Spleenlab Partners With Uxs Alliance To Forge Ai-Driven Autonomous Dominance
Spleenlab Joins the UXS Alliance: Powering the AI Brain Behind Next-Gen Autonomous Missions
In a …
25. July 2025
Nvidia Gpu Smuggling Scandal: Billion-Dollar Heist Exposed Amid Trump-Era Export Controls
The Great NVIDIA Heist: $1 Billion Worth of GPUs Smuggled to China Amid Trump-Era Export Controls
The Financial Times has uncovered that at least $1 billion worth of NVIDIA GPUs have been smuggled to China despite US President Donald Trump’s efforts to tighten export controls on the tech giant. This brazen operation highlights the cat-and-mouse game between governments and companies in the high-stakes world of AI hardware exports.
NVIDIA’s B200 systems, based on the company’s high-end Blackwell architecture, are widely used by leading AI companies such as OpenAI, Google, Meta, Microsoft, and others. The direct sales of this chip are prohibited in China due to US government’s strict export controls aimed at curbing the country’s growing military capabilities.
However, FT’s investigation revealed that a significant portion of these chips has found its way into Chinese hands through illicit means. A prominent Chinese company is suspected of being involved in the smuggling scheme, and packaging and installation pictures show product logos of reputable hardware providers such as Supermicro, Dell, and Asus.
Supermicro, a US-based hardware infrastructure provider, is at the center of this operation. According to FT’s analysis, most B200 racks sold by a Chinese company originated from Supermicro. The involvement of these hardware providers raises questions about their due diligence and adherence to export regulations.
AliCloud, owned by Alibaba Group, has also been implicated in the smuggling scheme. This company has received significant shipments of B200 systems from Supermicro, despite US government’s export controls. AliCloud has reportedly used more than 100 business partners each, many of whom are believed to be receiving NVIDIA GPUs through illicit channels.
ByteDance’s Huoshan Cloud, another major AI player in China, is also accused of receiving NVIDIA GPUs through unauthorized channels. As one of China’s largest AI companies, ByteDance has faced scrutiny over its use of foreign technology in developing its cutting-edge AI capabilities.
The combination of factors contributed to this success story, including the widespread use of B200 systems in various hardware configurations, making it difficult to track, and NVIDIA’s high-end Blackwell architecture. Additionally, China’s complex web of middlemen and distributors has allowed these chips to be diverted into the black market.
In response to this development, President Joe Biden’s administration announced plans to strengthen export controls on sensitive technology, including AI hardware. The US government has approved the sale of modified H20 chip versions to comply with new regulations, which were previously deemed too powerful for export to China.
This move is seen as a strategic attempt by the Biden administration to balance its commitment to national security while promoting responsible trade practices. The incident highlights the complex and often fraught nature of international trade in sensitive technologies. As the global AI landscape continues to evolve, it is essential that governments and companies work together to strengthen regulation and prevent the illicit flow of sensitive technology.
The smuggling of NVIDIA GPUs to China raises critical questions about the effectiveness of US export controls and the ease with which sensitive technology can be diverted into the black market. While this incident highlights the need for greater vigilance, it also underscores the complexities and challenges involved in regulating international trade in advanced technologies.
As the world grapples with the implications of AI-powered innovation, it is essential that governments, companies, and regulators work together to promote responsible trade practices while ensuring national security interests are protected.