Advanced AI Chips Found in Chinese State Network Despite Trade Rules

For many people following the rapid expansion of artificial intelligence, the story has been one of strict boundaries between technology providers and major users. That narrative faced a challenge this week when a security firm revealed that 32 high-end Nvidia processors were found operating inside the network of China Telecom, a state-owned telecommunications company. These specific chips, known as the Blackwell Ultra B300, have been barred from export to China by U.S. regulations. The discovery highlights how difficult it remains to keep the world's most powerful AI hardware out of the hands of restricted buyers, even when official trade rules are in place.

The B300 is a significant piece of hardware. Announced last year, it offers a 50 percent increase in memory and compute power compared to its predecessor, the B200. Nvidia originally designed the chip for sophisticated AI models and expected it to ship by the end of 2025. Because of its power, the chip became a highly sought-after item. Export controls require that such chips only be sold to approved buyers who promise not to transfer them to sanctioned parties. Despite these safeguards, the chips ended up in a Chinese data center, a situation that has occurred before. In 2025, reports indicated that Chinese companies smuggled one billion dollars worth of Nvidia AI chips into the country in just three months.

Nvidia has responded to the news with a measured tone. A spokesperson stated that the diversion of controlled products is a small fraction of the total, describing it as a drop in the bucket compared to the massive amount of computing power China already possesses. The company argued that with tens of millions of GPUs in service in approved countries, smuggling remains a non-factor. Nvidia did not deny the findings but suggested that the issue is being overstated by governments, researchers, and media outlets.

The method used to find the chips is also noteworthy. The security firm, Lava, used a monitoring tool called DCGM Exporter to identify the processors. This software is typically installed by data center operators to track server health and performance. While intended for internal use, it can sometimes be accessed from outside the network. In some cases, it does not even require a password for authentication. It is not yet clear exactly how the chips entered the network or where specifically they are located within the company's infrastructure.

The difficulty of stopping these transfers is evident in recent legal actions. In August 2026, nine people in Taiwan were indicted for illegally exporting B300 GPUs to China. This followed an internal investigation at Supermicro, which led to the firing of several employees. The smuggling operation reportedly used a Taiwanese server trading firm with approved status to move the hardware. The high demand for these chips has also driven up prices for older models. Nvidia's A100 servers, which are five years old, have seen their prices triple and can now cost up to 82,000 dollars. Some companies have even begun retooling gaming graphics cards to handle AI workloads as a temporary solution.

The situation illustrates the complex global market for AI hardware. While export controls aim to manage the flow of advanced technology, the high value of these chips creates strong incentives for alternative routes. For now, the discovery serves as a reminder that the landscape of AI computing is still evolving, with security and supply chain challenges continuing to shape the industry.