The Chinese government is preparing to authorize artificial intelligence companies to purchase Nvidia's H200 graphics processing units, according to a report by The Straits Times citing Information magazine.

The H200 is part of Nvidia's Hopper architecture, designed specifically for training and running large-scale AI models, and remains one of the most powerful chips available for export to China under current US restrictions.

The decision marks a notable shift in the regulatory landscape for semiconductor trade between the two nations.

While Beijing has been systematically advancing a strategy to reduce reliance on US technology and displace firms like Nvidia, Anthropic, and OpenAI from its domestic AI infrastructure, this move suggests a pragmatic approach to immediate hardware needs.

The report indicates that authorities are still determining the specific volume of chips that will be permitted for import, leaving the scale of the potential market open to interpretation.

For Nvidia, the ability to sell H200 units in China represents a significant revenue stream, given the country's massive demand for AI computing power.