Nvidia CEO Defends Chinese AI Models, Calls Backdoor Fears a Misconception

Nvidia CEO Jensen Huang defended Chinese artificial intelligence models at a moment when pressure to restrict them is building in Washington. In an interview with Axios published on July 22, 2026, Huang said plainly that American companies should be able to use Chinese open source models, calling concerns about built-in backdoors for the Chinese government a misconception.

What Huang Said

The Nvidia CEO said Chinese language models have reached a level that deserves recognition, regardless of country of origin. In his view, open models that are good should be used by companies around the world, and blocking them for political reasons hurts Americans more than it hurts China.

These Chinese models are excellent. Open-source models that are excellent should be used. - Jensen Huang, CEO of Nvidia

The Backdoor Argument

A key element of Huang's remarks was his rejection of the claim that downloading and running a Chinese model opens a door to surveillance by Beijing. He pointed out that open source models, unlike cloud services, can be fully inspected, modified, and run in an environment completely cut off from the internet.

Huang argued that transparency of code and model weights works in favor of security rather than against it. In his view, the more independent researchers who can look inside a model and search for problems, the harder it is to hide malicious mechanisms in it. He also warned that relying on a single closed system creates a single point of failure, which is an easier target than a diverse ecosystem of open models.

Background of the Washington Dispute

Huang's comments come amid heightened political debate over Chinese artificial intelligence in the US. Treasury Secretary Scott Bessent had previously threatened Beijing with sanctions over the alleged theft of American AI models, and some politicians in Washington are calling for restricting or even banning the use of Chinese models by American companies and institutions.

The immediate trigger for the current wave of concern was the launch of Moonshot AI's Kimi K3 model on July 16, 2026. The model immediately unsettled investors in Silicon Valley, since its high quality at a lower cost of use undermines the assumption that the technological edge in AI safely belongs to the US side. The drop in chipmaker shares following Kimi K3's debut showed how sensitively the market reacts to Chinese progress in artificial intelligence.

Nvidia's Stake in the Debate

Huang's position is not disinterested. Cheaper and more accessible models, regardless of country of origin, increase overall demand for computing power, and therefore for Nvidia's graphics processors. The more companies and developers turn to any AI model, open or closed, the greater the demand for the infrastructure those models run on.

Nvidia has spent months balancing between the two sides of the US-China trade dispute. The company depends both on chip sales in the Chinese market and on good relations with Washington, where decisions on export restrictions are made. Huang's public defense of Chinese models fits his established line, in which he has repeatedly opposed political barriers that hinder the global flow of AI technology.

What It Means for Polish Companies

For Polish companies and developers, this dispute has practical significance, since it concerns whether and under what conditions they will be able to freely use cheaper, open Chinese models such as Kimi K3 or DeepSeek's models, without regulatory risk stemming from decisions made in Washington. Potential US export restrictions or sanctions against Chinese AI providers could indirectly affect the availability of these tools in Europe as well.

For now, the position of the US administration remains ambiguous, and the remarks of the head of one of the world's most important technology companies add weight to the side favoring openness, against proponents of tough restrictions on Chinese artificial intelligence.