Technology
Nvidia optimizes hardware for Chinese AI models while pointing to potential White House regulatory risks
The chipmaker is enhancing support for open-source systems like DeepSeek and Qwen to retain global developers despite potential U.S. restrictions.
The short version
- Nvidia is tailoring its hardware and software to support open-source Chinese AI models, including Alibaba's Qwen and DeepSeek's V4 Flash, to compete against Chinese chip manufacturers like Huawei.
- The company warned in an SEC filing that potential White House restrictions on supporting Chinese-origin open-source models could negatively impact its business.
- A White House official dismissed reports of prospective tariffs on U.S. semiconductors as baseless speculation absent an official announcement.
Key facts
- Nvidia highlighted optimizations for open-source AI models developed in China alongside its own and Google's systems as part of a local AI initiative.[CNBC]
- In a second-quarter SEC filing, Nvidia warned that prospective U.S. regulatory controls limiting support for third-party applications built on Chinese models could harm its operations.[CNBC]
- Chinese hardware makers, including Huawei and Alibaba, have recently rolled out their own optimizations for DeepSeek and Qwen open-source models.[CNBC]
- Nvidia shares rose after posting second-quarter financial results and future revenue guidance that exceeded analyst estimates.[CNBC]
- In July, Nvidia alongside Microsoft, Meta, Palantir, and more than 20 other companies issued a public statement urging policymakers against imposing premature restrictions on open-weight models.[CNBC]
What remains uncertain
- Whether the White House will enact reported tariffs on U.S. semiconductors or formally restrict U.S. tech companies from supporting Chinese open-source AI models.[CNBC]