Nvidia Concedes Chinese Chip Smuggling Persists

Israeli security firm Lava has identified 32 Nvidia Blackwell Ultra B300 graphics processors within the network of China Telecom, a state-owned telecommunications provider. The discovery highlights the ongoing difficulty of enforcing US export controls on high-performance AI hardware. Although these specific chips are barred from export to sanctioned parties, including China, their presence in a domestic data center raises questions about the integrity of the global supply chain.

The B300 is one of the most powerful data center GPUs currently available, offering significant performance gains over its predecessor. It features 50% more memory and FP4 compute capabilities than the previous-generation B200 chip. Nvidia announced the hardware in March of last year, positioning it for the era of advanced reasoning models. At the time, the company projected shipments would begin by the end of 2025, quickly making the unit a highly sought-after asset for AI developers.

Nvidia has responded to the findings with a mix of acknowledgment and minimization. A spokesperson told Tom's Hardware that while the diversion of controlled products is real, it remains statistically small. The company noted that recent court cases have shown less than half of one percent of controlled US products were diverted. The spokesperson described this as a drop in the bucket compared to the vast amount of domestic compute capacity China already possesses. Nvidia insisted that with tens of millions of GPUs in service in approved countries, smuggling remains a non-factor.

Background on the issue reveals a persistent pattern of leakage. In 2025, reports indicated that Chinese companies smuggled approximately $1 billion worth of Nvidia AI chips into the country over a three-month period. During that time, B200 chips were the primary target, but distributors were already preparing for the release of the B300 series. The current discovery in China Telecom's infrastructure adds to the evidence that export bans alone have not fully halted the flow of advanced silicon.

The detection method used by Lava adds a layer of complexity to the story. The firm identified the chips using Nvidia's own DCGM Exporter tool. This software is typically installed by data center operators to monitor server health and performance. While designed for internal use, the tool can often be accessed from outside the data center network and does not always require password authentication. This vulnerability allowed Lava to pinpoint the location of the hardware, though the exact placement within China Telecom's network remains unconfirmed.

Reactions to the smuggling have escalated recently. In August of this year, nine individuals were indicted in Taiwan for the illegal export of B300 GPUs. This legal action followed an investigation by Supermicro, which fired several employees linked to the smuggling operation. The illicit trade involved a Taiwanese server trading firm known as Flying Tiget Tech, which held whitelisted status for GPU purchases. This case illustrates the sophisticated methods used to bypass regulatory checks.

Market dynamics further complicate the enforcement landscape. Demand for Nvidia's silicon has driven up prices for older hardware, with A100 servers tripling in cost despite being five years old and reaching up to $82,000. Some buyers have even resorted to retooling gaming GPUs for AI workloads. Nvidia is aware of these trends and has confirmed it is working on a tool to help locate its chips, though the release is still pending.

What happens next remains uncertain. While Nvidia continues to downplay the impact of these leaks, the presence of B300 chips in a major Chinese network suggests that enforcement gaps persist. The company's upcoming chip-location tool may provide better visibility into the hardware's whereabouts. For now, the tension between US export policy and Chinese demand for advanced AI compute continues to shape the global technology landscape.