American firms should be free to use Chinese AI models, Nvidia's Jensen Huang says
Against the backdrop of intensifying US-China technology controls, the question of which artificial intelligence software American companies may legally use has grown nearly as contested as chip export rules. Nvidia (NVDA) chief…
Key takeaways
- Nvidia CEO Jensen Huang has stated that American companies should be permitted to use Chinese AI models.
- Huang's position runs counter to the current direction of Washington's technology policy.
- Huang's argument concerns the software layer (AI models), which is distinct from the existing US export restrictions on advanced chips sold into China.
- Because AI model selection sits upstream of compute procurement and Nvidia sits in that procurement chain, restricting American firms from foreign-built models narrows the competitive software set and shapes future hardware demand.
- The AI sector is in an active capital-expenditure cycle, with data center investment running across North America, Europe, and Asia, making cross-border model access a live planning variable.
Against the backdrop of intensifying US-China technology controls, the question of which artificial intelligence software American companies may legally use has grown nearly as contested as chip export rules. Nvidia (NVDA) chief executive Jensen Huang has stated that American companies should be permitted to use Chinese AI models, a position that runs counter to the current direction of Washington's technology policy.
What Huang said and where the argument sits
Huang's case is at the software layer. Nvidia already operates under export restrictions on advanced chips sold into China. The question he is now raising publicly is whether the flow should be permitted in the other direction, with American enterprises drawing on models built by Chinese developers. Those two debates are related but distinct, and conflating them obscures where the actual constraint sits in the AI supply chain.
The sector read-through for compute demand
AI model selection sits upstream of compute procurement. When American companies are restricted from using models built outside the United States, the competitive software set narrows. Over time, a narrower software environment shapes which hardware configurations get specified and by which developers. Nvidia sits directly in that procurement chain, which is part of why Huang's view on model access carries industry weight beyond chip policy alone.
The macro caveat
The AI sector is in an active capital-expenditure cycle. Data center investment is running across North America, Europe, and Asia, and cross-border access to models has become a live variable in how companies plan their AI programs. Huang's public stance reflects the view that software access rules and hardware demand are connected. How Washington answers the model-access question will define the competitive software environment for American AI development and, by extension, the procurement decisions Nvidia and its peers are counting on.