Urgent.News

What's breaking now, across thousands of outlets.

AI

U.S. Extends AI Controls to China's Overseas Computing Access, Raising Questions for Korean HBM Demand

The United States is expanding export controls on advanced artificial intelligence (AI) chips to cover computing services provided by overseas data centers, targeting a route through which Chinese companies can access U.S. AI technology without importing the chips directly. The shift extends regulatory scrutiny beyond the physical destination of semiconductors to the companies ultimately using…

The United States is expanding export controls on advanced artificial intelligence (AI) chips to cover computing services offered by overseas data centers. This move aims to restrict a method through which Chinese entities can access U.S. AI technology without directly importing semiconductor chips. The regulation broadens the scope of scrutiny from the physical destination of semiconductors to the companies ultimately using their computing power.

The U.S. Commerce Department's Bureau of Industry and Security (BIS) announced the interim final rule on November 16, introducing supplementary licensing requirements for advanced AI chips headed towards Singapore, Malaysia, Thailand, Indonesia, and Brazil. The measures do not target South Korean high-bandwidth memory (HBM) directly; however, restrictions affecting U.S. AI accelerators and overseas cloud services might influence GPU procurement, server deployment timetables, and customer contracts within a supply chain encompassing Samsung Electronics and SK hynix.

Traditionally, semiconductor export controls have primarily focused on preventing advanced chips from reaching restricted countries or entities. However, cloud computing introduces complexity as the location of hardware and its users can diverge. For example, U.S.-designed GPUs in a Malaysian data center could potentially be accessed remotely by a Chinese firm.

The recent U.S. measures aim to tackle this indirect access avenue. Reports indicate that Chinese AI companies are seeking to secure overseas Nvidia-based computing resources due to apprehensions about increased American restrictions.

The new rule applies to overseas data centers offering substantial computing capacity to companies based in so-called Country Group D:5, or in Macau. It also factors in indirect ownership and control, making the ultimate customer's identity a crucial compliance aspect. Therefore, the rule's applicability relies not only on the data center's country of operation but also on additional factors such as computing capacity, customer location, ownership structure, and service nature. The regulations' applicability hinges on more than just the geographical location of a data center.

Written by urgent.news from Korea IT Times's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at koreaittimes.com →

More in AI

More from Tuesday 17 November →