Nvidia B300 Chips Discovered in China Telecom Network Despite Export Bans

The global effort to restrict high-performance AI hardware from reaching certain markets has encountered a significant hurdle. Israeli security firm Lava announced today that it has identified 32 Nvidia Blackwell Ultra B300 chips operating within the network of China Telecom, a state-owned telecommunications provider.

These specific chips are currently barred from export by U.S. regulations, raising immediate questions about how they ended up in Chinese data centers. Under current rules, such hardware can only be sold to whitelisted buyers who agree not to transfer the technology to sanctioned parties, including entities in China. The discovery highlights the persistent challenge of enforcing export controls on some of the most powerful processors available.

This incident follows a broader trend of smuggling activity in the AI chip market. In 2025, reports indicated that Chinese companies moved approximately $1 billion worth of Nvidia AI chips into the country over a three-month period. While the B200 model was the primary target then, distributors were already preparing to move B300-series units as soon as they became available, suggesting a well-organized pipeline for illicit trade.

Nvidia has responded to the news with relative calm, downplaying the impact without denying the presence of the chips. A spokesperson stated that recent court cases have shown insignificant diversion of controlled U.S. products, citing figures of less than 0.5 percent. The company characterized these instances as a drop in the bucket compared to the vast amount of domestic compute power China already possesses, noting that with tens of millions of GPUs in approved countries, smuggling remains a non-factor.

The method of detection offers a glimpse into the technical vulnerabilities of current monitoring tools. Lava identified the chips using Nvidia’s own DCGM Exporter, a software utility designed for data center operators to monitor server health and performance. Although intended for internal use, this tool can sometimes be accessed from outside the data center network and, notably, does not always require password authentication. Nvidia has acknowledged working on a new tool to help locate its chips but has not yet released it.

The desire for the B300 is driven by its superior specifications. Announced in March of last year, the chip offers 50% more memory and FP4 compute capabilities than its predecessor, the B200. Nvidia marketed the B300 as built for the age of reasoning, specifically citing advanced large language models like DeepSeek R1. Originally expected to ship by the end of 2025, the chip quickly became a high-value target for those seeking an edge in AI processing power.

Recent legal actions underscore the intensity of the smuggling trade. In August 2026, nine individuals in Taiwan were indicted for illegally exporting B300 GPUs to China. This followed closely after Supermicro fired several employees following an internal investigation into smuggling operations. The illicit network involved a Taiwanese server trading firm, Flying Tiget Tech, which held whitelisted status for purchasing these GPUs, complicating the chain of custody.

The scarcity and high demand for these chips have driven up prices across the board. Nvidia’s A100 servers, which are five years old, have seen their prices triple, with costs reaching up to $82,000. This market pressure has even led some organizations to retool older gaming GPUs to handle AI workloads, a testament to the critical role these processors play in modern data infrastructure. As the search for banned hardware continues, the gap between policy and practice remains a focal point for industry observers.