Policy2026-11-17
TechCrunch AI
US tightens AI chip export controls
The U.S. government has issued an interim rule that requires companies to obtain licenses before exporting AI chips in situations where the hardware could provide compute to China or other restricted countries. The measure is designed to close a loophole: even if physical chips do not ship directly to a restricted destination, foreign cloud services and overseas data centers might still let restricted users access controlled AI compute remotely. By expanding enforcement beyond shipments, the rule reaches cloud providers, chipmakers and multinational enterprises that operate global infrastructure. The immediate effect is a heavier compliance burden. Companies will need to assess customers, workloads and data-center locations more carefully, and some transactions may face delays or licensing uncertainty. The rule could also reshape how AI compute is sold internationally, encouraging stricter partitioning of services and more detailed end-use monitoring. For Washington, the goal is to slow the transfer of advanced AI capabilities to strategic rivals. For industry, the challenge is to comply without fracturing global operations or losing legitimate customers. The interim rule may escalate U.S.-China tensions and prompt retaliation or new efforts to build alternative supply chains. It also signals that export controls are increasingly focused on compute access, not just hardware ownership. Companies affected should expect closer scrutiny of cloud contracts, remote access and overseas subsidiaries. The policy could become a template for future controls as governments treat AI compute as a strategic resource.