On March 19, Alibaba Group released its Q3 fiscal year 2026 financial results (covering October to December 2025). Total revenue increased 2% year-over-year to 284.843 billion yuan, falling short of market expectations of 289.795 billion yuan. Adjusted net profit was 16.71 billion yuan, down 67% compared to the same period last year, against market expectations of 29.579 billion yuan. Net profit attributable to shareholders was 16.322 billion yuan, also declining by 67%. The reasons for the profit decline are widely understood: Alibaba is heavily investing in the instant retail and artificial intelligence sectors. Against this backdrop, Alibaba's free cash flow in the fourth quarter of last year dropped significantly by 71% to 11.346 billion yuan.
Alibaba Cloud is currently the second fastest-growing business segment for Alibaba, with fourth-quarter revenue increasing 36% year-over-year to 43.284 billion yuan, accelerating from the 34% growth recorded in the third quarter. The fastest-growing business for Alibaba is instant retail, which saw a 56% year-over-year increase in the fourth quarter.
During the earnings call, Alibaba Group CEO Wu Yongming revealed that Alibaba Cloud's external commercial revenue for fiscal year 2026, as of the end of February, had officially surpassed 100 billion yuan. Looking at the long-term growth momentum of the AI market, Alibaba Group's AI strategy has a clear commercial objective: over the next five years, combined cloud and AI commercial revenue, including MaaS, is targeted to exceed $100 billion.
The explosion of AI applications is undoubtedly the core driver of Alibaba Cloud's revenue growth. Revenue from its AI-related products has now achieved triple-digit growth for the tenth consecutive quarter. Wu Yongming stated that Alibaba has developed comprehensive, full-stack AI capabilities spanning from AI infrastructure to applications and is accelerating the construction of its MaaS platform. Over the past three months, token consumption volume on the public model service market within the Bailian MaaS platform has increased sixfold. It is projected that commercial MaaS revenue will become Alibaba Cloud's largest revenue-generating product.
T-Head, which had previously maintained a low profile, made its first appearance in an Alibaba earnings report. T-Head's self-developed GPUs have achieved mass production, supporting Alibaba's internal operations and providing commercial services to hundreds of external enterprise customers via Alibaba Cloud, already making a substantial contribution to cloud infrastructure supply. Wu Yongming disclosed that, as of February 2026, T-Head had cumulatively delivered 470,000 self-developed GPU chips on a large scale. Within Alibaba Cloud's actual business scenarios, over 60% of T-Head chips serve external commercial customers. The company has completed the adaptation of AI tasks for large-scale external clients, supporting the AI workloads of more than 400 enterprise customers across various industries, including internet services, financial services, and autonomous driving.
At the model and application level, on March 16, Alibaba officially announced the establishment of the Alibaba Token Hub business group. This new organization is focused on the core objectives of "creating tokens, transmitting tokens, and applying tokens," and is directly led by Wu Yongming. The Alibaba Token Hub integrates five business lines, covering a complete layout from underlying models to the application end: Tongyi Lab is responsible for multimodal foundational model research, continuously pushing the boundaries of model capabilities; the MaaS business line builds an open model service platform to support the AI ecosystem across all industries; the Qianwen division focuses on personal AI assistants; the AI Innovation division is tasked with exploring new models and markets; the Wukong division is a new unit appearing in the public eye for the first time in this restructuring. The Wukong division is positioned as a "B2B AI-native work platform," aiming to deeply integrate model capabilities into enterprise workflows.
During the earnings call, Wu Yongming also elaborated on Alibaba's full-stack AI layout: using chips and cloud computing as the AI infrastructure layer; and using the Token Hub as the main thread, comprising the large model, MaaS business, and "B2B + B2C" applications to form the AI model and application layer. Together, these two layers constitute Alibaba's complete capabilities from AI infrastructure to applications. Alibaba's ambitious AI strategic blueprint has now been presented more clearly to the outside world.
Comments