Nvidia Blackwell Ultra Chips Discovered in China Telecom Network
Israeli security firm Lava found 32 banned B300 GPUs in a Chinese state-owned carrier's infrastructure, prompting Nvidia to call the smuggling incident a minor anomaly.
By Gabriela Flores ·

Nvidia Blackwell Ultra B300 graphics processing units, which are subject to strict U.S. export controls, have been identified inside the network of China Telecom. The discovery was made by Lava, an Israeli security company, which located 32 of the high-performance chips within the infrastructure of the state-owned telecommunications provider. The findings raise significant questions about how hardware barred from export ended up in a Chinese data center, as these specific chips are only permitted for sale to whitelisted buyers who agree not to pass them on to sanctioned parties.
The presence of the B300s is not an isolated incident in the broader context of chip smuggling. In 2025, reports indicated that Chinese firms moved approximately $1 billion worth of Nvidia AI chips into the country over a three-month period. While the B200 series was the primary target at that time, distributors were already positioning themselves to sell the newer B300 models as they became available. The B300 offers substantial performance improvements over its predecessor, featuring 50% more memory and FP4 compute capability. Nvidia originally announced the chip in March 2025 as a solution for advanced reasoning AI, with shipments initially expected by the end of that year.
Nvidia responded to the latest report by downplaying the significance of the find, though it did not dispute the facts. A spokesperson for the company stated that the diversion of controlled products has been found to be insignificant, citing recent court cases where the amount was less than 0.5%. The spokesperson described this leakage as a drop in the bucket compared to the vast amount of domestic compute capacity China already possesses. Nvidia noted that with tens of millions of its GPUs in service in approved countries, smuggling remains a non-factor that is effectively blocked by governments on both sides of the Pacific.
The mechanism for the chips' arrival remains unclear. Lava confirmed the presence of the B300s using Nvidia's DCGM Exporter tool, a software utility designed for data center operators to monitor server health. This tool is generally intended for internal use but can often be accessed from outside the data center network. Notably, it does not always require password authentication, which may have facilitated the discovery. It is currently unknown exactly where within China Telecom's network the servers are located. Nvidia has previously acknowledged work on a tool to help locate its chips, but that solution has not yet been released.
The smuggling of these powerful processors has become a complex operation involving multiple intermediaries. In August 2026, nine individuals were indicted in Taiwan for the illegal export of B300 GPUs to China. This legal action followed an investigation by Supermicro, which fired several employees after uncovering the smuggling scheme. The process involved a Taiwanese server trading firm called Flying Tiget Tech, which held whitelisted status for GPU purchases. The demand for Nvidia silicon is so intense that even older models like the five-year-old A100 servers have seen prices triple, reaching costs of up to $82,000. Some buyers have even resorted to retooling gaming GPUs to handle AI workloads where possible.
The stakes for the AI industry continue to rise as access to top-tier hardware becomes a critical competitive advantage. The discovery of banned chips in a major state-owned network highlights the persistent challenges in enforcing export controls. As China continues to develop its own AI capabilities, the flow of U.S. hardware remains a focal point for both government regulators and corporate strategists. The situation underscores the high value placed on Nvidia's technology, where even a small number of diverted units can have a disproportionate impact on local AI development efforts.
Source: Tom's Hardware


