'Zero possibility': Nvidia CEO says US shouldn't fear Chinese AI models

Nvidia CEO Jensen Huang has pushed back against growing concerns over the rapid rise of affordable Chinese AI models and the increasing calls in Washington to restrict their use. In a recent interview with Axios, Huang argued that open-source Chinese AI models are a net positive for the industry, saying they are "excellent and should definitely be used."

The world's leading AI chipmaker said American companies should be free to use Chinese AI models, even as Trump administration officials and several leading US AI labs continue lobbying Washington to curb their adoption.

"There's no scenario where China runs US companies off the road," Huang told Axios. "Zero possibility."

More from Tech

His remarks come amid renewed debate over China's AI progress following the release of Kimi K3, a new open-source model from Beijing-based Moonshot AI. The model has sparked one of the biggest waves of concern in the AI industry since DeepSeek rattled financial markets in January 2025.

Kimi K3 combines near-frontier performance with significantly lower costs and open weights that developers can download and customize. Those capabilities have intensified questions about whether companies can continue justifying the massive spending on AI infrastructure and proprietary models.

Huang also dismissed concerns that downloading Chinese AI models could create a "backdoor" for Beijing. He argued that organizations deploying these models retain full control over how they are used by running them inside secure environments and customizing them to suit their needs.

He further said that open-source development enables researchers to identify vulnerabilities and fix them more quickly, ultimately making AI models stronger and more secure.

The Nvidia chief also criticized Anthropic's decision to limit access to some of its cybersecurity-focused AI models. Huang argued that powerful AI systems should be strengthened through rigorous testing and rapid security fixes rather than restricting access from the outset.

His views stand in sharp contrast to those of US Treasury Secretary Scott Bessent, who recently told Fox Business that the administration is examining whether Chinese AI models incorporate stolen US intellectual property and is considering sanctions against companies found to be violating American interests.

China remains a critical market for Nvidia

Huang's comments also come at a time when Nvidia's business in China continues to face significant pressure.

China has long been one of Nvidia's most important markets. Last year, the company generated nearly $20 billion in revenue from the country, accounting for roughly 9% of its total sales. However, tighter US export restrictions have taken a heavy toll. Revenue from China fell by about half year over year to approximately $4.5 billion in the fiscal first quarter, and Huang has since indicated that sales have continued to deteriorate. Although the US has approved some export licenses for Nvidia's H200 AI chip, the company has yet to recognize any revenue from those shipments and has excluded China data center sales from its forward guidance.

Despite the slowdown in China, Nvidia has been rapidly expanding into new growth areas. The company is emerging as a major force in the CPU market alongside its long-standing dominance in AI GPUs. Management expects to generate nearly $20 billion in CPU revenue this year—an amount that could effectively offset the revenue once generated from China.

By combining its Vera CPUs with GPUs, high-speed networking, and AI computing racks into a unified platform, Nvidia is challenging long-established rivals such as Intel and AMD. The strategy has helped fuel explosive growth in its data center business, where first-quarter revenue nearly doubled year over year to $75 billion, putting the company on track for yet another record year.