Alibaba Unveils China's Most Powerful AI Chip in Direct Challenge to Nvidia

Alibaba Group used its annual Apsara Conference in Hangzhou on Tuesday to unveil what it calls China's most powerful artificial intelligence chip, escalating the country's effort to reduce its dependence on Nvidia and other Western semiconductor makers.

A new accelerator built for scale

The chip, named Zhenwu V900, comes from Alibaba's in-house semiconductor unit, T-Head, and is said to triple the performance of its predecessor. Chief Executive Eddie Wu said the accelerator can be linked together in clusters of up to 500,000 units, a configuration intended to support the training of large, frontier-scale AI models. T-Head has already shipped 560,000 units across its Zhenwu chip family to more than 400 external customers, according to the company.

Alongside the chip announcement, Wu laid out an ambitious infrastructure target: expanding Alibaba Cloud's global data centre capacity to more than 20 gigawatts by 2032. He attributed the scale-up to what he described as "exponentially rising demand" for AI computing power. While large by Chinese standards, the figure is still smaller than some Western buildouts, such as SoftBank's planned 10-gigawatt facility in Ohio alone.

Big spending, bigger ambitions

Alibaba has committed more than $53 billion toward AI capacity expansion over three years and raised roughly $10.2 billion through a follow-on share sale in Hong Kong in August to help fund the push. The company is also reportedly working on a new AI model with between 5 trillion and 10 trillion parameters, a scale well beyond what is common among openly available models today, aimed at handling longer and more complex tasks.

Markets responded positively to the announcement. Alibaba shares rose more than 5% in Hong Kong trading, while Tencent, another major Chinese tech name, surged over 7% on the back of broader optimism around AI stocks following encouraging early signs for Meta's new personal AI agent.

Timing matters

The unveiling comes just two days before a scheduled summit between US President Donald Trump and Chinese President Xi Jinping in Washington, where artificial intelligence is expected to be high on the agenda. Senior US technology executives, including Nvidia's Jensen Huang and Microsoft's Satya Nadella, are expected to attend related events. Washington has previously accused Alibaba and other Chinese AI developers, including DeepSeek, of improperly accessing American models to accelerate their own development — allegations Beijing has rejected.

Talks between US Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng this week reportedly touched on establishing a broader dialogue between the two countries on AI goals and risks, suggesting the technology has become as much a diplomatic flashpoint as a commercial one.

The bigger picture

Alibaba's aggressive AI roadmap comes despite a rocky quarter financially — the company reported a more than 75% drop in quarterly profit after its heavy AI-related spending. Still, executives maintain that AI-linked revenue is on track to reach an annualised $10 billion this quarter, with a goal of quintupling combined cloud and AI revenue to $100 billion within five years.

The pace of Alibaba's rollout also stands in contrast to cautionary notes struck by some Western AI leaders. Anthropic CEO Dario Amodei has urged a slower, more careful approach to advanced AI development, citing safety concerns — a call that Wu's Tuesday remarks appeared to directly push back against, reaffirming his belief that "the more capable AI becomes, the more powerful humanity will be."

This is an original summary based on public reporting. See our editorial policy for how we source, write, and correct our stories.