Nvidia Downplays Discovery of Banned Blackwell Ultra Chips in China Telecom Network

Nvidia has responded to the discovery of 32 Blackwell Ultra B300 graphics processing units within the infrastructure of China Telecom by describing the incident as a negligible portion of the broader market. The state-owned telecommunications company operates across China and now hosts these high-performance chips, which are subject to strict U.S. export controls. The identification of these specific units raises significant questions about how restricted hardware is continuing to reach Chinese data centers despite official bans on their sale to non-whitelisted buyers in the region.

The chips were located by Lava, an Israeli security firm, using Nvidia’s own DCGM Exporter tool. This software is typically installed by data center operators to monitor server health and performance. Although intended for internal use, the tool can sometimes be accessed from outside the data center network and does not always require password authentication. While Lava confirmed the presence of the GPUs, it could not determine exactly where within China Telecom’s network they are situated or how they were originally acquired. Nvidia has previously acknowledged working on a dedicated tool to locate its chips but has not yet released it.

The Blackwell Ultra B300 represents some of the most powerful data center hardware currently available. Announced in March of last year, the chip offers substantial performance improvements over the previous generation B200, including 50% more memory and higher FP4 compute capabilities. Nvidia marketed the hardware as built for the age of reasoning, specifically highlighting its suitability for sophisticated large language models. The chip was expected to ship by the end of 2025 and quickly became a highly sought-after asset in the artificial intelligence sector.

This latest finding adds to a pattern of restricted Nvidia hardware appearing in China. In 2025, reports indicated that Chinese companies smuggled approximately one billion dollars worth of Nvidia AI chips into the country over a three month period. At that time, the B200 was the primary target, but distributors were already preparing for the arrival of the B300 series. The difficulty in controlling the flow of these chips has led to complex smuggling operations. For example, in August of this year, nine individuals in Taiwan were indicted for the illegal export of B300 GPUs. That indictment followed an investigation by Supermicro, which fired several employees and identified a Taiwanese server trading firm with whitelisted status as part of the chain.

A spokesperson for Nvidia told Hot News Digest that the diversion of controlled U.S. products remains insignificant. The company noted that recent court cases have shown less than half of one percent of products are diverted. Nvidia argued that this figure is a drop in the bucket compared to the vast amount of domestic compute power China already possesses. With tens of millions of GPUs in service in approved countries, the company maintains that smuggling is not a major factor and is effectively blocked by governments on both sides of the Pacific.

The high demand for Nvidia silicon has driven up prices across the board, even for older generations. The A100 servers, which are five years old, have seen their prices triple and can now cost up to 82,000 dollars. The scarcity of newer chips has also led some operators to retool gaming GPUs to run AI workloads where possible. As the global competition for AI infrastructure intensifies, the ability to trace and control the distribution of these critical components remains a central challenge for both technology providers and regulatory bodies.

The situation highlights the ongoing tension between export control policies and the rapid pace of AI development. While Nvidia insists that the leakage of a small number of chips does not fundamentally alter the competitive landscape, the continued appearance of banned hardware in key state-owned networks suggests that enforcement gaps remain. As new generations of chips are developed, the methods of smuggling and the tools used to detect them will likely evolve in parallel.