Semiconductor powerhouse Nvidia is confronted with unforeseen additional U.S. export restrictions targeting its H20 chips.

Nvidia disclosed in a Tuesday filing that the U.S. government mandated a license for the export of its H20 AI chips to China, specifying this requirement would be permanent due to the potential for the chips' use in Chinese supercomputers.

Nvidia foresees incurring charges amounting to $5.5 billion in its first fiscal quarter of 2026, concluding on April 27, leading to an approximate 6% decline in the company's stock during extended trading.

The H20 is the best AI chip Nvidia is allowed to send to China under the current and past US export rules. Last week, NPR said that CEO Jensen Huang might have avoided new rules about the H20 during a dinner at President Donald Trump's Mar-a-Lago resort. This might have happened because Nvidia promised to put money into AI data centers in the US.

It may not be a mere coincidence that Nvidia declared on Monday its intention to allocate hundreds of millions of dollars over the subsequent four years towards the domestic production of certain AI chips, a commitment critics swiftly noted as lacking specific information.

Several government officials had asked for stricter rules on exporting the H20 chip. They said the chip was supposedly used to train models by DeepSeek, a new AI company in China. This included their R1 "reasoning" model, which caused a lot of confusion in the U.S. AI market in January.

Nvidia refrained from offering any commentary on the matter.