Alibaba Group has unveiled what it calls “the most powerful AI chip in China,” part of a grand plan to expand its global data center capacity beyond 20 gigawatts by 2032.
Speaking at the company’s 2026 Apsara Conference in Hangzhou, southwest of Shanghai, CEO Eddie Wu said the Zhenwu V900 can deliver three times the performance of its predecessor, and a single cluster built on the chip can support up to 500,000 cards to power frontier artificial intelligence model training and inference.
“Backed by the proven maturity of T-Head product lines and widespread adoption across our clients, we anticipate a significant growth in the annual AI chip shipment volumes,” said Wu, referring to products from its chip design subsidiary.
The e-commerce conglomerate is doubling down on AI at a moment when the race to build frontier models and the infrastructure to support them is intensifying in both China and the West.
In February last year, Alibaba committed to capital spending of RMB 380 billion (USD 56.5 billion) over the next three years, and as of the end of the June quarter this year, the company had already spent RMB 190 billion (USD 28.3 billion). The figure is much lower than those of US tech giants but it is one of the largest single AI infrastructure pledges among Chinese companies.
Last month, the company also raised HKD 80 billion (USD 10.2 billion) through a new share placement to fund its AI investments.
Exceeding 20 GW of data center capacity by the target date would mark a significant leap. Last year, Goldman Sachs estimated that Alibaba’s capacity was 3–4 GW, implying a need to add around 2 GW a year.
Citi analyst Alicia Yap expects Alibaba’s planned 20 GW of AI data center capacity by 2032 to generate USD 160 billion in external cloud revenue by then. She said in a note that the company’s capital expenditure could remain elevated in the coming years, while some capacity will be secured through partnerships and accounted for as operating expenses rather than capital investment.
Wu laid out a case for the company’s ambitious data center expansion. “Machine thinking still has an enormous growth runway,” he said. “As machines are becoming the primary force behind thinking, turning intelligence into a commodity supplied at scale, the truly groundbreaking products of the machine intelligence era have not yet arrived.”
His remarks stand in contrast to calls to slow down the pace of AI development by US frontier AI labs such as Anthropic. Earlier this month, Huawei’s rotating chairman Eric Xu expressed a similar view that Chinese companies should move faster in AI so that they can feel the same risks that US AI leaders already sense.
The Alibaba chief’s comments also came just before a summit between Chinese President Xi Jinping and his US counterpart, Donald Trump.
The US continues to prohibit exports of top-tier AI accelerators to China, and analysts do not expect the summit to produce breakthroughs on US export controls, sanctions against Chinese companies or China’s curbs on shipments of rare earths.
“We are still in Chapter 1 of AI deployment, and there is significant room for further growth in AI-related data center operations,” said Simon Woo, Asia technology research coordinator at BofA Global Research.
“Any company can claim they have developed the best chips, but what matters is the commercial impact—how many are actually being deployed. For now, global AI data center training remains predominantly reliant on US high-end chips, followed by non-US alternatives,” he said.
Still, the landscape is shifting. “Going forward, AI semiconductor chips will become more diversified,” Woo added.
This article first appeared on Nikkei Asia. It has been republished here as part of 36Kr’s ongoing partnership with Nikkei.
Note: HKD, RMB figures are converted to USD at rates of HKD 7.84 = USD 1 and RMB 6.72 = USD 1 based on estimates as of September 24, 2026, unless otherwise stated. USD conversions are approximate and, where appropriate, rounded for ease of reference. They may not fully match prevailing exchange rates.