Nvidia plans to release a downgraded version of its H20 artificial intelligence chip for China in July, following US export restrictions on the original model, reported Reuters citing sources.

Two of the sources revealed that the US chipmaker has informed key Chinese clients, including top cloud service providers, of plans to release the modified H20 in July 2025.

The adjustment forms part of Nvidia’s efforts to maintain operations in China.

Previously, the H20 was Nvidia’s most advanced AI chip available to Chinese buyers.

However, US officials last month informed the company that the chip would now require an export licence, leading to its withdrawal from the market.

In response, Nvidia has established new technical specifications that will significantly scale back the capabilities of the chip, particularly by reducing its memory capacity, one source told the news agency.

GlobalData Strategic Intelligence

US Tariffs are shifting - will you react or anticipate?

Don’t let policy changes catch you off guard. Stay proactive with real-time data and expert analysis.

By GlobalData

Another source said that some customers may be able to adjust the chip’s module configuration to tailor its performance to their specific needs.

Nvidia declined to comment, and the US Commerce Department did not immediately respond to a request for comment, Reuters said.

China represents a significant portion of Nvidia’s business, contributing $17bn in revenue, which is 13% of the company’s total sales, in the fiscal year ending 26 January.

Since 2022, the US has imposed restrictions on the export of Nvidia’s most advanced chips to China , citing potential military uses as a concern.

The H20 chip was introduced in late 2023, after the US implemented stricter export rules.

Major Chinese tech firms such as Tencent, Alibaba and ByteDance increased their orders for the chip in early 2024, driven by demand from AI-focused companies including startup DeepSeek.

In April 2025, Reuters reported that Nvidia received orders worth $18bn for the H20 chip since January 2025.

In April 2025, Nvidia said it plans to manufacture up to $500bn of AI supercomputers within the US over the next four years.