U.S. — Nvidia showed a "local AI initiative with optimizations for top open models," including DeepSeek's V4 Flash and Alibaba's Qwen 3.8, alongside systems developed by Google and Nvidia itself. The company stated that the U.S. is considering restrictions that would limit its ability to support "third-party applications and models built on open-source foundation models originating in China."

An SEC filing linked to Nvidia's second-quarter earnings warned of business risks from potential White House restrictions on AI developed in China. The filing indicated that any "regulatory control or other restriction that limits our ability to provide products and services" supporting apps built on models "such as DeepSeek, Qwen or Kimi" could have a material impact on business.

In its second-quarter SEC filing, Nvidia stated that restrictions on supporting applications built on Chinese open-source foundation models could reduce its revenue by up to $5 billion in the next fiscal quarter, depending on geopolitical conditions. Nvidia shares rose after the company reported better-than-expected second-quarter results and issued revenue guidance that topped estimates.

A Nvidia employee told CNBC that providing support for models worldwide allows developers to build on the American tech stack. The employee said developers using popular American and Chinese models will choose the stack that the model is optimized for, making the need to optimize for the American stack critical.

Daniel Remler, senior fellow at the Center for a New American Security, told CNBC that if Chinese AI technology becomes the default for developing countries, those countries may be more likely to align themselves politically with Beijing and Chinese AI companies get a beachhead in their markets. A White House official told CNBC that "unless officially announced by the Administration, any reporting about tariffs should be regarded as baseless speculation."

Chinese chipmakers including Huawei and Alibaba have announced their own optimizations for open models, including the DeepSeek and Qwen series, in recent months. In 2023, Huawei launched the Ascend 910C chip, which was optimized for running DeepSeek and other Chinese AI models, signaling an effort to compete directly with Nvidia in the AI hardware space. DeepSeek is owned and funded by High-Flyer, a Chinese hedge fund.

Why It Matters

The debate over supporting Chinese AI models centers on maintaining U.S. technological influence in global markets. Microsoft, Meta, Palantir and more than 20 other companies joined Nvidia in issuing a statement in July urging policymakers to avoid premature restrictions on open-weight models.

Nvidia's stock has risen more than 10% so far in 2026. In 2025, Nvidia CEO Jensen Huang personally visited the White House in July and August to secure export licenses for the H20 chip, which is used for Chinese AI models, as part of a deal that would see 15% of China sales of the H20 go to the U.S. government. The U.S. government has not officially announced any new tariffs or restrictions on AI models developed in China as of August 2026, though officials have warned that such measures are under consideration.

Timeline

Nvidia said in August that updates to its software made it "easy to cluster multiple DGX Spark systems together," which was important for running Chinese models like Z.ai's GLM 5.2 and DeepSeek V4 Flash. Nvidia announced in August that it was optimizing its hardware stack, including day-zero support on its RTX GPU systems for Qwen3.8-27B.

What's New

Forrester reported that the number of developers in China working on open-source AI models grew by 37% between 2023 and 2025, reinforcing Nvidia’s argument about the importance of maintaining influence in the region. Research titled "Strategic Fit and Growth Logic: NVIDIA in the AI Era" was published in 2025 in Proceedings of the 1st International Conference on Branding, Marketing, and Media Communication.

A 2025 internal memo from the U.S. Department of Defense outlined concerns that Chinese AI models could be used in military applications, raising questions about the security implications of supporting such models through foreign hardware like Nvidia’s chips. The Center for a New American Security (CNAS), where Daniel Remler serves as a senior fellow, has previously issued reports showing the strategic risks of Chinese AI dominance in developing markets, including concerns about political alignment with Beijing.