Nvidia privately admitted it can track its GPUs against smuggling
In closed-door conversations, Nvidia told Illinois Democrat Bill Foster that it could track the location of its GPUs, even though the Semiconductor Industry Association (SIA) called such measures "potentially infeasible" in March. In writing, the company confirmed nothing; its engineers simply replied, "Of course we can." For GPU buyers, the takeaway is straightforward: location verification could become a condition of supply.
AI-processed from MIT Technology Review; edited by Hamidun News
The debate over whether AI accelerators can be technically tracked to stop smuggling has run for more than a year. Supporters of the law say the technology already exists. The industry responds that it is immature and possibly infeasible. Now the debate has an unexpected witness: according to MIT Technology Review, Nvidia itself signaled in private conversations that tracking is possible.
What Nvidia said behind closed doors
Illinois Democratic Congressman Bill Foster told MIT Technology Review that Nvidia representatives acknowledged in technical discussions that the measures in the Chip Security Act can be implemented. Foster co-authored the bill and sits on the House Committee on Science, Space, and Technology. According to him, the company put nothing in writing, but its reaction to the idea of protective mechanisms was: "Yes, of course we can do that."
This sharply contradicts the industry's public position. In March 2026, the Semiconductor Industry Association (SIA), of which Nvidia is a member, came out against the bill. Its letter described the requirements as "new, unproven, and potentially infeasible." Nvidia left a request for comment and confirmation of Foster's account unanswered.
An important caveat: this is a secondhand account of private conversations, relayed by a politician, not an official company statement. Still, there is no shortage of indirect evidence. In December 2025, Nvidia announced Fleet Intelligence, a tool that helps customers monitor chip operations across different data centers. As the company confirmed to Reuters, it can show which country an accelerator is in. Nvidia declined to comment on whether it could be used against smuggling.
How a chip can be tracked
The bill allows for three approaches: site inspections, written assurances from companies, and location verification via signal. The third option is the hardest to get around. It works like smartphone geolocation: when powered on, the accelerator sends a "ping" with its position data. Analysts at the Center for a New American Security (CNAS) concluded in September that this method is also the cheapest.
For developers and infrastructure teams, this is not abstract. If the requirement lands in the U.S. defense budget (the bill is being considered for inclusion in the Senate NDAA), server vendors, cloud providers, and cluster operators will face new mandatory telemetry. They will need to account for how it is built into the hardware, who receives the data, and what happens when the communication link fails.
Why the topic is back in focus
The reason is the scandal surrounding Nvidia. CEO Jensen Huang has resigned, and reportedly had an affair with Alice Huang, who runs the Singapore-based company Megaspeed. Megaspeed is now under investigation on suspicion of smuggling advanced GPUs into China. Reportedly, a U.S. Commerce Department review of the firm was halted after Nvidia intervened. Megaspeed allegedly bought chips from Supermicro, which builds servers based on Nvidia processors.
In October, as concern in Congress grew over Nvidia's role, the company answered a question about Fleet Intelligence with a general formula: by its account, the volume of leaked controlled goods is negligible, "a drop in the ocean" compared with the computing power China already has.
Limits and what's next
Even if the law passes, it does not solve everything. Tracking only shows investigators where chips are. It does not allow the disabling of an accelerator that has already been smuggled out. Foster intends to introduce a tougher bill in the new year. It would require a license to operate advanced GPUs, and a valid license would have to be verified before a chip is connected.
Foster did not ask Nvidia representatives why the company does not voluntarily deploy such tools if it can. In his assessment, after the scandal there will be "significant political pressure" to make geographic chip verification truly robust.
The practical takeaway: the market is moving toward a model in which access to powerful GPUs depends on verified location and licensing. Those building clusters, selling servers, or planning international accelerator purchases should build compliance and telemetry into their plans now. Rules may arrive faster than the industry expected.
- Nov 22, 2026SEC Probes Nvidia's Accounting as Shares Drop 14% in One Day
- Nov 22, 2026Nasdaq 100 Falls 10.4% in a Week, Its Worst Stretch Since the Pandemic
- Nov 22, 2026LangChain, a Side Project Turned $3 Billion Startup in Four Years, Raises $300 Million
- Nov 22, 2026Nvidia Stops Selling H200 in China and Rejects Musk's Push to Bring Back Huang
Key Questions Answered
What Nvidia said behind closed doors?
Illinois Democratic Congressman Bill Foster told MIT Technology Review that Nvidia representatives acknowledged in technical discussions that the measures in the Chip Security Act can be implemented. Foster co-authored the bill and sits on the House Committee on Science, Space, and Technology. According to him, the company put nothing in writing, but its reaction to the idea of protective mechanisms was: "Yes, of course we can do that." This sharply contradicts the industry's public position.
How a chip can be tracked?
The bill allows for three approaches: site inspections, written assurances from companies, and location verification via signal. The third option is the hardest to get around. It works like smartphone geolocation: when powered on, the accelerator sends a "ping" with its position data. Analysts at the Center for a New American Security (CNAS) concluded in September that this method is also the cheapest. For developers and infrastructure teams, this is not abstract. If the requirement lands in the U.S.
Why the topic is back in focus?
The reason is the scandal surrounding Nvidia. CEO Jensen Huang has resigned, and reportedly had an affair with Alice Huang, who runs the Singapore-based company Megaspeed. Megaspeed is now under investigation on suspicion of smuggling advanced GPUs into China. Reportedly, a U.S. Commerce Department review of the firm was halted after Nvidia intervened. Megaspeed allegedly bought chips from Supermicro, which builds servers based on Nvidia processors.
Test Frontier LLMs & Autonomous Agents in One Workspace
Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.