Top-Tier Nvidia Chips Found in Chinese State Network, Raising Questions on Export Controls
A security firm has detected banned high-performance graphics processors inside the infrastructure of a major Chinese telecom, prompting Nvidia to argue that such leaks are minor compared to the country's total computing power.
By Adewale Fletcher ·
For tech enthusiasts and business leaders tracking the global AI race, a new discovery suggests that the flow of high-end computing power into China may be harder to stop than previously thought. An Israeli cybersecurity company named Lava has reported finding 32 units of Nvidia’s Blackwell Ultra B300 graphics processing units within the network of China Telecom. This state-owned telecommunications provider is a key player in the country’s digital infrastructure. The presence of these specific chips is notable because they have already been barred from export by United States regulations, which restricts sales to only those buyers who are on an approved list and promise not to sell to sanctioned parties.
The B300 is one of the most powerful data center GPUs currently available. It was announced in March of last year and offers significant performance improvements over its predecessor, the B200. Specifically, the B300 provides 50% more memory and double the FP4 computing capacity. Nvidia described the chip as being built for the age of reasoning, highlighting its utility for advanced large language models. Given these capabilities, it is no surprise that the chip became highly sought-after after its launch, with expectations for it to ship by the end of 2025.
Lava identified the chips using Nvidia’s own DCGM Exporter tool. This software is designed for data center operators to monitor server health and performance. While it is intended for internal use, the tool can often be accessed from outside the data center’s local network. In some cases, it does not even require password authentication, which may have facilitated the discovery. It remains unclear exactly how the chips arrived in China Telecom’s server racks or their precise location within the network, but their presence has been confirmed through this monitoring software.
Nvidia has responded to the finding by downplaying its significance. A spokesperson stated that the diversion of controlled U.S. products has been found to be less than one-half of one percent in recent court cases. The company described this amount as a drop in the bucket compared to the vast amount of domestic computing power China already possesses. With tens of millions of GPUs in service in approved countries, Nvidia argues that smuggling is a non-factor, noting that governments on both sides of the globe are actively working to block such transfers.
This incident is part of a broader pattern of attempts to move high-performance chips into China despite export controls. In 2025, reports indicated that Chinese companies smuggled $1 billion worth of Nvidia AI chips into the country in just three months. At that time, the B200 was the most desired model, but distributors were already preparing for the B300 series. The latest finding underscores the continued challenge of preventing the proliferation of export-controlled technology.
The smuggling of these chips involves complex logistics. In August 2026, nine individuals were indicted in Taiwan for the illegal export of B300 GPUs to China. This legal action occurred shortly after Supermicro, a major server maker, fired several employees following an internal investigation into smuggling activities. The process reportedly involved a Taiwanese server trading firm called Flying Tiget Tech, which held whitelisted status for GPU purchases, adding a layer of complexity to the supply chain.
The high demand for Nvidia’s silicon has had a ripple effect on the broader hardware market. Even older chips, such as the A100 servers that are five years old, have seen their prices triple, with some units costing up to $82,000. This scarcity has led some organizations to retool gaming GPUs to run AI workloads where possible, highlighting just how vital these processors have become for modern computing infrastructure.
While the discovery of 32 chips may seem small in the grand scheme of a national network, it highlights the persistent gap between regulatory intent and operational reality. As AI models become more sophisticated and resource-intensive, the fight over access to top-tier hardware continues to be a central theme in the global technology landscape. Nvidia maintains that the overall impact remains manageable, but the incident serves as a reminder that export controls are an ongoing battle rather than a finished task.
