What Is This New Rule Exactly?
The regulation, issued by the Commerce Department's Bureau of Industry and Security (BIS), is an 'Interim Final Rule' (IFR) that took effect immediately upon its announcement. Its primary goal is to close a major loophole in technology export controls.
Previously, restrictions focused on the physical shipment of advanced AI chips, such as those made by Nvidia. This new rule extends controls to cover 'remote access'. It means companies in restricted nations can no longer simply rent computing power from data centers in third-party countries to train their AI models. The rule requires a US license to export advanced AI chips to any data center in the world if there's knowledge it will supply significant computing power to companies headquartered in China, Macau, or other restricted nations.
Why Target Remote Access Now?
US policymakers grew concerned that their existing chip export bans were becoming less effective. While it was illegal to sell a high-performance AI chip directly to a Chinese firm, that same firm could legally rent time on a server packed with those very chips located in, for example, Malaysia or Singapore. This allowed them to continue developing sophisticated AI models that could have military applications, undermining the original intent of the export controls. The Commerce Department stated the rule is designed to ensure that 'existing restrictions on access to sensitive American technology cannot be circumvented through remote computing arrangements'. The move aligns the rules for remote access with the existing controls on physical hardware.
Who Does This Rule Affect Most?
The impact is widespread. Major US cloud providers who operate data centers globally will need to implement robust 'Know-Your-Customer' (KYC) checks to ensure their computing power isn't being rented by restricted entities. Data center operators worldwide that use US chip technology are also affected. The rule specifically adds new licensing requirements for advanced chip shipments to five countries identified as having a 'higher-than-normal risk of diversion': Brazil, Indonesia, Malaysia, Singapore, and Thailand. This puts a significant compliance burden on chip exporters and data center operators to track not just who they sell to, but who their customers' customers are. Chinese AI companies are the primary target and will find it much harder to access the high-end computing power needed for advanced AI development.
What Happens Next for Tech Companies?
The rule is 'interim,' which means it's in effect now but is also open for public comment for 60 days following its official publication, expected around November 19. This gives the industry a chance to provide feedback, though major changes are not guaranteed. Existing data centers that are already providing services to restricted companies have a 90-day grace period to obtain the necessary licenses or cease the service. For tech firms, particularly cloud service providers and chipmakers, the immediate next step is a massive compliance and due diligence effort. They must scrutinize their client lists and their ownership structures to avoid violating the new, more stringent regulations. The rule also formally rescinds a previous, unenforced regulation from January 2025 known as the 'AI Diffusion Rule'.
















