BEIJING, January 29 (TMTPost)-- China issued a warning about new signs that the Biden administration will expand its curbs on China’s high-tech industries including artificial intelligence (AI).
Credit：Visual China
China urges the U.S. side not to act contrarily to the laws of sci-tech advancement, earnestly respect the principles of market economy and fair competition, and create favorable conditions for strengthening international AI coordination and cooperation, Chinese Foreign Ministry spokesperson Wang Wenbin commented on the recent report on recent U.S. proposed cloud computing requirements at a regular press on Monday. Wang noted AI development and governance bears on the future of humanity, and requires concerted and coordinated response, not decoupling, severing of supply chains nor fence-building.
The aforementioned report involves a draft rule proposed by the Bureau of Industry and Security, Department of Commerce on Sunday. The propsed regulation asks U.S. Infrastructure as a Service (IaaS) providers to verify the identity of foreign persons that sign up for or maintain accounts that access or utilize U.S. IaaS providers' IaaS products or services (Accounts or Account)—that is, a know-your-customer IKYC) program or Customer Identification Program (CIP). Such program requires cloud serivces like Amazon.com Inc. and Microsoft Corp. to reveal foreign customers’ names and IP addresses. The draft regulation also requires U.S. IaaS providers to submit to the Department a report when a foreign person transacts with the IaaS provider to train a large AI model with potential capabilities that could be used in malicious cyber-enabled activity. The report, at a minimum, must include the identity of the foreign person and the existence of a training run that meets the criteria set forth in the regulation.
Though not elaborating how the American IaaS providers to deal with their customers in China, the draft regulation is evidently deemly as a blow to Chinese companies which may access the data centers and servers crucial to training and hosting AI. Days prior to the draft released, U.S. Commerce Secretary Gina Raimondo said the proposal targets foreign entities including Chinese ones’ access to U.S. data centers to train AI models. "We can't have non-state actors or China or folks who we don’t want accessing our cloud to train their models," Raimondo told Reuters in an interview on Friday. "We use export controls on chips. Those chips are in American cloud data centers so we also have to think about closing down that avenue for potential malicious activity," the Secretary said.
The United States has imposed months ago a new export control on AI offerings to China. The U.S. Department of Commerce introduced a rule entitles 「Implementation of Additional Export Controls: Certain Advanced Computing Items; Supercomputer and Semiconductor End Use; Updates and Corrections」, on October 18. The department said the rule reinforces the controls launched in October 2022 to restrict China’s both purchase and manufacture certain high-end chips critical for military advantage. The agency believes the rules are necessary to maintain the effectiveness of these controls, close loopholes, and ensure they remain durable. The goal is to limit China’s 「access to advanced semiconductors that could fuel breakthroughs in artificial intelligence and sophisticated computers,」 said Raimondo.
China opposes US politicizing, instrumentalizing and weaponizing trade and tech issues, and will closely follow the developments and firmly safeguard our legitimate rights and interests, Chinese Foreign Ministry spokesperson Wang Wenbin said in October. Arbitrarily placing curbs or forcibly seeking decoupling to serve political agenda violates the principles of market economy and fair competition, undermines the international economic and trading order, disrupts and destabilizes global industrial and supply chains and will eventually hurt the interests of the whole world, Wang said.
The rule was supposed to come into effect following a 30-day public comment period. But Nvidia Corporation, the AI chip leader,disclosed on October 25 that the U.S. government informed the licensing requirements of the rule applicable to products having a 「total processing performance」 of 4800 or more and designed or marketed for data centers, is effective immediately.
TMTPost learned in November month that Nvidia will launch a new set of chips under the name of HGX p0, L20 PCle, L2 PClel for China market, though all of them are weakened version. Compared with p00, the HGX p0 has limitations on bandwidth and computing speed, and its overall computing power is about 80% less than that of p00, namely, the comprehensive computing performance of p0 is equal to 20% of p00. Nvidia was reported late November that it informed customers it is delaying the launch of p0, the most powerful of three new chips for the Chinese market until the first quarter of 2024, due to issues that took place in server manufacturers’ integrating the chip.
