The United States has approved exports of Nvidia’s H200 artificial intelligence chips to China under newly revised licensing rules, marking a shift in U.S. policy toward advanced semiconductor sales to Beijing.
China’s state broadcaster CCTV reported on Wednesday that U.S. authorities had approved Nvidia’s shipments of the H200 to Chinese customers.
The approval follows a rule published on Tuesday in the U.S. Federal Register by the Commerce Department’s Bureau of Industry and Security (BIS), which revised its export licensing policy for semiconductors bound for China. The new rule replaces the previous default denial policy with a case-by-case review system for advanced chips, including Nvidia’s H200 and equivalent models, as well as lower-performance processors.
The move is a decision previously signalled by U.S. President Donald Trump in December and represents a partial reversal of export restrictions imposed under the Biden administration in 2022.
However, the approval comes with conditions designed to limit China’s access and protect U.S. supply chains.
Under the new rules, exporters must demonstrate that sufficient quantities of the relevant chips remain available to U.S. customers and that exports will not divert global foundry capacity away from U.S. end-users for comparable or more advanced products. Chinese recipients must also implement robust security safeguards.
The regulation further states that shipments to China may not exceed 50% of the total volume supplied to U.S. customers for the same products.
In addition, chips exported to China must undergo independent testing by U.S.-based third-party organisations to verify performance specifications. Nvidia must also show that at least twice the volume exported to China is available for the U.S. market and submit documentation detailing the security and protective mechanisms used by Chinese customers.
“These rules look like a compromise designed to manage political disagreements within Washington,” said Jay Goldberg, an analyst at Seaport Research. “On paper they cap exports, but in practice they may be difficult to enforce.”
Nvidia has been ramping up production of the H200 in anticipation of regulatory approval and had hoped to begin shipments before the Lunar New Year in mid-February.
At last week’s CES 2026 conference, Nvidia Chief Executive Jensen Huang and Chief Financial Officer Colette Kress said the company was still awaiting the final outcome of the U.S. review process.
Huang said at the time that demand from China was “extremely strong” and that Nvidia expected to return to the Chinese market.
“The H200 will make a contribution in the future,” Huang said. “The new year is something to look forward to.”
Nvidia declined to comment on specific customers or shipment volumes.
The revised rules mark a significant shift from the technology containment approach pursued by Washington in recent years, which sought to limit China’s access to advanced AI hardware over national security concerns.
Bloomberg reported earlier this week that the new policy not only reflects Trump’s decision to allow limited sales to China but also signals a broader recalibration of U.S. export controls.
However, the entry of Nvidia’s H200 into China still faces regulatory and political uncertainty on the Chinese side.
Nvidia has previously faced scrutiny in China over alleged “security backdoors,” and the country has accelerated efforts to develop domestic chip alternatives under its industrial policy push for technological self-reliance.
Public debate in China has intensified in recent weeks over whether the H200 should be permitted into the market and what impact it could have on domestic semiconductor firms.
At CES, Huang acknowledged that Nvidia’s technological lead in China would not last indefinitely and that continued success would depend on bringing newer chips to the market.
He praised Chinese engineers and entrepreneurs as “world-class” and said China’s technology sector would continue to grow.
“I look forward to competing with you,” Huang said, addressing Chinese chipmakers.
Nvidia’s current U.S. shipments have already shifted toward its newer Blackwell architecture, while its next-generation Rubin chips are expected to enter production later this year. As a result, the H200 is increasingly positioned as a product aimed at overseas markets, including China.
Whether the revised rules will translate into large-scale commercial shipments remains uncertain, analysts said, given the regulatory conditions, geopolitical tensions and China’s parallel push for domestic alternatives.
