US authorities have arrested an American man over an alleged scheme worth roughly $300 million to smuggle Nvidia AI servers into China, according to a report by the South China Morning Post.

The case centers on servers built around Nvidia's AI accelerators — hardware that sits at the center of Washington's export control regime. Since 2022, the US Commerce Department has restricted sales of Nvidia's top data-center GPUs to Chinese customers, with the controls tightened repeatedly to cover successive product generations. The arrest signals that enforcement has now moved beyond fines and entity-list designations to criminal prosecution of individuals.

The reported $300 million figure, if confirmed in court filings, would rank the alleged scheme among the largest publicly known attempts to route restricted US AI hardware to Chinese buyers. It would represent hundreds of server systems given typical per-unit pricing for eight-GPU AI server platforms, which sell for several hundred thousand dollars each depending on configuration.

Smuggling cases like this one sit at the intersection of two commercial realities. First, Chinese demand for Nvidia's AI compute remains enormous: domestic alternatives have not yet matched the performance of Nvidia's flagship parts, and Chinese cloud providers and AI labs have continued to seek restricted hardware through intermediaries. Second, Nvidia's China data-center revenue has collapsed under successive export rules, turning what was once a multi-billion-dollar market into a legally constrained one — and creating the arbitrage conditions that black-market channels exploit.

Each high-profile arrest also carries a compliance message to the supply chain itself. Distributors, system integrators, resellers and freight forwarders handling Nvidia server hardware now face heightened scrutiny of end-customer verification, and the risk profile of gray-market transactions has shifted from commercial to criminal. For US export-control authorities, prosecuting individuals is a comparatively cheap enforcement tool that raises the cost of circumvention across the entire network of brokers moving restricted silicon.

The case will be watched closely for what it reveals about the mechanics of the alleged pipeline: which intermediaries were used, how the servers were routed, and which Chinese entities allegedly received the hardware. Those details will matter to Nvidia, which has insisted it complies with all export regulations, and to regulators assessing how much restricted compute is actually reaching China despite the controls.

Prosecutors will need to prove the elements of the alleged scheme in court, and the $300 million figure remains an allegation rather than an adjudicated fact. But if the case proceeds to conviction, it will likely prompt tighter know-your-customer requirements across the AI server distribution chain — and give both Nvidia and its channel partners a strong incentive to police where their highest-margin hardware ends up.