Chinese Telecom Found Holding 32 Banned Nvidia Chips

For the average person watching the technology news, this discovery highlights just how difficult it is to control the flow of high-end computer parts across borders. Security firm Lava recently identified 32 Nvidia Blackwell Ultra B300 graphics cards operating within the network of China Telecom. This state-owned telecommunications provider operates domestically in China, making the presence of these specific chips notable because U.S. export controls have already barred them from leaving the country. These restrictions are designed to ensure that only approved buyers, who agree not to resell to sanctioned parties, can purchase such powerful hardware.

The finding raises questions about how the chips arrived in Chinese data centers. This is not the first time export bans have faced challenges. In 2025, reports indicated that Chinese companies smuggled one billion dollars' worth of Nvidia AI chips into the country over a three-month period. While the B200 model was the primary target at that time, distributors were already preparing for the B300 series. The new discovery confirms that the movement of these controlled chips into China remains a persistent issue for regulators and manufacturers alike.

Nvidia has responded to the news with a measured tone, choosing to downplay the significance of the find without denying the facts. A spokesperson stated that while governments and researchers have found some diversion of controlled products, the amount involved is less than half of one percent in recent cases. The company described this as a drop in the bucket compared to the vast amount of domestic computing power China already possesses. With tens of millions of its GPUs in service in approved countries, Nvidia maintains that smuggling is not a major factor in the broader market.

The method of detection offers an interesting technical insight. Lava located the chips using Nvidia’s own DCGM Exporter tool, a software utility intended for internal use by data center operators to monitor server health. Although designed for internal monitoring, this tool can sometimes be accessed from outside the data center network and does not always require a password. While it is not yet clear exactly where within China Telecom’s infrastructure the chips are located, their presence has been confirmed. Nvidia is currently working on a dedicated tool to help locate its chips, but that feature has not yet been released.

The B300 chip is highly sought after because it offers significant performance improvements over its predecessor. Announced in March of last year, it provides 50% more memory and FP4 compute capabilities than the B200. Nvidia originally marketed the chip as being built for the age of reasoning, citing advanced AI models like DeepSeek R1. It was expected to ship by the end of 2025, and its popularity led to increased scrutiny. In August 2026, nine individuals were indicted in Taiwan for illegally exporting B300 GPUs to China. This followed an investigation by Supermicro, which fired several employees after discovering that a whitelisted trading firm was involved in the smuggling process.

The intense demand for these chips has also affected the secondary market. Older Nvidia hardware, such as the A100 servers, have seen prices triple despite being five years old, with some units costing up to $82,000. Because new chips are so vital for AI data centers, some operators are even retooling gaming graphics cards to handle AI workloads when possible. The ongoing effort to track and control these components shows that the global market for AI hardware remains complex and tightly contested.