The Race to Build a “Find My” for AI Chips

Chips & Policy

Washington Chases a 'Find My' for AI Chips as Smuggling Persists

A $300 million server smuggling case and a House-passed bill are pushing chipmakers toward ping-based location verification, cryptographic certificates and hardware fingerprints.

By
Grace Kim
Filed
Channel
Chips & Policy
Read
4 min read

On 1 October 2026, U.S. authorities arrested a man accused of shipping more than $300 million worth of GPU-equipped servers to China on paperwork meant for Malaysia and Singapore. The case underscores why Washington now wants a way to locate advanced AI chips after they leave the factory. A bill requiring location verification for those chips cleared the House Foreign Affairs Committee last March, and legislators pressed for a floor vote before the late-September U.S.-China summit.

The technology to comply may already exist. A September report from Georgetown's Center for Security and Emerging Technology (CSET) lays out a feasible approach: ping-based location verification (PLV), supplemented by physical inspections.

How would ping-based location verification work?

PLV measures the round-trip time of a ping traveling from a chip to multiple trusted "landmark" servers around the world. Timing several such messages places the chip within a radius. Kyle Miller, senior research analyst at CSET and report coauthor, says manufacturers could build PLV into new AI chips in hardware, while a firmware update could cover chips already in circulation.

Not every chip warrants tracking. "The priority should be for the highest-end data center chips—the cream of the crop in high-end data center GPUs," said Jacob Feldgoise, senior data research analyst at CSET and coauthor of the report.

A parallel effort comes from secure computing firm Lucid Computing, which has published a draft specification for encrypted "sovereignty certificates." These are "cryptographic credentials that provide verifiable, unforgeable proof of the physical location of a computing device, such as a CPU or GPU, at a specific point in time," per the specification.

Lucid CEO and cofounder Kristian Rönn described the mechanism: "If you can use a private key that exists on a particular chip, and you can sign a message with that cryptographic key, and then that message is received by a so-called landmark server, then you can send a couple of those messages and time those messages." He added: "You get a radius of where the chip has to be." Many modern chips already ship with trusted execution environments and a hardware root of trust capable of signing such messages.

Miller believes deterrence has value. "They know if they turn this chip on, it's going to start pinging these landmark servers. That could deter some smuggling." When the system flags a chip in the wrong place, "the PLV data gives you a better sense of when and where it was moved," he said.

Where does PLV fall short?

The method only detects diversion after it happens. It cannot see offline chips in storage, in transit, or before activation. A failed ping leaves the system unable to distinguish a valid outage from tampering. Physical follow-up remains essential. "Essentially, [PLV] generates leads, and you track down those leads," Miller said.

Adversaries with physical access could block transmissions or extract cryptographic keys and spoof a signal from the intended location. One proof-of-concept study took a different route, generating unique fingerprints for Nvidia H200 GPUs from manufacturing variations and random seeds. Rönn supports that breadth: "You might want to use some statistical fingerprinting to know that the chip is what it's supposed to be. It also requires good physical security around the chips, such as tamper-proofing."

The CSET report also warns PLV is costly, and Miller noted a strategic limit: if China achieves the ability to craft its own advanced AI chips, "then tracking the export of U.S. chips is less of a strategic priority."

Report coauthor Hanna Dohmen pointed to another gap in the export control regime: "Currently, Chinese companies are legally able to remotely access AI chips abroad through cloud service providers." Advanced chips increasingly sit in data centers across Southeast Asia, including Malaysia, that serve Chinese firms.

Regardless of legislation, Feldgoise expects tighter export terms over time. "The chips that generate compute are being seen as a strategic resource, so I would be unsurprised if the due diligence requirements on these devices increase as time passes," he said. Rönn likens sovereignty certificates to a "lie detector" for data centers, one that answers where, how, what and who in verifying compliance.

Original: theinformation.com

Share this article:

More from Grace Kim

Grace Kim

Show full bio

Market editor covering people, launches and day-to-day practice in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.

763 articles

Related articles

« Previous article