Nvidia Downplays Discovery of Banned B300 Chips in China Telecom Network

Nvidia has characterized the recent discovery of banned Blackwell Ultra B300 chips inside a Chinese state-owned network as a negligible anomaly in the broader market. The company maintains that such instances of diversion do not significantly impact its global supply or demand dynamics, describing the volume as a small fraction of the total compute resources already available to China.

The catalyst for this response is a report from Lava, an Israeli security firm, which claims to have identified 32 B300 processors within the infrastructure of China Telecom. These specific chips are currently under U.S. export controls, meaning they should only be sold to whitelisted buyers who agree not to resell them to sanctioned parties. Their presence in a Chinese data center raises questions about how these high-performance units bypassed strict regulatory checks.

This incident is not an isolated event in the ongoing trade friction surrounding artificial intelligence hardware. In 2025, reports indicated that Chinese entities smuggled approximately $1 billion worth of Nvidia AI chips into the country in just three months. While the B200 series was the primary target at that time, distributors were already positioning themselves to sell the newer B300 models as soon as they became available. The continued flow of these chips suggests that export bans remain difficult to enforce completely.

An Nvidia spokesperson addressed the findings directly, noting that government and media investigations have historically found diversion rates of less than 0.5 percent in recent legal cases. The company argued that with tens of millions of GPUs in service across approved countries, this specific report further demonstrates that smuggling is a nonfactor. The firm insists that both the U.S. and Chinese governments are actively working to block these unauthorized transfers.

The technical detection method used by Lava provides a clear audit trail. The firm identified the chips using Nvidia's DCGM Exporter tool, a software utility installed by 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 frequently lacks mandatory password authentication. Nvidia has previously confirmed it is developing a dedicated tool to help locate its chips, though that solution has not yet been released.

The Blackwell Ultra B300 is a highly sought-after component in the current AI market. Announced in March 2025, the GPU offers a 50 percent increase in memory and FP4 compute capabilities compared to the previous generation B200. Nvidia positioned the chip as being built for the age of reasoning, citing advanced large language models like DeepSeek R1. Its high performance explains why it has become a target for illicit trade despite the regulatory hurdles.

The pressure to acquire these chips has led to significant legal and corporate actions in recent months. In August 2026, nine individuals in Taiwan were indicted for the illegal export of B300 GPUs to China. This followed an internal investigation by Supermicro, which resulted in the termination of several employees linked to the smuggling ring. The scheme involved a Taiwanese server trading firm, Flying Tiget Tech, which held whitelisted status for purchasing the GPUs.

The scarcity of modern AI hardware has also driven up the cost of older technology. Nvidia's A100 servers, which are five years old, have seen their prices triple, with some units commanding up to $82,000. This price surge has led some operators to retool gaming GPUs for AI workloads as a cost-effective alternative. The market dynamics continue to shift as companies seek to balance the high cost of new silicon with the necessity of maintaining competitive AI capabilities.