Many Chinese technology companies are ready to develop the next advanced chip for artificial intelligence (AI) applications to meet demand before the next US sanctions package takes effect from November 16.
According to Xataka, the next US sanctions package aimed at China makes Nvidia unable to sell its A800 and H800 artificial intelligence chips in China. Even the most powerful GeForce RTX 4090 GPU for the consumer market is affected. This not only impacts Nvidia now but also in the future when they may lose many Chinese customers after the ban is lifted, as commented by Nvidia CEO Jensen Huang when he said: “If China It is impossible to buy GPUs for artificial intelligence from the US, they will simply produce them themselves.”
In the short term, Nvidia’s decline in China is especially beneficial for Huawei – the company already has Ascend AI chips available to try to fill the gap in the market that Nvidia must leave in about 10 days. In addition, it is impossible not to mention MetaX, Alibaba, Biren Technology, Moore Threads, Innosilicon, Zhaoxin, Iluvatar CoreX, DenglinAI, or Vast AI Tech that are developing advanced chips for AI applications.
Recently, several reputable media outlets in Asia reported that several Chinese companies specializing in graphics hardware and AI such as MetaX or Moore Threads have recruited engineers from AMD and Nvidia to serve their tasks. which they aim for.
magazine Everything gained more attention after an article published in Nature by some Chinese scientists from Tsinghua University. According to the article, researchers have described in detail the architecture of a chip for computer vision with strong power and extremely fast operation, at least based on theory. Not talking about graphics power, scientists say that the chip’s computer vision task is 3.7 times faster than Nvidia’s A100 chip when both face the same intensive task of image classification. image.
Tsinghua University is one of the most prestigious scientific institutions in China, so the article’s content is completely valid. The most interesting thing about this chip is that its theoretical performance is higher than Nvidia’s A100 GPU.
The chip is manufactured using 180nm CMOS integration technology – a very mature lithography technique that is much less complex than the most advanced technologies currently being produced. It is difficult to determine to what extent their performance would improve if much more advanced lithography techniques were used in the manufacturing process. Of course, these integrated circuits cannot currently be produced on a large scale, so they still have to overcome the challenge of leaving the laboratory and entering mass production.