China has formally urged local companies to avoid purchasing and using Nvidia’s H20 AI processors, particularly for government or security-related projects, citing possible security risks and strategic self-reliance concerns
The H20, a variant based on Nvidia’s older Hopper architecture, was permitted for export under U.S. controls. However, China’s push reflects caution regarding potential backdoors or vulnerabilities — despite Nvidia’s assurances there are none
Government Strategy and Domestic Alternatives
This guidance appears to be another facet of Beijing’s broader campaign to strengthen its semiconductor industry. Chinese agencies are encouraging companies to favor homegrown solutions over foreign chips like the H20
Reports suggest companies received direct inquiries asking:
- Why they choose Nvidia’s chips over domestic alternatives
- Whether security concerns influenced their choice
- If domestic chips could meet their needs instead
Implications for Nvidia and China’s AI Landscape
- China’s Tech Goals: The move underscores the country’s goal of bolstering semiconductor independence and reducing reliance on U.S. technology.
- Revenue Impact: Nvidia’s recovery attempts in China—including a recent deal to pay the U.S. 15% of revenues from H20 sales—could face obstacles if domestic firms shift supply chains Reuters
- Strategic Caution: While not a formal ban, the guidance carries significant weight in China’s regulated environment—effectively narrowing market opportunities for the H20 chip.
Summary:
China is calling on companies—especially those in government or security sectors—to avoid using Nvidia’s H20 AI chips amid rising self-reliance and security concerns. This strategy aligns with broader tech sovereignty initiatives and may affect Nvidia’s business momentum in its second-largest market.
