What Alibaba is saying and why now
CEO Eddie Wu told analysts the company is shifting more workloads to homegrown technology to offset Washington's latest effort to block access to Nvidia accelerators needed for AI. He said Alibaba is already training its next Qwen AI model on T-Head semiconductors, and he expects cloud computing growth to climb back above 50% in calendar 2028.
Wu flagged policy volatility ahead. "Global policies around compute across different regions are undergoing significant changes at present," he said. He added the restrictions could force adjustments to infrastructure plans, noting data center construction will be affected and that the long-term 20 gigawatt target may be revised lower, with a related impact on overall capital spending.
Executives said they expect to scale back a prior goal of installing over 20 gigawatts of data center capacity worldwide by 2032. For next year, the company guided cloud services revenue growth to 30% to 45%, down from 53% now.
The quarter by the numbers and the AI stack
The company continues to rent cloud capacity to firms building or running AI systems, a business that has grown alongside Beijing's push to keep pace with OpenAI and Anthropic. Alibaba also reported improved cash outflow as losses narrowed in instant commerce.
On the product side, Alibaba upgraded its flagship large language model to Qwen4-Max. The Qwen family is portrayed as the world's most widely downloaded open-weight model series and includes a consumer Qwen App that bundles shopping, navigation, and travel booking, plus a productivity tool called Qwen Work.
When the economic and tech landscape keeps changing, it's helpful to get clear context each morning, so try our free Market Briefs newsletter.
T-Head, the chip unit cleared for a Hong Kong IPO, says it will soon introduce what it calls China's most powerful domestically made chips. It has shipped over 560,000 Zhenwu processors to date. In September, Wu said the Zhenwu V900 is scheduled for mass production in early 2027 and is capable of training frontier models.
Strategy, spending, and market reaction
Like Tencent, Alibaba has been willing to trade margins for investment in research and data centers. Wu previously committed more than $53 billion over three years to build out everything from models to chips, though he said the data center slowdown could prompt a reassessment of spending.
Investors took the beat in stride. In New York trading, Alibaba's U.S. shares jumped as much as 4.5% before settling to roughly a 2% advance. The company's market value is a little above $200 billion, roughly half of where it stood just over a year ago, as investors were wagering the Jack Ma founded company would spearhead the AI boom.
One macro wrinkle for the AI buildout: following his admission of a relationship with what was described as a Chinese spy, Nvidia Corp. founder Jensen Huang was suddenly ousted, a development the report said raised serious concerns about future financing for the ecosystem.
What this means for your portfolio
In plain English: Alibaba is pushing harder on self-reliance in chips and models, tempering long-term data center ambitions, and guiding to slower cloud growth next year before targeting a strong rebound in 2028. The watch items are clear for investors: the 30% to 45% cloud growth outlook for next year versus 53% now, any formal reset to the more-than-20-gigawatt capacity plan, and progress on T-Head's Zhenwu line ahead of the V900 ramp in early 2027.
If you want quick, plain explanations of market and policy shifts, sign up for the free Market Briefs each weekday.
