Smuggled Nvidia Chips Found in China Telecom
Israeli security firm identifies 32 export-controlled B300 GPUs within a major Chinese state telecom network.
By Brianna Jackson ·

Israeli security firm Lava has identified 32 Nvidia Blackwell Ultra B300 graphics processing units operating within the network of China Telecom, a state-owned telecommunications provider. The discovery is significant because U.S. export controls have recently barred these specific chips from being sold to Chinese buyers. According to a report by The Information, the chips are restricted to whitelisted buyers who agree not to export them to sanctioned parties, including those based in China. The presence of these units in a Chinese data center raises questions about how they bypassed these regulatory frameworks.
This incident highlights a persistent challenge in enforcing export bans on high-performance AI hardware. In 2025, reports indicated that Chinese companies smuggled approximately $1 billion worth of Nvidia AI chips into the country over a three-month period. While the B200 was the most sought-after model at that time, distributors were already preparing for the arrival of the B300 series. The latest findings confirm that the proliferation of export-controlled chips into China remains an ongoing issue despite intensified scrutiny.
Nvidia has downplayed the significance of the report, though it did not dispute the findings. An Nvidia spokesperson stated that governments, researchers, and media have found only "insignificant diversion of controlled US products, less than one half of one percent in recent court cases." The company described this volume as a "drop in the bucket compared to the ocean of domestic compute China already has." With tens of millions of GPUs in service in approved countries, Nvidia maintains that smuggling is a non-factor, largely blocked by regulatory actions on both sides of the globe.
The exact method by which the B300 chips entered China Telecom's server racks remains unknown. Lava also could not confirm the precise location of the chips within the company's broader network. However, the firm identified the hardware using Nvidia's DCGM Exporter tool. This software is typically installed by data center operators to monitor server health and performance. Although designed for internal use, the tool can often be accessed from outside the data center's network and frequently does not require password authentication. Nvidia has acknowledged working on a dedicated tool to locate its chips, but it has not yet been released.
Chinese firms are highly motivated to acquire the Blackwell Ultra B300 due to its superior capabilities. Announced in March of last year, the data center GPU offers substantial performance gains, including 50% more memory and FP4 compute power compared to the previous-generation B200. Nvidia marketed the chip as "built for the age of reasoning," citing its utility for sophisticated AI large language models such as DeepSeek R1. Originally expected to ship by the end of 2025, the B300 quickly became a highly sought-after asset in the AI market.
The demand for these chips has driven intense illicit activity. In August 2026, nine individuals in Taiwan were indicted for the illegal export of B300 GPUs to China. This legal action followed days after Supermicro confirmed it had fired several employees following an internal investigation into smuggling activities. The complex smuggling process reportedly involved a Taiwanese server trading firm called Flying Tiget Tech, which held whitelisted status for GPU purchases, allowing it to legally acquire the hardware before it was moved to China.
The scarcity of Nvidia's advanced silicon has also impacted the broader market, causing prices for older hardware to skyrocket. A100 servers, despite being five years old, have tripled in price and can cost up to $82,000. The ongoing crackdown on smuggling has led some operators to retool gaming GPUs to run AI workloads where possible. As the market continues to tighten, the balance between export controls and global demand for compute power remains a critical focus for both technology companies and regulators.
Source: Tom's Hardware


