Source-reported · Independent verification status is not recorded for this article; claims remain attributed to their sources.
U.S. restricts remote access to advanced AI chips, tightening control over who can run powerful models
The U.S. Commerce Department has moved to limit remote access to advanced American AI chips, aiming to close a loophole that let foreign users tap high‑end hardware via the cloud. The rule lands amid a broader political push to control advanced chip use and secure AI supply chains.
The United States is moving to curb who can remotely use its most advanced AI hardware, issuing a rule that restricts access to high‑end American chips when they are offered over networks rather than shipped abroad.
The U.S. Commerce Department announced the measure, which targets arrangements where users outside the country could effectively rent time on powerful U.S. AI chips hosted in domestic data centers. Those chips are widely used to train and run large‑scale artificial intelligence models.
Officials have not yet released the full technical text, but the stated aim is to bring remote access under the same kind of control as physical exports. In other words, a chip’s physical location in the United States should no longer automatically mean it can be used from anywhere in the world.
The rule arrives as Washington lawmakers and regulators debate tighter limits on advanced chip exports, with a particular focus on China. Nvidia CEO Jensen Huang’s sudden resignation has sharpened that debate, as rival chipmakers and AI firms move to influence how far new restrictions should go.
According to recent reporting, AMD and Qualcomm are working to expand their presence in Washington just as Congress and agencies consider new rules. At the same time, major AI developers OpenAI and Anthropic are backing legislation that would restrict China’s access to advanced chips and manufacturing tools.
In parallel to these domestic moves, the U.S. and nine international partners have signed the San Francisco Declaration, a joint commitment to strengthen the security of AI technology supply chains and protect critical technologies and infrastructure. The new Commerce rule fits into that broader push to manage risks around who can access the computing power needed for cutting‑edge AI.
For big cloud providers, the rule means tighter compliance systems to verify customers and monitor where powerful AI workloads originate. Research groups and startups abroad may find it harder or slower to tap U.S.‑based hardware for training large models, depending on how the restrictions are implemented and which countries are affected.
For governments that see themselves as targets of U.S. controls, this step underlines that Washington is focused not just on where advanced chips are sold, but on who is allowed to run them from afar.
Key points to watch now are the detailed wording of Commerce’s rule once it is published, how U.S. cloud providers adjust their access policies, and whether countries such as China publicly respond or look for alternative sources of high‑end AI computing power.