Nvidia Downplays Discovery of 32 Banned B300 Chips in Chinese State Network

Nvidia has responded to a new report revealing that 32 of its high-end Blackwell Ultra B300 GPUs are operating within the network of China Telecom, a major state-owned telecommunications provider in China. The discovery comes from Lava, an Israeli cybersecurity company, which claims to have identified these specific chips despite U.S. export controls that strictly limit their sale to sanctioned parties. The presence of these processors in a Chinese data center raises immediate questions about how they bypassed the regulatory hurdles designed to keep them out of the country's hands.

The B300 chips are part of Nvidia's latest generation of data center hardware, which are currently barred from export to China. Under current rules, these advanced AI accelerators can only be sold to buyers on a whitelist who agree not to resell them to sanctioned entities. The fact that 32 units ended up in China Telecom’s infrastructure suggests that the current enforcement mechanisms may have gaps. This is not the first time such leaks have occurred, as previous reports indicated that significant volumes of Nvidia hardware had been smuggled into the region earlier in the year.

According to Lava, the chips were located using Nvidia’s own DCGM Exporter tool. This software is typically installed by data center operators to monitor the health and performance of their servers. While intended for internal use, the tool can sometimes be accessed from outside the data center’s local network. Notably, it does not always require password authentication, which may have facilitated the identification of the hardware. Nvidia has acknowledged working on a dedicated tool to help locate its chips, but that solution has not yet been released to the public.

Nvidia has sought to minimize the impact of this finding, arguing that it represents a small fraction of the total computing power available in China. A company spokesperson stated that recent legal cases have shown that the diversion of controlled U.S. products amounts to less than half of one percent of the total volume. The spokesperson described this leak as a drop in the bucket compared to the massive amount of domestic compute capacity China already possesses. The company emphasized that with tens of millions of GPUs in service in approved countries, such smuggling incidents are a non-factor in the broader market dynamics.

The motivation for Chinese firms to acquire B300 chips is clear given the performance advantages they offer. Announced in March of last year, the B300 was marketed as a processor built for the age of reasoning, with significant improvements over its predecessor. It features 50% more memory and double the FP4 compute power compared to the B200. These specifications make it highly attractive for running sophisticated large language models, driving intense demand from AI developers who are looking to gain a competitive edge.

The smuggling of these chips has become an increasingly complex operation, often involving intermediaries with legitimate purchasing rights. In August of this year, nine individuals in Taiwan were indicted for illegally exporting B300 GPUs to China. This legal action followed an investigation by Supermicro, which fired several employees after finding evidence of smuggling. The operations reportedly utilized a Taiwanese server trading firm with whitelisted status to move the hardware, highlighting the intricate nature of the supply chain loopholes.

The scarcity of these advanced chips has also had a ripple effect on the pricing of older Nvidia hardware. Because demand for new silicon is so high, the value of previous generations has surged. For example, A100 servers, which are five years old, have seen their prices triple, with some units now costing up to $82,000. This price inflation underscores the critical role Nvidia’s chips play in modern AI data centers, even as companies look for ways to stretch older equipment.

Despite Nvidia’s insistence that the recent discovery is a minor issue, the ongoing efforts to smuggle hardware into China demonstrate the persistent tension between U.S. export controls and global AI demand. The industry continues to grapple with how to balance the need for cutting-edge technology with the strategic interests of protecting intellectual property and maintaining a competitive advantage in the AI race.

As the market evolves, both government regulators and private companies are working to tighten the screws on chip sales. The release of Nvidia’s upcoming location-tracking tool for its GPUs may provide a new layer of oversight, though it remains to be seen if it will be sufficient to stop the flow of hardware. For now, the presence of 32 B300 chips in China Telecom’s network serves as a reminder that the enforcement of export controls is an ongoing challenge for the tech industry.