Two Arrested for Smuggling Nvidia AI Chips to China Amid US Export Crackdown
The U.S. Department of Justice (DOJ) announced on Tuesday the arrest of two Chinese nationals, Chuan Geng and Shiwei Yang, for their alleged involvement in the illicit shipment of high-performance artificial intelligence chips valued at tens of millions of dollars to China. The arrests, which took place in California on August 2, led to charges under the Export Control Reform Act, a felony carrying a potential sentence of up to two decades in prison.
Geng and Yang stand accused of knowingly and deliberately exporting “sensitive technologies,” including powerful Graphics Processing Units (GPUs)—the specialized processors crucial for advanced AI computations—from the United States to China. These shipments were allegedly routed through their California-based company, ALX Solutions. While the DOJ’s complaint did not explicitly name the chip manufacturer, it described the smuggled components as “the most powerful chip in the market” and “designed specifically for AI applications,” a description that strongly points to Nvidia’s high-performance GPUs. A Reuters report subsequently confirmed the chips in question were indeed Nvidia’s H100 GPUs.
An examination of export documentation by the DOJ revealed that ALX Solutions dispatched chips and other technological hardware to shipping and freight-forwarding companies in Singapore and Malaysia. In return, payments for these transactions were received from entities based in Hong Kong and mainland China. Investigators also uncovered communications indicating a deliberate intent to route the technology through Malaysia specifically to circumvent U.S. export restrictions.
Responding to the arrests, an Nvidia spokesperson asserted that “smuggling is a nonstarter.” The company stated it primarily sells its products to established partners, including Original Equipment Manufacturers (OEMs), who assist in ensuring compliance with U.S. export control regulations. Nvidia emphasized that even relatively small exporters and shipments undergo rigorous review and scrutiny, and any products diverted from legitimate channels would be ineligible for service, support, or updates.
This incident unfolds against a backdrop of complex U.S. policy challenges, as Washington endeavors to balance the promotion of global AI innovation with the imposition of stringent export restrictions on China—a nation many in the West perceive as a formidable competitor in the burgeoning AI race. The recently unveiled AI Action Plan by the Trump administration underscored the necessity of robust export controls but offered limited specific details on implementation.
Amidst these policy discussions, the U.S. government has floated the concept of integrating tracking technology into chips as a potential deterrent against smuggling. However, this proposition has met with significant resistance from chipmakers. Nvidia, in a blog post, vehemently opposed the idea of embedding “kill switches” or “backdoors” into its GPUs. The company argued that such features would inherently compromise security, effectively creating vulnerabilities that could be exploited by hackers and hostile actors.
“Nvidia has been designing processors for over 30 years,” the company stated, asserting that “embedding backdoors and kill switches into chips would be a gift to hackers and hostile actors.” It further contended that such measures would undermine global digital infrastructure, erode trust in U.S. technology, and contradict established legal principles that mandate companies fix, rather than create, vulnerabilities. Nvidia concluded that such a policy would be an “overreaction” that would “irreparably harm America’s economic and national security interests.” The ongoing debate highlights the intricate tension between national security imperatives and the foundational principles of technological trust and innovation in a globalized economy.