U.S. to Empower Companies as Gatekeepers for AI Chips

By Alexandra Alper and Karen Freifeld

WASHINGTON (Reuters) – In its latest move against Beijing, the U.S. plans to allow companies such as Google and Microsoft to act as global gatekeepers for highly sought-after access to AI chips, according to two sources familiar with the draft plan.

Under this scheme, expected to be released soon, these companies will need to comply with strict requirements, including reporting key information to the U.S. government and blocking Chinese access to AI chips. This would enable them to offer artificial intelligence capabilities within the cloud overseas without a license, the sources added.

The new rules come as officials scramble in the final days of the Biden administration to streamline the approval process for AI chip exports while preventing bad actors from gaining access. The U.S. is concerned that China could leverage AI technology to enhance its military, launch cyberattacks, or develop bioweapons.

The Commerce Department has refrained from commenting on the content and timing of the new regulations, and sources suggest that the administration's plans are still subject to change.

Google and Microsoft have not yet responded to requests for comment.

This measure aligns with a national security agreement Microsoft reached with the U.S. government in April, allowing it to supply AI technology to Emirati firm G42.

According to the draft rules, companies beyond those designated as gatekeepers will also have the opportunity to compete for licenses to import a limited number of high-end AI chips from Nvidia and AMD in various countries.

Nvidia, a leading producer of powerful AI chips, stated its readiness to collaborate with the administration on the new rules. AMD has yet to respond to comment requests.

Countries such as the Netherlands and Japan, plus Taiwan, will be exempted from the caps, maintaining unlimited access to AI chips and their capabilities. Conversely, countries under nuclear embargo include Russia, China, Iran, and Venezuela, which are already prohibited from acquiring U.S. AI semiconductors and will continue to face those restrictions.

These caps may create tensions with some nations. Geoffrey Gertz, a former White House official, warns that a global program of country caps could provoke significant concerns among U.S. partners and allies, who are apprehensive about the U.S. acting unilaterally regarding access to critical advanced chips for AI.

The U.S. government is currently reviewing a proposed "Artificial Intelligence Diffusion" rule drafted by the Commerce Department, indicating progress towards publication. Three sources noted that the government posting has referenced these AI caps.

However, the Information Technology Industry Council, which includes members like AMD and Google, has expressed concern over the Biden administration's hurried rollout of this complex rule without sufficient industry input, urging caution to prevent unintended consequences. Naomi Wilson, the council's senior vice president of Asia and global trade policy, stated, "If reports are accurate, such changes would dramatically expand the scope of export controls and have significant global implications."

These rules build upon a framework introduced in September that allows pre-approved overseas data centers to receive AI chips without needing a license. To achieve this status, data centers must share information about their customers, business activities, access restrictions, and cybersecurity measures.