Alibaba sails into the "unmanned zone"

Author | Zhou Zhiyu

Editor | Zhang Xiaoling

In the light drizzle of early autumn in Hangzhou, Alibaba CEO Eddie Wu took the stage at the Yunqi Conference and delivered a speech titled "The Road to Super Artificial Intelligence." This was not only his annual report two years into his leadership at Alibaba but was also interpreted by the outside world as a decisive declaration.

Previously, there had been rumors in the market about Alibaba increasing its capital expenditure, and at the Yunqi Conference, Eddie Wu's response far exceeded market expectations.

Building on the plan proposed in February to invest 380 billion in AI infrastructure, Eddie Wu stated plans for even greater investment. He further added a stunning goal: by 2032, the energy consumption scale of Alibaba Cloud's global data centers will increase tenfold from the 2022 level.

The capital market quickly provided the most direct response. On September 24, Alibaba's Hong Kong stock continued to rise, surging over 9% during the session, reaching a new high since October 2021. Fund managers and analysts holding billions in capital began to frantically update their models in internal emails and reports. The core action was singular: re-evaluate Alibaba.

In addition to capital expenditure, Eddie Wu put forth two highly disruptive new judgments: that large models are the next generation of operating systems, and that super AI cloud is the next generation of computers.

This means that large models will consume traditional software, and in the future, almost all software will be generated by large model-produced Agents. In this new paradigm, natural language is the programming language, Agents are the software, and Alibaba Cloud's goal is to build this "super AI cloud" to provide a smart computing power network globally.

This entirely new narrative demands that the capital market completely abandon the old valuation coordinate system. Alibaba is attempting to tell investors: do not price Alibaba Cloud using the traditional IDC (Internet Data Center) logic; instead, use a higher-dimensional coordinate system belonging to the ASI era to re-examine it.

The so-called ASI is a new concept. Eddie Wu directly elevated the goal from the industry's hot topic of AGI (Artificial General Intelligence) to ASI (Super Artificial Intelligence) at the conference. He clearly outlined the three stages toward ASI: from the "intelligent emergence" of learning human knowledge, to "autonomous action" assisting humans, and finally achieving autonomous learning and "surpassing humans."

Supporting this narrative is a saturated investment in technology and products. At this conference, Alibaba Cloud launched a "model seven-in-one" with seven heavyweight new products, including the flagship model Qwen3-Max, the next-generation architecture Qwen3-Next, and the visual programming model Qwen3-VL. Among them, the flagship model Qwen3-Max has ranked among the top three on the global authoritative list LMArena, surpassing top models like GPT-5.

These are no longer visions on a PPT but have already transformed into verifiable and callable business capabilities, directly reflected in the financial data. Alibaba's latest financial report shows that Alibaba Cloud's quarterly revenue surged 26% year-on-year, with AI-related revenue achieving triple-digit growth for eight consecutive quarters Previously, Goldman Sachs raised the valuation of Alibaba Cloud from $36 per ADS to $43. The core reason is the strong demand from enterprises for large models, presenting new development opportunities for Chinese cloud service providers.

However, painting a grand blueprint and turning that blueprint into reality are two entirely different things. Eddie Wu pointed out to Alibaba a "no man's land" filled with both opportunities and risks.

On the technical front, by 2025, the global large model field is progressing amidst entanglements, with even OpenAI's GPT-5 not meeting expectations, and comments on technological stagnation are incessant. The ASI path proposed by Alibaba, with its third phase goal of "surpassing humans," is quite similar to the AGI Level 6 defined by Google DeepMind, which remains a frontier to be explored globally.

Eddie Wu asserted, "Tokens are the electricity of the future AI world." Behind this metaphor lies a fundamental transformation of the business model.

Xu Dong, General Manager of Alibaba Cloud's Tongyi large model business, revealed that a year ago, large model calls were mostly offline tasks; a year later, the volume of online task calls has increased by dozens of times, with enterprises embedding large models into production processes. This means that customers are shifting from purchasing "computing power duration" to purchasing "model call volume," necessitating the re-establishment of rules and pricing systems for this new market.

The upcoming competition will no longer be won solely by models acting independently. Alibaba Cloud CTO Zhou Jingren candidly stated, "The current competition among models is already a competition between systems."

Currently, global tech giants are aggressively investing in AI, with Nvidia's $100 billion investment plan in OpenAI and Eddie Wu's prediction of over $4 trillion in global AI investment over the next five years indicating the brutality of this race. Domestically, companies have also diverged in their approaches: Tencent focuses more on scenario implementation, while Alibaba relies on open-source to carve out a path in the global market.

Alibaba Cloud's confidence lies in its chosen "Android" ecological position. Its strategy is similar to Google's, conducting full-stack self-research from computing power and cloud computing to models, ensuring that each layer achieves international leadership. At the same time, Alibaba is the most aggressive giant in China regarding the open-source route. As of now, Tongyi Qianwen has open-sourced over 300 models, with global downloads exceeding 600 million times and over 170,000 derivative models, all ranking first globally. This builds a strong ecological moat for Alibaba.

This is inseparable from the helm of Eddie Wu. Since taking over as CEO of Alibaba Cloud in 2023 and proposing the "AI-driven, public cloud first" strategy, Eddie Wu has implemented a series of decisive reforms: returning to the public cloud, cutting low-profit projects, and concentrating resources heavily on AI.

This is a gamble with no way back. If successful, Eddie Wu will prove that Alibaba can not only seize the opportunities of the previous e-commerce era but can also become a defining force in the next AI era. In the landscape of tech stocks in China and globally, Alibaba will reclaim its core position. However, the exploration of the "no man's land" also entails extremely high risks, with massive investments, uncertain commercial returns, and fierce competition. Any misstep in any link could cause this giant ship to lose its way in the storm The pointer of history seems to have turned back to that afternoon in the lakeside garden of 1999. More than twenty years later, Alibaba once again stands at a crossroads that will determine its fate. Eddie Wu and his team have chosen the most difficult yet most exciting path. The capital market has already heard their declaration, but the real test has just begun