI business has entered the stage of large - scale returns.
Taobao Flash Sale Needs to Spend Money Wisely, and E - commerce Fights with Qianwen Ecosystem
After a year of fierce competition, Taobao Flash Sale has new missions and goals.
36Kr learned that Taobao Flash Sale's goal in the new fiscal year is to achieve a positive unit economics (UE) while maintaining its market share. The executives also made corresponding commitments at the financial report meeting, which means that the stage of expanding the market by burning money in instant retail has ended.
Due to the fact that excessive investment in both offline and online channels has led to profit losses, the e - commerce business really needs to spend money more carefully to ensure that there is support for Alibaba's two - front battlefields. In Q1, the adjusted EBITA of Alibaba's Chinese e - commerce business decreased by 40% year - on - year, with a loss of more than 20 billion yuan. If the loss of the instant retail business is excluded, the EBITA actually remained stable year - on - year. In addition, the cash flow in Q1 turned negative again, with a net outflow of 17.3 billion yuan due to huge investments.
The CMR growth of Taotian has returned to normal. Under the new calculation method (the e - commerce business launched a "marketing plan". When merchants reach the GMV target, the platform will return part of the marketing fees. Previously, merchants' advertising and commissions were included in CMR, and marketing fees were included in market expenditure), the CMR of the Chinese e - commerce business in Q1 was 1%, but in a comparable calculation, the year - on - year growth was actually 8%, exceeding market expectations. This is mainly due to the lag in Spring Festival stockpiling and the continuous growth of Flash Sale business revenue.
To achieve a positive UE for Flash Sale, it is necessary to continuously reduce costs and increase merchants' revenue.
Currently, the user subsidies and rider subsidies for Taobao Flash Sale have been reasonably adjusted. After a year of user accumulation, Alibaba's subsidy focus has shifted to high - spending users, and rider subsidies have also been adjusted according to the order - volume business layout. Compared with last year, more Flash Sale riders will support retail businesses such as 4 - hour delivery and half - day delivery.
Some leading merchants now have a strong willingness to pay because the merchants who took the lead in trying have achieved results. Take Decathlon as an example. Since it launched the brand - direct - sales + in - store self - pickup model on Flash Sale, the average daily order volume has exceeded 5,000, and it can reach 20,000 orders at the peak of major promotions. The single - store fulfillment radius has expanded from 3 kilometers to 10 kilometers, and the sales volume has exceeded 100 million yuan.
This is inseparable from Flash Sale's promotion of the retail business. Since the end of last year, Flash Sale has continuously expanded the territory of its retail business, increasing investment in formats such