32 Banned Nvidia Chips Found in Chinese State Network
A security firm detected restricted Blackwell Ultra hardware within China Telecom's infrastructure, prompting Nvidia to argue the leak is minor relative to China's total compute capacity.
By Daniela Vargas ·

Thirty-two Nvidia Blackwell Ultra B300 graphics processing units have been identified inside the network of China Telecom, a state-owned telecommunications provider. These specific chips are subject to U.S. export controls that prohibit their sale to sanctioned parties, raising immediate questions about how restricted hardware ended up in a major Chinese data center. The discovery was made by Lava, an Israeli cybersecurity firm, which utilized Nvidia’s own DCGM Exporter tool to detect the presence of the hardware.
The DCGM Exporter is a software utility designed for data center operators to monitor server health and performance. Although intended for internal use, the tool can sometimes be accessed from outside the immediate network and does not always require password authentication. While Lava confirmed the existence of the 32 chips, it could not specify their exact location within China Telecom’s infrastructure. Nvidia acknowledged the findings but noted that it is still developing a dedicated tool for locating its chips in the field, which has not yet been released.
Nvidia has responded by downplaying the significance of the discovery, arguing that it represents a small fraction of the broader market. A spokesperson stated that recent court cases have shown the diversion of controlled U.S. products to be less than half of one percent. The company described this as a "drop in the bucket" compared to the vast amount of domestic compute power China already possesses, asserting that with tens of millions of GPUs in approved countries, smuggling remains a non-factor managed by governments on both sides.
The presence of these chips is notable because the B300 series offers significant performance advantages over its predecessor, the B200. Announced in March 2025, the B300 provides 50% more memory and FP4 compute capability. At launch, Nvidia positioned the chip as being "built for the age of reasoning," highlighting its suitability for advanced large language models like DeepSeek R1. Despite these high-end features, the chips were expected to ship by the end of 2025, making them a highly sought-after commodity for AI development.
The leak follows a trend of smuggling that has plagued the GPU market. In 2025, reports indicated that Chinese companies imported one billion dollars worth of Nvidia AI chips in just three months. While the B200 was the primary target at that time, distributors were already preparing for the B300 release. The enforcement challenge is compounded by the complex logistics of moving high-value hardware across borders.
Recent legal actions have targeted the smuggling networks. In August 2026, nine individuals in Taiwan were indicted for the illegal export of B300 GPUs. This occurred shortly after Supermicro fired several employees following an internal investigation. The smuggling operation reportedly involved Flying Tiget Tech, a Taiwanese server trading firm with whitelisted status for purchasing GPUs. This case illustrates the difficulty in tracking restricted hardware through legitimate supply chains.
The scarcity of these advanced chips has also impacted the secondary market. Older Nvidia hardware, such as the five-year-old A100 servers, has seen prices triple, with units now fetching up to $82,000. As demand for AI compute continues to outpace supply, some organizations have even resorted to reconfiguring gaming GPUs to handle AI workloads, highlighting the persistent pressure on the market.
Source: Tom's Hardware


