Semiconductor powerhouse Nvidia is confronting unforeseen new U.S. export regulations concerning its H20 chips.

In a filing on Tuesday, Nvidia disclosed that the U.S. government has imposed an indefinite licensing requirement for the export of its H20 AI chips to China, citing the potential risk of their deployment in a Chinese supercomputer.

Nvidia foresees approximately $5.5 billion in associated costs during its first quarter of the 2026 fiscal year, concluding on April 27, with the company's stock experiencing a decline of roughly 6% in after-hours trading.

The H20 is the most advanced AI chip that Nvidia is allowed to send to China under the export rules the U.S. has now and had before. Last week, NPR reported that CEO Jensen Huang may have avoided new rules about the H20 during a dinner at President Donald Trump's Mar-a-Lago resort. He might have done this partly by promising that Nvidia would put money into AI data centers in the U.S.

Perhaps tellingly, Nvidia disclosed on Monday its commitment to investing hundreds of millions of dollars over the ensuing four years in manufacturing certain AI chips domestically, a move analysts promptly highlighted as lacking specific information.

Several government officials had asked for stricter rules on exporting the H20 chip. This was because it was reportedly used to train models by the Chinese AI company DeepSeek, including their R1 'reasoning' model. This model caused a lot of disruption in the U.S. AI market in January.

Nvidia opted to refrain from offering any commentary.