Beijing (Reuters) – Nearly every major tech business operating public clouds or cutting-edge artificial intelligence training modules in China is expected to be impacted by a U.S. directive banning sale of certain sophisticated semiconductors to that nation, according to analysts.
Nvidia Corp, a chip manufacturer, said on Wednesday that American authorities had ordered it to cease sending two of the best processing processors for AI development to China.
Additionally, Advanced Micro Devices said that it had acquired additional licencing restrictions that would prevent the shipment of its cutting-edge AI chip, designated MI250, to China.
A representative for the Chinese Commerce Ministry, Shu Jueting, said on Thursday that Beijing opposes the regulations because they threaten to disrupt global supply chains and violate the rights of Chinese businesses.
The directives highlight escalating tensions between the US and China for access to cutting-edge microprocessor technology.
According to Jay Goldberg, CEO of D2D Advisory, a finance and strategy consulting business, “We’re moving from barring specific U.S. corporations from providing to a given company, as was the case with Huawei, to banning certain U.S. goods from selling to China generally.”
The worst-case scenario, according to analysts at Jefferies, would be if Washington expanded the embargo to prevent contract chipmakers like Taiwan Semiconductor Manufacturing Co. and Samsung from producing semiconductors for Chinese chip designers.
“We are not there yet, and the U.S. will probably assess the efficacy of each gradual move before considering dramatic measures,” it said.
Market observers predict that a large number of Chinese IT heavyweights, including Alibaba Group Holding Ltd, Tencent Holdings Ltd, Baidu Inc., and Huawei Technologies Co Ltd., would be negatively impacted by the most recent ban.
Affected firms might employ several low-end chips to mimic the processing capacity of the outlawed, high-end chips, or depend on cloud services like Alphabet Inc.’s Google or Amazon.com Inc.’s AWS to create AI software and ship it back to China, according to Jefferies.
According to a former senior employee of AMD in China, the limits won’t prevent Chinese tech firms from furthering their AI research, but they will increase costs and decrease efficiency in the near run.
“It affects resources. Even if it slows them down, they will continue work on the same tasks and advance “He informed Reuters.
Requests for comment from Reuters were not immediately answered by Alibaba, Tencent, or Baidu. Huawei opted not to respond.
The AI and machine learning applications that the Nvidia and AMD processors targeted by Washington are employed, primarily for creating training modules for tasks like natural language processing
The military could find these components helpful for constructing weapons and simulating bombs.
Few Chinese businesses, according to Goldberg at D2D, could produce AMD and Nvidia replacement chips rapidly, and the limits would probably encourage greater financing for local semiconductor startups to close the gap with American companies.
Many firms with the goal of competing with Nvidia and AMD are based in China. Former employees of these organisations started a lot of them, although few have grown to a significant size.
Hygon Information Technology Co. and Loongson Technology Corp., two Chinese manufacturers of AI chips, saw their shares soar on Thursday, increasing 10% and 6%, respectively.
A 7nm device, which experts believe represents significant advancement for China’s semiconductor industry, was introduced last week by Biren, a business created by former employees of Nvidia and Alibaba.
The order books of the several dozen Chinese semiconductor firms developing various types of AI accelerators will soon be filled, according to Goldberg.