US splits world into three tiers for AI chip access

While close US allies get unrestricted AI chip access, the rest of the world has numerical limits.

On Monday, the US government announced a new round of regulations on global AI chip exports, dividing the world into roughly three tiers of access. The rules create quotas for about 120 countries and allow unrestricted access for 18 close US allies while maintaining existing bans on China, Russia, Iran, and North Korea.

AI-accelerating GPU chips, like those manufactured by Nvidia, currently serve as the backbone for a wide variety of AI model deployments, such as chatbots like ChatGPT, AI video generators, self-driving cars, weapons targeting systems, and much more. The Biden administration fears that those chips could be used to undermine US national security.

According to the White House, "In the wrong hands, powerful AI systems have the potential to exacerbate significant national security risks, including by enabling the development of weapons of mass destruction, supporting powerful offensive cyber operations, and aiding human rights abuses."

The new rules build on previous chip controls from September 2022 and October 2023. The regulations will take effect in 120 days, extending into the incoming Trump administration.

A game of numbers

The new regulations set specific numerical limits on AI chip exports. While first-tier countries (the 18 key US allies) face no restrictions, countries in the second tier can receive up to 50,000 so-called "advanced computing chips," with the possibility to double that cap to 100,000 if they sign technology security agreements with the US.

For most buyers, orders of up to 1,700 advanced chips will not require licenses or count against these national caps—a policy designed to speed up purchases by universities, medical institutions, and research organizations.

The Commerce Department also announced two new verification systems for trusted buyers. Companies based in allied nations can apply for "Universal Verified End User" (UVEU) status, allowing them to deploy up to 7 percent of their global AI computing capacity in individual countries. Companies outside restricted nations can seek "National Verified End User" status to purchase computing power equivalent to 320,000 advanced GPUs over two years.

According to Reuters, cloud service providers received special consideration under the rules. Companies like Microsoft, Google, and Amazon can apply for global authorizations to build data centers, bypassing country quotas after meeting security and human rights requirements. However, US-headquartered providers must maintain at least 50 percent of their AI computing power inside the US, with no more than 25 percent outside tier-one countries.

The rules also introduce controls on AI model weights—the neural network files that determine how AI models work. The regulations restrict transfers of these weights for "closed-weight" models to non-trusted actors while allowing unrestricted sharing of open-weight model information.

Nvidia responds

Nvidia wasted no time reacting to the news on Monday, calling the regulations "sweeping overreach" that restricts technology found in common gaming PCs. Others in the tech industry are not pleased with the regulations, either. Last week, Oracle warned that chip export restrictions could push the global AI and GPU market toward Chinese competitors.

Commerce Secretary Gina Raimondo disagrees with that position, according to Reuters, and connected the new chip export rules to maintaining US technological leadership. "The US leads AI now—both AI development and AI chip design, and it's critical that we keep it that way," she said.

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+

RIP Matrix | Farewell my friend