Sharp Take | Alibaba's "Ice and Fire": E-commerce Stalling Under AI's Hundred-Billion Blueprint

Guancha.com Despite Alibaba Cloud’s rapid 36% growth and its ambitious goal of surpassing $100 billion in revenue within five years, the cloud giant remains overshadowed by underwhelming performance.

On March 19, Alibaba announced its financial results for Q3 FY2026 (ending December 31, 2025), reporting revenue of 284.843 billion RMB, up 2% year-over-year. Excluding the disposed businesses of Gaoxin Retail and Intime, the same-store revenue grew 9% YoY. Non-GAAP net profit was 16.71 billion RMB, down 67% YoY.

Market expectations had forecasted Alibaba’s Q3 revenue at 290.14 billion RMB, a 3.6% increase, and adjusted net profit at 31.04 billion RMB, a 39.2% decline.

Both revenue and net profit fell short of market expectations.

The earnings report showed Alibaba’s net cash flow from operating activities for 2025 declined 49% YoY to 36.032 billion RMB. Free cash flow plummeted from 39.02 billion RMB last year to 11.346 billion RMB, a 71% drop.

Following the release, Alibaba’s U.S. stock opened down more than 9.31%, reflecting market sentiment.

However, nearly all institutional forecasts highly recognize Alibaba’s AI strategy: Morgan Stanley upgraded Alibaba to industry top pick; China International Capital Corporation maintained an outperform rating with a target price of HKD 197, praising the newly established Alibaba Token Hub (ATH) business group for accelerating the commercialization of models; First Eagle Investment pointed out that the market’s valuation ignores Alibaba’s highly promising AI business potential.

Notably, JPMorgan recently stated that the intensive departure of core AI talent in March (including Lin Junyang, former head of Qianwen, and Yu Bowen, former head of post-training for Qianwen) raises concerns about the future development and sustainability of the Tongyi Qianwen large model, increasing short-term execution risks.

In contrast to AI’s rapid progress, the financial report shows that as of the end of December 2025, Alibaba’s Customer Management Revenue (CMR, including advertising and commissions on e-commerce platforms) was 102.664 billion RMB, a mere 1% increase YoY. In the same quarter, Alibaba’s China e-commerce segment’s adjusted EBITA was 34.613 billion RMB, down 43% YoY.

On one side, AI-driven cloud business is booming; on the other, the e-commerce engine is clearly slowing down.

AI “Advances”

During the earnings call, Alibaba Group CEO Daniel Zhang stated that with the advent of the AI agent era in 2026, Alibaba has built a complete full-stack AI capability, from chips and cloud computing to models and applications.

In the quarter, Alibaba Cloud’s revenue growth reached 36%, surpassing market expectations, with “AI-related product revenue achieving triple-digit YoY growth for the tenth consecutive quarter.”

Zhang outlined specific goals: “In the next five years, the commercialization revenue from cloud and AI, including MaaS (Model-as-a-Service), will exceed $100 billion.”

He further noted that Alibaba Cloud’s external commercialization revenue as of the end of February 2026 has officially surpassed 100 billion RMB.

Of particular interest, this earnings call revealed key data about Pingtouge for the first time at scale. As of February 2026, Pingtouge had delivered 470,000 chips, over 60% of which serve external commercial clients across internet, finance, autonomous driving, and more than 400 enterprise customers.

Zhang explained that Pingtouge’s products cover the entire AI workflow from training and fine-tuning to inference, and are deeply integrated into Alibaba Cloud’s training and inference scenarios.

He believes Pingtouge’s value lies not only in cost optimization but also in providing critical “supply assurance” amid global computing power shortages. He also mentioned that an IPO for Pingtouge is not ruled out in the future, though no specific timeline has been set.

On the model and application layer, the newly formed Alibaba Token Hub (ATH) business group is the core organization for strategic implementation. Its goal is to “create tokens, deliver tokens, and apply tokens,” integrating Tongyi Laboratory, MaaS business line, Qianwen division, Wukong division, and AI Innovation division.

Zhang explained that this organizational design is based on the fact that AI has entered an agent-driven era, where close integration of models and applications is crucial.

Combining model development with application scenarios aims to gather tight data feedback from customer use cases, continuously enhancing model capabilities.

“The Qwen 3.5-Plus model performs excellently in inference, programming, and agent benchmarks,” Zhang revealed. A next-generation model optimized for coding and agentic scenarios will be launched soon.

Of note, Alibaba’s enterprise-oriented MaaS platform, BaiLian, saw a sixfold increase in token consumption in the public model service market over the past three months. Zhang expects “commercial MaaS revenue to become Alibaba Cloud’s largest revenue product.”

Meanwhile, Alibaba launched Wukong, the world’s first enterprise AI-native agent platform, compatible with different enterprise data permissions and management workflows, integrating Alibaba’s entire B2B ecosystem as “skills,” aiming to become Alibaba’s unified enterprise AI capability interface.

“Slowing” E-commerce

While AI business advances rapidly, Alibaba’s core e-commerce business is slowing.

In the quarter, Alibaba’s China e-commerce revenue was 159.347 billion RMB, up 6% YoY, but core Customer Management Revenue (CMR) grew only 1% to 102.664 billion RMB.

Alibaba management explained that the slowdown was mainly due to consumption trends, a warm winter, and a later Spring Festival this year. Additionally, extended promotional periods and increased consumer rights investments contributed. The China e-commerce segment’s adjusted EBITA fell 43% YoY to 34.613 billion RMB.

Management pointed out that this was mainly due to investments in instant retail, user experience, and technology.

To counter competitors like Meituan in local life services and to build new consumption scenarios, Alibaba is investing unprecedentedly in instant retail.

This quarter, revenue from instant retail, including “Taobao Flash Sale” (formerly Ele.me), grew 56% YoY to 20.842 billion RMB, with scale continuing to expand.

Alibaba E-commerce CEO Jiang Fan stated that, alongside scale expansion, improvements in fulfillment logistics efficiency, order structure, and high customer retention have led to better unit economics and average order value (AOV). The synergy between instant retail and the overall e-commerce market has driven double-digit YoY growth in Taobao’s monthly active buyers.

However, this is a costly “blood transfusion”: to secure future consumption access, Alibaba is sacrificing current profitability.

Alibaba management said free cash flow dropped from 39.02 billion RMB to 11.346 billion RMB, mainly due to investments in instant retail.

Jiang Fan provided a clear profitability timeline for instant retail, maintaining a target of over 1 trillion RMB in total transaction volume by FY2028, with profitability expected to be achieved after reaching that scale, likely by FY2029.

This suggests that heavy investments in instant retail will continue to weigh on overall profits for at least the next two to three years.

Cost structure changes further confirm the investment pressure. Sales and marketing expenses soared from 42.675 billion RMB last year to 71.934 billion RMB, accounting for 25.3% of revenue, up from 15.2%, mainly due to investments in Alibaba China e-commerce user experience.

This indicates that, beyond the visible battlefield of instant retail, Alibaba is also increasing hidden investments in user retention and platform stickiness.

Meanwhile, share-based compensation expenses rose from 3.865 billion RMB to 4.859 billion RMB. Alibaba management explained this was mainly due to the replacement of incentives related to Ele.me, aimed at talent retention.

Notably, the number of 88VIP members continued to grow double-digit YoY this quarter, exceeding 59 million.

Alibaba International Digital Commerce Group (AIDC) saw a significant narrowing of losses, driven by logistics optimization and efficiency improvements, with unit economics of AliExpress Choice also improving sequentially.

Future Outlook

On one hand, the AI business has outlined a billion-dollar blueprint; on the other, the core e-commerce business is paying real costs for strategic transformation. Alibaba stands at a critical crossroads.

Zhang expressed optimism about the technological trend during the AI strategy discussion, believing that Alibaba is ready to support exponential growth in AI, from infrastructure to applications.

In response to analyst questions, Zhang further explained the design logic of TokenHub: integrating models, applications, and MaaS to create a virtuous data flywheel of “more scenarios – more data – more users – stronger models.”

He emphasized that the priority is to develop the most capable models, which requires joint efforts from both model and application layers for long-term improvement.

The large-scale deployment of Pingtouge chips is not only a cost-saving and efficiency tool but also a “ballast” for future computing power supply.

Zhang highlighted that, as AI shifts from “selling resources” to “selling intelligence,” combined with the scale effects of self-developed chips, cloud profit margins will experience “non-linear, abrupt improvements.”

He expects that in 2026 and 2027, Pingtouge will continue expanding its high-quality AI chip production capacity, providing sufficient computing power for Alibaba’s AI growth. Pingtouge’s software stack is well compatible with the CUDA ecosystem, lowering migration costs for customers.

However, financial realities are pressing. The company’s adjusted EBITA has fallen by 57%, and free cash flow has sharply contracted, clearly indicating that Alibaba is in a high-investment, low-return strategic phase.

Despite holding 560.175 billion RMB (USD 80.104 billion) in cash and liquid investments, whether capital markets will have enough patience to wait two, three, or even longer for instant retail profitability and the realization of the AI billion-dollar goal remains uncertain.

Disclaimer: This content and data are compiled by Guancha based on publicly available information and do not constitute investment advice. Please verify before use.