US Nvidia ban reportedly bypassed by China's military and AI institutes

NVIDIA, the world’s leading company in designing chips for artificial intelligence tools, has an impressive line up of semiconductors but they have been banned by the US for the last year and for a major part of 2022 from being supplied to China. But, as it turns out, over the preceding year, Chinese military entities, state affiliated artificial intelligence research institutes, and academic institutions have procured restricted batches of NVIDIA semiconductors that are prohibited from export to the dragon country.

These transactions underscore the challenges faced by Washington in its attempts, via imposed bans, to sever China's access to advanced US chips, as revealed by an examination of tender documents conducted by . These chips are instrumental in propelling advancements in artificial intelligence (AI) and sophisticated computing systems, particularly for military applications, which is why the US has been raining hellfire (in ways of sanctions) on the Chinese AI development front.

The sanctions are hurting NVIDIA The company hit a major snag in supplying its chips to China, whose 90% AI chip market was dependent on NVIDIA. Despite the constraints, the supply of high end US chips remains legally permissible within China. The tender documents with Reuters indicate that numerous Chinese entities have successfully acquired Nvidia semiconductors after the implementation of restrictions.

These include the A100 and the more potent H100 chip, subject to export bans to China and Hong Kong in September 2022, along with the subsequently developed A800 and H800 chips, which were banned in October of the same year. In response to , a spokesperson of NVIDIA said, "If we learn that a customer has made an unlawful resale to third parties, we'll take immediate and appropriate action." In the comprehensive review carried out by Reuters, spanning over 100 tenders, it was revealed that state entities have actively procured Nvidia's A100 chips.

A tender claims to have created a more powerful chip One that stood out among them was Tsinghua University, called China's Massachusetts Institute of Technology, which procured two H100 chips and around 80 A100 chips since the 2022 ban. Recently, Chinese scientists from the same university claimed to have developed a chip more powerful and efficient than NVIDIA’s A100 chip, one of the most commercially used AI chips in the world today.

Called an all analog chip combining electronic and light computing (ACCEL), the Chinese Marvell chip can perform 74.8 quadrillion operations per second using just one watt of power, reported NVIDIA also disclosed a delay in the release of its cutting edge AI chip, the HGX H20, specifically made for China.

Originally scheduled for a November 2023 launch, the H20 chip, touted as the most potent among three China specific chips developed by NVIDIA to align with new US export regulations, is now anticipated to debut in the first quarter of 2024. The company communicated the postponement to its Chinese clientele, indicating that the H20 chip's launch is now expected in February or March of the coming year..