US Empowers Tech Giants to Control Global Access to AI Chips

The United States government is planning to empower major companies like Google and Microsoft to act as gatekeepers for accessing highly coveted AI chips on a global scale. This development comes as part of a broader strategy to curb China's access to advanced technologies that could enhance its military and cyber capabilities.

According to sources familiar with the draft plan, which is expected to be unveiled soon, these companies would need to adhere to stringent requirements. This includes reporting key information to the US government and actively blocking Chinese firms from gaining access to AI chips.

As a result, Google and Microsoft will be able to provide AI services through cloud platforms internationally without needing additional licenses. These new regulations are being prepared in the final days of the Biden administration, as officials seek to streamline the export approval process while simultaneously preventing unauthorized users from obtaining AI technology.

The US government has raised concerns that China could leverage advanced AI capabilities to strengthen its military, launch significant cyberattacks, or even develop biological weapons. In response, the Department of Commerce has declined to comment on the specifics or the timing of the new rules, noting that plans may still change.

When approached for comments, both Google and Microsoft did not provide immediate responses.

This initiative seems to take inspiration from a national security agreement Microsoft established earlier this year with the US government. This agreement permitted Microsoft to provide AI technology to the Emirati company G42.

Furthermore, under the proposed regulations, other companies without gatekeeper status could still compete for licenses to import a smaller quantity of cutting-edge AI chips made by companies like Nvidia and AMD in different countries.

Nvidia, recognized for creating some of the world's most powerful AI chips, has expressed its willingness to cooperate with the US administration regarding these new regulations. Meanwhile, AMD has not yet responded to inquiries about this matter.

Significantly, 19 allied countries, including the Netherlands, Japan, and Taiwan, will not face these restrictions and will have unrestricted access to AI chips and their capabilities, according to two sources. Countries under nuclear embargoes, such as Russia, China, Iran, and Venezuela, will remain barred from acquiring US-made AI semiconductors.

Experts caution that these caps could provoke dissatisfaction among various nations. Geoffrey Gertz, a former official in the White House, remarked that a global program imposing country limits might raise concerns among US allies who are uneasy about the US taking unilateral decisions on who can access critical advanced chips.

The US government is in the final stages of reviewing a draft regulation termed "Artificial Intelligence Diffusion," suggesting an imminent announcement. Several sources indicate that this regulation relates to the proposed AI chip restrictions.

The Information Technology Industry Council, which represents companies like AMD and Google, is worried that the Biden administration is hastening the rollout of complex regulations without sufficient industry consultation, potentially leading to negative outcomes. Naomi Wilson, the council's senior vice president for Asia and global trade policy, commented that the changes could considerably widen the scope of export controls with far-reaching global implications.

This initiative builds upon a program launched in September that permits pre-approved data centers outside the US to receive AI chips without needing licenses. To qualify for this status, these data centers must provide detailed information about their customers, business activities, access limitations, and cybersecurity measures.

AI, Chips, Regulations