The U.S. is considering a drastic new move to control the use of NVIDIA AI chips—adding a built-in “kill switch.” This comes after export controls failed to stop China from accessing advanced chip technology through loopholes and third-party sellers.
For years, Washington has tried to slow China’s tech rise. But those efforts have struggled. Despite restrictions, Chinese firms like Huawei continue to thrive, using domestically-made alternatives or acquiring U.S. chips through backdoor channels.
Now, U.S. lawmakers want to go beyond blocking exports. They’re aiming to control how NVIDIA AI chips are used after they’ve left American hands. Representative Bill Foster, a physicist and Democrat from Illinois, is drafting a bill that could reshape AI chip regulation. His proposal would mandate built-in location tracking and a remote shutdown system—what he calls a “kill switch.”
This legislation targets a key weakness. Once AI chips are exported, there’s little oversight. NVIDIA has admitted it can’t trace where all its chips end up. That gap is exactly what Foster wants to fix.
“This isn’t a future problem. It’s happening now,” Foster told Reuters. He warned that the Chinese Communist Party could already be using American chips to develop AI weapons or artificial general intelligence (AGI). To prevent this, his bill would require that chips verify their location regularly and shut down if used without a valid U.S. export license.
The technology for this already exists. Google, for example, uses similar location verification on its in-house AI chips. They ping secure servers. By measuring the response delay, the system can estimate where the chip is located. Experts say this could be adapted for use in NVIDIA AI chips with minimal changes.
Tim Fist, a former engineer and director at the Institute for Progress, supports the idea. He argues that U.S. regulators lack real-time insight into where chips go. With location tracking, enforcement becomes far more effective.
The proposed kill switch would go further. It would prevent chips from operating unless they pass a software check confirming their license. While harder to implement, this method could stop unauthorized use even after the chip is sold.
Foster’s plan is already gaining bipartisan support. Lawmakers from both parties, including those on the House Select Committee on China, are backing tighter controls. Although the bill isn’t officially out yet, Foster expects to introduce it soon.
The timing is key. In February, U.S. regulators froze shipments of NVIDIA’s H20 chip—a model designed for China—by tightening license rules. Major Chinese tech giants like Baidu and ByteDance are now scrambling to respond. At the same time, Huawei has reemerged as a player, thanks to a chip partnership with SMIC, China’s leading semiconductor maker.
Chip smuggling has also resurfaced. In one recent case, a Chinese national was among three people charged in Singapore for fraud tied to banned NVIDIA AI chips. Meanwhile, Chinese AI firm DeepSeek has built powerful models using restricted chips—raising further concern in Washington.
Foster’s bill would give the Commerce Department six months to finalize the rules. While exact details are still being debated, the concept is clear. The U.S. wants more than border checks. It wants control at the hardware level—inside the NVIDIA AI chips themselves.
The economic impact is also significant. NVIDIA earned over $17 billion from the Chinese market last fiscal year, about 13% of its total revenue. Yet much of that revenue may now be at risk. Chips sold to Hong Kong often end up in mainland China, bypassing restrictions.
As AI becomes a national security issue, policymakers are shifting their strategy. The old export rules weren’t enough. Now, the focus is on embedding compliance into the chips—so they can’t be used without approval.
If passed, Foster’s legislation could mark a major turning point. It would reshape how NVIDIA AI chips are sold, tracked, and activated around the world. And it might become the standard for regulating advanced AI hardware in the years ahead.
