Alibaba's AI narrative is approaching a turning point.

On the evening of March 19, Alibaba released its financial results for the third quarter of fiscal year 2026. Revenue for the quarter reached 284.843 billion yuan, a 2% increase year-over-year. Excluding revenue from disposed businesses like Gaoxin Retail and Intime, revenue on a comparable basis grew by 9% year-over-year.

However, during the earnings call, a statement from Yongming Wu, known for his rigorous, engineering-oriented demeanor, was more impactful than the financial figures themselves. He projected that "external cloud revenue is expected to achieve a compound annual growth rate of over 40% in the coming five years, with a revenue target of $100 billion in the fifth year."

For the full fiscal year 2025, Alibaba's total revenue was 996.347 billion yuan. This target implies that within five years, AI and cloud-related businesses are expected to generate over half of Alibaba's current total revenue, positioning them as the group's definitive growth engine.

Wu's confidence stems from the strong growth momentum of Alibaba Cloud this quarter. Financial reports show that revenue for the Cloud Intelligence Group grew 36% year-over-year to 43.284 billion yuan, continuing an accelerating trend. Adjusted EBITA was 3.911 billion yuan, up 25% year-over-year.

More noteworthy for the market is the potential valuation re-rating implied by this target. As Alibaba Cloud revenue approaches $100 billion, with improving profit margins and an increasing share of the group's overall revenue, its valuation logic is expected to shift from a lower-multiple Price-to-Earnings ratio to a higher-multiple Enterprise Value-to-Sales ratio. This could lead to a standalone valuation for the cloud business in the range of $500 billion to $700 billion, already surpassing Alibaba's current total market capitalization.

Wu expressed strong confidence in this path, stating that based on the existing business and product foundation, the visibility towards achieving this goal is high.

"We believe the biggest growth driver comes from breakthroughs in large AI model capabilities. Since 2026, a clear trend has emerged—large models are beginning to handle complex B2B workflows. As more enterprises start deploying AI Agent-driven systems internally to manage end-to-end tasks, the fundamental nature of the IT budget market targeted by AI and cloud services has changed. When enterprises consume tokens, they no longer view it as an IT expense, but rather as a production cost, R&D cost, or part of their means of production," Wu explained.

Several other significant data points support this outlook: Revenue from Alibaba Cloud's AI-related products has achieved triple-digit year-over-year growth for ten consecutive quarters. Over the past three months, token consumption on the Bailian MaaS platform surged sixfold. T-Head, making its debut in the financial reports, has already delivered 470,000 of its self-developed GPU chips at scale. Furthermore, on March 18, Alibaba Cloud announced price increases of up to 34% for AI computing power and cloud storage products, citing "exploding global AI demand and rising supply chain costs."

These signals collectively indicate that the prospect of $100 billion in external cloud revenue is not unfounded. Previously, the market was concerned that Alibaba's AI business would entail heavy capital expenditures without translating into substantial profits. However, demand growth now appears certain, Alibaba's full-stack AI capabilities are in place, and pricing power has returned. Alibaba's AI narrative seems poised to cross the inflection point from the investment phase to the value realization phase.

While Alibaba's transformation over the past three years has gained mainstream recognition, and the perception has shifted from viewing it as merely an "e-commerce company" to one with strong technological foundations, its valuation in the capital markets has remained anchored to the low P/E multiples typical of traditional e-commerce. Metrics like growth rate, profitability, and market share continued to be the primary focus for scrutiny.

If Wu's $100 billion target is achieved on schedule, and the valuation of the cloud and AI business surpasses that of the entire Alibaba group, AI will completely replace e-commerce as the company's core value foundation. This would mark the official start of Alibaba's "second venture," freeing it from the valuation constraints of traditional e-commerce and initiating a comprehensive transformation from an "e-commerce giant" to an "AI technology giant."

**Transitioning from "Selling Computing Power" to "Selling Intelligent Capabilities"**

In the second half of 2025, Wu repeatedly emphasized that Alibaba Cloud aims to be a "full-stack artificial intelligence service provider." The recent establishment of the Alibaba Token Hub business group underscores Alibaba's commitment to integrating these full-stack capabilities.

On March 16, Alibaba announced a new round of organizational restructuring, forming the ATH business group, which will be directly managed by Wu Yongming. This department encompasses core AI businesses including the Tongyi Lab, MaaS business line, Qianwen business unit, Wukong business unit, and the AI Innovation business unit, covering the entire industrial chain from foundational model research and development to model service platforms and AI applications for both individual and enterprise users.

Post-restructuring, Alibaba Group's business structure consists of: Alibaba China Commerce Group, Alibaba International Digital Commerce Group, Cloud Intelligence Group, Alibaba Token Hub business group, and other businesses.

During the earnings call, Wu explained the purpose of establishing the ATH business group: "From the second half of 2025 to the beginning of 2026, over these two to three months, we have observed AI entering a new era driven by agentic capabilities. The biggest difference from the early stages of AI is the tight integration between models and applications, which is crucial for developing effective models and applications."

Alibaba has now completed its full-stack AI layout: with chips and cloud computing forming the AI infrastructure layer; and the AI model and application layer, centered around the Token Hub and comprising large models, MaaS services, and both B2B and B2C applications. Together, these layers create a complete capability stack from AI infrastructure to applications—a logic somewhat similar to Google's approach.

If the popularity of DeepSeek during the 2025 Spring Festival accelerated progress in large models, then the rapid rise of Agent products like OpenClaw after this year's Spring Festival has directly driven a surge in token usage demand.

In fact, token usage demand began to skyrocket in the latter half of last year. According to a report by the international market research firm Frost & Sullivan, the average daily token usage by enterprise-level large models in China surged to 37.0 trillion tokens in the second half of 2025, a 263% increase from 10.2 trillion tokens in the first half. The market share of leading large models increased, with Alibaba Cloud's Qwen showing the most significant growth, its share jumping to 32.1%, nearly double the 17.7% share in the first half.

Against this backdrop, Alibaba Cloud's business model is undergoing a substantive transformation: shifting from "selling computing power" to "selling intelligent capabilities." The core of traditional cloud computing is resource leasing, whereas in the AI era, users increasingly prefer to acquire intelligent capabilities directly through "Model-as-a-Service." The delivery of these capabilities relies not just on underlying computing power, but more critically on the performance and efficiency of the models themselves. Leveraging its leading domestic cloud infrastructure and large model capabilities represented by Qwen, Alibaba Cloud is beginning to offer intelligent services with both performance and cost advantages. Compared to merely selling computing power, this model offers higher profit margins and builds more sustainable long-term competitiveness.

More importantly, the token business is reshaping the user base. It is no longer confined to traditional enterprise clients but is expanding massively to developers, individual users, and even "one-person companies." As the user base grows exponentially, the costs of underlying computing power and models will be continuously diluted, scale effects will emerge, and this will subsequently drive optimization of the overall profit structure.

**The True Value Re-rating**

A critically important component of Alibaba's full-stack capability is its self-developed chip technology at the foundational layer, and the signals here are now very clear.

For the first time, management highlighted "T-Head" during the earnings call. Regarding potential listing plans for T-Head, Wu stated plainly that an IPO is not ruled out in the future.

He also revealed that T-Head has successfully achieved commercialization, with chip deployment exceeding 470,000 units and annualized revenue surpassing 10 billion yuan. In 2026 and 2027, its production capacity for high-quality AI chips will continue to expand, providing sufficient computing power assurance for the group's AI business. This indicates that T-Head has crossed the critical threshold from research and development to commercialization and is beginning to function as an infrastructure component within Alibaba's AI ecosystem.

Wu stated: "Beyond overall efficiency improvements and cost reduction, another significant value for us is the assurance of AI computing power supply. Over the next 3 to 5 years, global AI computing power will be in a very tight supply phase. As the only cloud computing company in the Chinese market with self-developed chip capabilities, T-Head is also extremely important for the Alibaba Group."

This point is particularly crucial—it signifies that T-Head is not merely a closed "internal technology department" but is evolving into a commercial entity capable of exporting capabilities externally and participating in market competition. On a deeper level, in the context of future computing power shortages, self-developed chips bring not just cost optimization, but supply certainty. The ability to shift computing power supply from external dependence to internal control will be significantly amplified during periods of supply-demand imbalance.

It is precisely under this logic that the synergy between chips, cloud, and models is beginning to reshape Alibaba's commercial structure. Chip research and development itself involves very high barriers to entry. When chip technology is synergistically optimized with the cloud platform and model capabilities, its value is further magnified: it can not only support model inference more efficiently but also enables an integrated capability across the entire chain from chips and hardware to cloud services and token output, thereby achieving a genuine "full-stack premium."

On March 18, Alibaba Cloud announced that due to exploding global AI demand and rising supply chain costs, prices for its AI computing power and cloud storage products would increase by up to 34%. This round of price adjustments includes a 5% to 34% increase for computing card products like T-Head's Zhenwu 810E, and a 30% increase for the CPFS file storage product. According to informed sources, this price hike is also based on the explosive growth in token usage.

Returning to the highly ambitious projection: within five years, the standalone cloud and AI business is expected to exceed the valuation of the entire group. Wu indicated that based on Alibaba's transition from "selling computing power" to "selling intelligent capabilities," combined with the high efficiency provided by self-developed T-Head chips, the company internally believes that as AI and cloud revenue scales, cloud profit margins will show a sustained and visible upward trend, potentially reaching a 20% level in the long term.

"However, it must be clarified that this improvement process is not linear. It may experience staged leaps driven by sudden changes in scale effects, significant increases in T-Head chip production scale, or the manifestation of overall product scale effects," Wu emphasized.

Only when the cloud and AI business truly becomes a core segment capable of independently supporting a fundamental valuation reassessment will the end goal of Alibaba's "second venture" come into sight.