Nvidia concedes China’s AI chip market to Huawei amid export restrictions

Nvidia once owned 95% of China’s AI chip market. Now it owns zero percent. That’s not a typo, and it’s not hyperbole. It’s the direct consequence of US export controls that have systematically locked America’s most valuable chipmaker out of the world’s second-largest economy.

CEO Jensen Huang has acknowledged the collapse himself, conceding that Nvidia has effectively lost its entire Chinese AI accelerator business. The beneficiary sitting across the table? Huawei, which is rapidly positioning itself as China’s primary supplier of advanced AI hardware.

How US export controls erased Nvidia’s China business

The unraveling started in October 2022, when the US Commerce Department rolled out sweeping export restrictions targeting advanced semiconductor sales to China. The initial rules barred Nvidia from shipping its flagship A100 and H100 GPUs to Chinese customers.

Nvidia tried to play the game. The company designed downgraded chip variants specifically to comply with the new performance thresholds. But Washington moved the goalposts. In October 2023, regulators tightened the rules again, effectively killing those workaround products too.

It wasn’t just the US government squeezing Nvidia from one side. Chinese authorities instructed the country’s major tech companies to stop purchasing Nvidia’s AI chips entirely. Both governments were pushing Nvidia out of the market simultaneously, from opposite directions.

Huawei steps into the vacuum

With Nvidia sidelined, Huawei has emerged as the obvious replacement, backed by significant state support and an increasingly capable domestic chip ecosystem.

Major Chinese tech firms, from Baidu to Alibaba to Tencent, are increasingly turning to Huawei’s Ascend series processors for their AI training and inference workloads.

The policy paradox nobody wants to talk about

The stated goal of US export controls was to slow China’s AI development by restricting access to cutting-edge chips. Policy analysts are now raising uncomfortable questions about whether that assumption was wrong. By cutting Nvidia off from the Chinese market entirely, the US may have accomplished something no Chinese industrial policy could have achieved on its own: it gave domestic chip companies a captive market of billions of dollars in annual demand, with zero foreign competition.

Before the export bans, Chinese companies had little incentive to switch away from Nvidia. The CUDA software ecosystem was deeply entrenched, the hardware was best-in-class, and the switching costs were enormous. The export controls eliminated all of those barriers overnight. Companies that would have happily stayed on Nvidia’s platform for another decade were forced to invest in Huawei’s ecosystem instead.