The US government has introduced and immediately started imposing a new rule that would restrict the export of computer chips used for supercomputers and artificial intelligence to Russia and China. NVIDIA has revealed in a SEC filing that the government has informed the company of the new license requirement affecting one of its current (A100) and one of its upcoming (H100) GPUs, which were designed to speed up machine learning tasks.
The government apparently indicated to NVIDIA that the new license requirement will address the risk that the affected products may be used for military purposes by either country. As The New York Times notes, AI and machine learning are used for various applications that include weapons development and surveillance. A US Department of Commerce spokesperson told Reuters:
“While we are not in a position to outline specific policy changes at this time, we are taking a comprehensive approach to implement additional actions necessary related to technologies, end-uses, and end-users to protect US national security and foreign policy interests.”
This requirement comes on the heels of another export control rule issued in mid-August that would prevent China from accessing advanced chip design software necessary to produce the next-generation processors.
NVIDIA has admitted in its filing that the restriction might hamper its ability to complete the development of the H100 chip. It doesn’t sell its products in Russia, but it’s expecting to earn around $400 million from sales to Chinese companies for the third fiscal quarter. That’s money it could use for the development of future products. The chipmaker could lose all those potential sales if its Chinese clients refuse to purchase its alternative offerings or if the government refuses to grant licenses to its biggest customers.
As for AMD, a spokesperson told Reuters that while the new license requirement will prevent the company from exporting its MI250 chips to China, it doesn’t believe it will have a big impact on its business.