Blackwell Ultra Chips Detected in State-Owned Chinese Network Despite Export Bans
Security researchers found 32 banned Nvidia GPUs inside a major Chinese telecom provider, prompting the chipmaker to label the discovery statistically insignificant.
By Kendra Konchanuk ·
For the first time, high-end Nvidia processors that were explicitly barred from entering China have been confirmed present in the infrastructure of a major state-owned telecommunications provider. The discovery highlights a persistent gap in global export controls, as these specific chips were restricted to approved buyers who agreed not to sell to sanctioned entities. While the immediate impact on the general consumer is limited, the finding underscores the ongoing difficulty in regulating the flow of advanced computing power, a factor that continues to influence the broader AI hardware market.
Israeli security firm Lava identified 32 Nvidia Blackwell Ultra B300 chips within the network of China Telecom. This state-owned entity provides essential telecommunications services across the country. The presence of these units is notable because U.S. export regulations have already prohibited the sale of this specific generation of hardware to Chinese buyers. Typically, these chips are restricted to a whitelist of purchasers who contractually agree to limit distribution to sanctioned parties, making the detection of 32 units a clear deviation from established protocols.
The issue of high-performance GPU smuggling is not new. Reports from 2025 indicated that Chinese companies moved approximately $1 billion worth of Nvidia AI chips into the country in just three months. At that time, the B200 model was the primary target, though distributors were already positioning themselves to move the newer B300 series as soon as it became available. The current detection suggests that despite previous crackdowns, the diversion of controlled technology remains a significant operational challenge for both U.S. regulators and Nvidia.
Nvidia has responded to the findings with a tone of measured confidence, downplaying the significance of the discovery without denying the data. A spokesperson noted that recent court cases have shown that the diversion of controlled U.S. products accounts for less than half of one percent of the total. They described this volume as a 'drop in the bucket' compared to the vast amount of domestic computing capacity China already possesses. The company emphasized that with tens of millions of GPUs in service in approved regions, such incidents are a non-factor in the overall landscape.
The method used to locate the chips offers an interesting technical insight. Lava identified the B300 units using Nvidia’s own DCGM Exporter tool. This software is designed for data center operators to monitor server health and performance. While intended for internal use, the tool can sometimes be accessed from outside the data center network and does not always require password authentication. This vulnerability allowed researchers to pinpoint the hardware even though the exact physical location within China Telecom’s facilities remains unknown.
The B300 chip is highly sought after due to its performance capabilities. Announced in March of last year, the data center GPU offers 50% more memory and FP4 compute power than its predecessor, 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, but its high performance quickly made it a target for illicit trade.
The smuggling of these chips has become a complex international affair. In August 2026, nine individuals were indicted in Taiwan for illegally exporting B300 GPUs to China. This legal action followed closely after Supermicro fired several employees following an internal investigation. The schemes often involve intermediaries, such as the Taiwanese server trading firm Flying Tiget Tech, which held whitelisted status for GPU purchases, allowing it to act as a conduit for the hardware.
The scarcity of Nvidia chips has had a tangible effect on pricing across the market. Older models, such as the five-year-old A100 servers, have seen prices triple, with units now costing up to $82,000. In response to the ongoing crackdown, some operators have resorted to retooling gaming GPUs to handle AI workloads where possible. As Nvidia continues to work on a dedicated tool to help locate its chips in the wild, the market will likely remain vigilant regarding the movement of its most powerful silicon.
Source: Tom's Hardware
