The ecommerce company turned AI-aspirant on Tuesday outlined a new Zhenwu V900 accelerator that it says triples the performance of its predecessor. From Alibaba’s T-Head division, it can be combined in clusters of up to 500,000 units to power frontier-model training, chief executive officer Eddie Wu said. 

Alibaba is one of the biggest spenders on AI among leading Chinese players. It’s committed to spending more than US$53 billion over a three-year period to expand its AI capabilities, and raised about US$10.2 billion from a follow-on share offering in Hong Kong in August. On Tuesday, Wu set a target of 20 gigawatts of data centre capacity for its Alibaba Cloud service globally by 2032, on the expectation of “exponentially rising demand.” 

Wu spoke as global AI stocks rallied on early signs of success for Meta Platforms Inc’s new personal agent, reviving investor optimism after recent market uncertainty. Alibaba’s shares leapt more than 5% in Hong Kong, while WeChat operator Tencent Holdings Ltd surged more than 7%.

“Last year, we asserted that the more capable AI becomes, the more powerful humanity will be,” Wu said. “Today, I remain steadfast in that conviction.”

Alibaba’s 20-gigawatt blueprint — while giant for a Chinese company — is relatively smaller than some of the world’s more ambitious buildouts. SoftBank Group Corp for instance envisions a 10-gigawatt-scale computing centre in Ohio, alone.

Alibaba has pledged to intensify expenditure on AI hardware and infrastructure as Wu makes the pursuit of artificial general intelligence the company’s lodestar. It also plans to list T-Head to tap interest in the white-hot AI accelerator market, Bloomberg News has reported. The company now plans to develop a model with between five and 10 trillion parameters, far larger than what’s common among open-weight peers today, for longer and more complex tasks.

That pace of development contrasts with recent warnings from top AI labs in the US, after Anthropic PBC CEO Dario Amodei urged a slower pace of development to better handle AI risks.

Alibaba’s event comes on the eve of a summit between US President Donald Trump and his Chinese counterpart Xi Jinping, when the leaders will discuss a swathe of issues. AI matters are set to figure prominently, as top-level US executives like Nvidia boss Jensen Huang and Microsoft Corp.’s Satya Nadella attend a White House state dinner.

Talks got underway this week between US Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng, where both sides agreed to set up a dialog on AI aimed at reaching a common understanding on goals and threats.

That’s despite Washington accusing Alibaba and other leading developers like DeepSeek of “illicitly accessing” American cutting-edge models to help develop their own products. Beijing has rejected such accusations as groundless and vowed to retaliate if Washington moves to target Chinese AI companies. 

Alibaba — till recently better known as the operator of the Taobao and T-mall marketplaces — created the world’s most popular AI model by downloads. The company has since rapidly expanded its Qwen series, while keeping pricing low to entice users. The Hangzhou-based company however saw quarterly profit plunge by more than 75% in the last quarter, after a nearly US$10 billion spending spree.

The company’s executives have in recent times talked about how its AI investment is increasingly bearing fruit, with annualised revenue from AI-related products expected to reach US$10 billion in the current quarter. Alibaba should be able to recoup its overall investment within a three-year period, Wu’s team has said. It also aims to quintuple annual cloud and AI revenue to US$100 billion in five years.

Alibaba — which last unveiled a new AI chip in May — plans to update that lineup every year. They’re designed to work in giant clusters to drive performance, a strategy widely adopted by Chinese chipmakers from Huawei Technologies Co to Moore Threads Technology Co.

T-Head has delivered 560,000 units of the Zhenwu product family, and its processors have been deployed by more than 400 external customers, Alibaba has said.

Uploaded by Liza Shireen Koshy