Alibaba is betting on artificial intelligence infrastructure with a strategic ambition comparable to the Industrial Revolution.
On September 22, at the 2026 Hangzhou Apsara Conference, Alibaba Group CEO Eddie Wu delivered his clearest AI expansion signal to date, positioning machine intelligence as the most profound transformation for human society since the Industrial Revolution, and anchoring a full-stack strategy spanning AI models, AI chips, and the AI cloud.
Wu announced that Alibaba aims to expand the global data center capacity operated by Alibaba Cloud to more than 20 gigawatts (GW) by 2032, while simultaneously disclosing plans to train a new-generation AI model with 5 to 10 trillion parameters.
Meanwhile, the company's chip division, T-Head, unveiled its latest AI chip, the Zhenwu V900, on the same day. Its computing power is three times that of the previous-generation M890, and a single cluster can scale to an impressive 500,000 cards. Alibaba stated that it expects a substantial increase in the annual shipment volume of T-Head AI chips.
Wu predicted that the total amount of machine-generated thinking will reach more than 1,000 times that of humans in the future. He noted that current long-term industry demand far exceeds supply capacity, and the global shortage in the AI data center supply chain is constraining the growth rate of computing power. This supply-demand gap is the core rationale behind Alibaba's large-scale expansion.
Following the announcement, Alibaba's Hong Kong-listed stock continued to strengthen, rising more than 4% at one point to hit a monthly high.
A Significant Supply-Demand Gap in Computing Power Sets a 20GW Data Center Expansion Target
Regarding the commercialization and order situation of its cloud business, Wu candidly stated that the current market is in a state of supply exceeding demand. He pointed out that customer AI demand is currently very strong, and Alibaba Cloud is fully deploying AI computing power to meet this demand, which has directly driven the continued acceleration of Alibaba Cloud's revenue growth.
However, current computing capacity still faces bottlenecks. Wu was blunt about the supply chain situation: "We see that the industry's long-term demand far exceeds our supply capacity. Currently, the global shortage in the AI data center supply chain limits the growth rate of our computing power."
Faced with this vast market potential, Alibaba provided aggressive infrastructure construction guidance. Wu emphasized that Alibaba will firmly invest in infrastructure represented by AI models, AI chips, and the AI cloud.
To build an AI cloud for the "machine intelligence" era, Alibaba has set a long-term goal: "Our goal is to expand the global data center capacity operated by Alibaba Cloud to more than 20GW by 2032, to meet the rapidly growing AI demand."
Driven by Both Hardware and Software: Launching the Strongest AI Chip and Developing a 10 Trillion-Parameter Model
As a core component in breaking the computing power bottleneck, T-Head provided a clear iteration roadmap and shipment expectations in terms of underlying hardware.
On the same day, T-Head officially released its new-generation AI chip, the "Zhenwu V900." The chip is positioned as the most powerful AI chip in terms of Chinese computing performance, with computing power performance reaching three times that of the previous-generation Zhenwu M890.
In terms of cluster scalability, a single cluster built on the Zhenwu V900 can be expanded to an astonishing 500,000 cards, which is sufficient to support the training and inference of cutting-edge supermodels. Regarding future performance guidance, Alibaba expects that as the T-Head chip product line matures and customer adoption broadens, its annual AI chip shipment volume will increase significantly.
As for progress on large AI models, the Qwen team is exploring a technical path toward superintelligent AI (ASI), known as recursive self-improvement (RSI).
Alibaba disclosed plans to train a new model with 5 to 10 trillion parameters, aiming to complete more complex and longer-horizon tasks. Simultaneously, multimodal models that integrate understanding and generation have become a core research direction.
"Total Machine Thinking Will Reach 1,000 Times That of Humans"
In addition to specific business data, Wu's macro-level assessment of the AI industry cycle in his speech also attracted market attention. He compared the current development of AI to the early era of electricity.
Wu stated: "Machines are becoming the main force of thinking, and intelligence is becoming a scalable commodity. In the future, the total amount of machine thinking will reach more than 1,000 times that of humans."
He believes that when the supply of machine intelligence becomes infinitely abundant, its significance to human society will far exceed the replacement of existing mental labor, and this change will be even more profound than the Industrial Revolution.
Regarding the various AI applications currently receiving widespread attention, Wu maintained a calm and objective perspective: "Today's AI Coding is like the electric light bulb in 1882. It replaces existing work but is not yet enough to create a new era. Helping people code or write reports is only the initial stage of machine intelligence."
Wu further emphasized: "No matter how many new inventions emerge later, the first step is to build enough power stations and lay a wide enough power grid."
He concluded that in the future era of machine intelligence, thinking will become a scalable commodity like power, and Alibaba's current strategic focus is to spare no effort in building this "super power grid" that will power the machine intelligence era.
As of the time of writing, Alibaba (HKEX: 09988) shares are up more than 4% in Hong Kong trading, reaching a new monthly high, reflecting strong market confidence in the company's aggressive AI infrastructure roadmap.
Comments