Nvidia is ramping up support for Chinese artificial intelligence models in a bid to beat domestic chipmakers in optimizing system hardware. The move comes amid growing concerns among US lawmakers about the adoption of Chinese AI models by US companies.
The chip giant on Wednesday highlighted “local AI initiatives with optimizations for top open models” such as DeepSeek’s V4 Flash and Alibaba’s Qwen 3.8, in addition to systems developed by Google and Nvidia itself.
But SEC filings related to the company’s second-quarter results also warned of business risks from potential White House restrictions on AI developed in China.
China’s AI models are set to leapfrog in their capabilities in 2026, and their increasing adoption by developers around the world, including the United States, is prompting a wave of anxiety from lawmakers in Washington.
Access to open AI models, of which China is the most capable, has become a key flashpoint in the escalating competition for technological superiority between the United States and China. In contrast to the dominant closed systems created by OpenAI and Anthropic, open AI models can be modified and self-hosted.
Both China and the United States are looking to use political influence to their advantage after reports surfaced on Thursday that President Donald Trump’s administration is considering new tariffs on American-made semiconductors.
A White House official told CNBC that “unless formally announced by the administration, any reports of tariffs should be considered baseless speculation.”
The U.S. is also considering restrictions that would limit Nvidia’s ability to support “third-party applications and models built on open source foundational models originating in China,” the company said, repeating messages from past financial results.
It warned that “regulatory controls or other restrictions that limit our ability to offer products or services” that support apps built on models “such as DeepSeek, Qwen, or Kim” could have a material impact on the business.
Nvidia stock rose Thursday after the company reported better-than-expected second-quarter results and issued a better-than-expected revenue outlook. The tech giant, which has a near monopoly on cutting-edge AI chips, has seen its stock price rise more than 10% by 2026.
nvidia stock
NVIDIA is not alone in campaigning against U.S. government intervention in an open model that companies can download, modify, and self-host.
Microsoft, Meta, Palantir and more than 20 other companies joined the semiconductor giant in issuing a statement in July urging policymakers to avoid “premature restrictions” on indiscriminate weight models.
“Providing support for models from around the world allows developers to build on the US technology stack,” an Nvidia employee who requested anonymity because he was not authorized to share the information told CNBC.
“The need to optimize for the US stack is important because developers using popular models from the US and China will choose the stack for which that model is optimized,” they added.
Chinese chipmakers such as Huawei and Alibaba, which produce the Ascend series of processors, have announced their own optimizations for open models such as the DeepSeek and Qwen series in recent months.
“Developers will play a critical role in building a winning AI ecosystem,” said an Nvidia employee. “China has the world’s largest developer population and is developing open source foundational models. Any model should work best on the U.S. technology stack, and countries around the world will choose the U.S..”
“Increasing influence”
China’s AI models are increasingly more capable and generally cheaper than their American alternatives, raising concerns that they could become the default for developing countries.
“If Chinese AI technology becomes the default for developing countries, those countries will be more likely to align themselves politically with China, potentially giving Chinese AI companies a beachhead in their markets,” Daniel Lemler, senior fellow in the Technology and National Security Program at the Center for a New American Security (CNAS) think tank, told CNBC.
Forrester vice president and principal analyst Charlie Dai told CNBC that NVIDIA’s optimization of hardware for China’s AI models reflects the company’s “growing influence” within the global AI ecosystem.
“By optimizing platforms like DeepSeek and Qwen, Nvidia is strengthening its position as the preferred infrastructure layer for AI developers and enterprises, regardless of where their leading models originate,” he added.
Nvidia announced in August that it was making optimizations to its hardware stack, including “day zero support” on the Qwen3.8-27B RTX GPU system. The company said in the same month that a software update made it possible to “easily cluster multiple DGX Spark systems,” which was important for running Chinese models such as Z.ai’s GLM 5.2 and DeepSeek V4 Flash.
