The federal indictment landed like a thunderclap. On March 20, 2025, the U.S. Department of Justice charged Wally Liaw — co-founder of Super Micro Computer — with orchestrating what prosecutors describe as a massive scheme to smuggle approximately $2.5 billion worth of Nvidia GPUs to China and other restricted destinations, evading export controls designed to keep America’s most advanced AI chips out of adversaries’ hands.
The case isn’t just about one man or one company. It’s about the enormous black-market demand for high-end AI processors, the fragility of U.S. export controls, and the lengths to which individuals will allegedly go to exploit the gap between Washington’s policy ambitions and its enforcement capacity.
According to the indictment unsealed in the Northern District of California, Liaw used a network of shell companies, falsified end-user certificates, and transshipment routes through Southeast Asia to funnel restricted Nvidia H100 and A100 chips to entities in China between 2020 and 2024. The charges include conspiracy to violate the Export Control Reform Act, money laundering, and wire fraud — carrying potential penalties of decades in federal prison, as reported by Slashdot.
Liaw, 65, co-founded Super Micro Computer alongside Charles Liang in San Jose in 1993. The company grew into a major server and storage solutions provider, eventually becoming one of the primary assemblers of AI server racks powered by Nvidia’s most advanced graphics processing units. Super Micro went public in 2007 and saw its stock price surge dramatically during the AI boom of 2023 and 2024, as demand for GPU-dense server configurations exploded worldwide. But Liaw had departed the company years earlier, and Super Micro has stated it is not a target of the current investigation.
That distinction matters — but only to a point. The case raises uncomfortable questions about supply chain oversight across the entire AI hardware industry.
Here’s what prosecutors allege happened. Liaw established or directed several entities in Taiwan, Singapore, and Malaysia that served as intermediary purchasers for Nvidia GPUs. These companies would place orders with authorized distributors, representing that the chips were destined for customers in permitted countries. Once the hardware arrived at warehouses in Southeast Asia, it was repackaged and rerouted to buyers in mainland China — including, according to the indictment, organizations with ties to Chinese government research institutions and military-adjacent AI labs.
The scale is staggering. $2.5 billion in chips. That’s not a rounding error or a handful of misrouted shipments. It represents tens of thousands of the most powerful processors on the planet, each one capable of training and running the large language models and other AI systems that the U.S. government has explicitly tried to deny to China.
The Biden administration imposed sweeping export restrictions on advanced semiconductors bound for China in October 2022, with subsequent tightening in October 2023. The rules specifically targeted Nvidia’s A100 and H100 GPUs, along with other high-performance chips from AMD and Intel. The rationale was straightforward: these processors are the engines of modern AI development, and allowing unrestricted access would accelerate China’s military and surveillance capabilities. Nvidia responded by designing downgraded chips — the A800 and H800 — specifically for the Chinese market, but even those were eventually restricted.
The restrictions created a colossal arbitrage opportunity. Nvidia’s top-tier GPUs, already in short supply due to surging global demand, became even more valuable on the Chinese gray and black markets. Industry analysts have estimated that H100s were trading at two to three times their list price in China throughout 2023 and into 2024. The financial incentive to circumvent controls was enormous, and apparently irresistible to some.
Liaw’s alleged operation was sophisticated. Prosecutors describe a layered system of corporate entities, each designed to obscure the ultimate destination of the chips. One Singapore-based company named in the indictment reportedly maintained legitimate contracts with data center operators in Southeast Asia, providing cover for the volume of GPU purchases. Another entity in Taiwan handled logistics and shipping documentation, generating falsified end-user certificates that listed fictitious or complicit companies in Malaysia and Thailand as final recipients.
Money moved through a parallel network. The indictment details wire transfers routed through banks in Hong Kong, Singapore, and the Cayman Islands, with funds ultimately traceable to accounts controlled by Chinese purchasing agents. Prosecutors say Liaw personally directed many of these transactions and received tens of millions of dollars in payments funneled back through the same circuitous financial channels.
None of this happened in a vacuum. The Commerce Department’s Bureau of Industry and Security, which administers export controls, has been struggling with enforcement since the restrictions took effect. A February 2025 report from the Center for Strategic and International Studies noted that the BIS had only about 400 employees dedicated to enforcement — a skeleton crew given the scope of the controls and the sophistication of evasion tactics. The Liaw case appears to be the largest single prosecution to emerge from the government’s efforts to crack down on GPU smuggling, but officials have signaled it won’t be the last.
“This case should send a clear message,” Attorney General Merrick Garland said in a statement accompanying the indictment. “We will pursue and prosecute anyone who undermines our national security by illegally exporting controlled technology to our adversaries.”
But will it? The enforcement challenge is immense.
Consider the numbers. Nvidia shipped an estimated 550,000 H100 GPUs in 2023 alone, according to analysts at New Street Research. Each chip is roughly the size of a paperback book. The sheer volume of global semiconductor trade — hundreds of billions of dollars annually — makes tracking individual chips from factory to final installation extraordinarily difficult. Once a GPU leaves an authorized distributor’s warehouse, the chain of custody becomes murky fast, particularly when intermediaries in multiple countries are involved.
Super Micro itself has had a turbulent recent history that adds context to the case, even though the company insists it’s not implicated. In August 2024, short-seller Hindenburg Research published a report alleging accounting irregularities at the company, which led to a delayed annual filing, the resignation of its auditor Ernst & Young, and a temporary threat of Nasdaq delisting. Super Micro eventually brought in a new auditor, filed its overdue reports, and stabilized its stock price — but the episode highlighted governance weaknesses that had persisted for years. The company had previously paid a $17.5 million fine to the SEC in 2020 to settle charges of widespread accounting violations between 2015 and 2017.
Liaw’s departure from Super Micro predates these recent troubles, and there’s no public evidence linking the company’s accounting issues to the smuggling allegations. Still, the overlap of characters and the company’s history of compliance failures create an unavoidable narrative thread. When a co-founder of a major AI hardware company is accused of running a multi-billion-dollar smuggling ring involving the very products that company sells, the industry takes notice.
Nvidia, for its part, has consistently maintained that it complies with all U.S. export regulations and cooperates with government investigations. The company has implemented a “know your customer” program and requires end-user certifications for sales of restricted chips. But critics argue these measures are insufficient. A November 2024 investigation by Reuters found that Nvidia GPUs were readily available for purchase in China through intermediaries in Southeast Asia and the Middle East, often at significant markups. The supply chains were complex but hardly invisible to anyone who looked.
The Liaw prosecution also arrives at a moment of broader geopolitical tension over AI technology. The Trump administration, which took office in January 2025, has signaled an even more aggressive stance on technology exports to China. Commerce Secretary Howard Lutnick has publicly discussed expanding the entity list — the roster of foreign organizations barred from receiving U.S. technology — and tightening restrictions on chip manufacturing equipment sold to Chinese foundries.
Meanwhile, China has been racing to develop domestic alternatives to Nvidia’s GPUs. Huawei’s Ascend 910B processor has emerged as the leading Chinese-made alternative, though industry benchmarks suggest it still lags behind Nvidia’s latest offerings in both performance and energy efficiency. The Chinese government has poured billions into semiconductor self-sufficiency programs, but closing the gap on leading-edge AI chips remains a multi-year endeavor at minimum.
Which is precisely why the smuggling market thrives. Every month that China lacks domestic equivalents to the H100, the demand for illicitly obtained Nvidia chips persists. And the profits are extraordinary — markups of 100% to 200% over list price create financial incentives that dwarf the perceived risks of prosecution, at least until cases like Liaw’s demonstrate real consequences.
The defense bar is watching this case closely. Liaw’s attorneys have not yet made detailed public statements, but legal experts expect them to challenge the government’s characterization of the transactions and potentially argue that some of the chips were destined for legitimate commercial uses in permitted countries. Export control law is notoriously complex, and proving that a defendant knew — or should have known — that goods would be diverted to a restricted destination requires detailed evidence of intent.
Prosecutors appear confident in their evidence. The indictment references intercepted communications, financial records obtained through international cooperation agreements, and testimony from cooperating witnesses — including at least one individual described as a former employee of one of Liaw’s intermediary companies who is now cooperating with the government.
The case also has implications for the broader semiconductor distribution chain. Authorized Nvidia distributors — companies like Arrow Electronics, Avnet, and WPG Holdings — face increasing pressure to implement more rigorous screening of their customers. Several distributors have already tightened their compliance programs in response to government pressure, but the Liaw indictment suggests that determined bad actors can still penetrate these defenses, particularly when they have deep industry knowledge and established relationships.
So where does this leave the industry? In an uncomfortable place. The AI hardware supply chain is global, complex, and under enormous demand pressure. Export controls are only as effective as the enforcement apparatus behind them, and that apparatus remains underfunded and outmatched by the scale of the challenge. Individual prosecutions, however dramatic, address symptoms rather than structural vulnerabilities.
The $2.5 billion figure in the Liaw indictment represents perhaps 1% to 2% of total Nvidia GPU revenue during the period in question. If this single alleged operation moved that volume, the total amount of restricted chips reaching China through all channels — gray market, black market, and everything in between — could be substantially larger. Some industry estimates, including figures cited by analysts at Bernstein Research, suggest that as much as 10% to 15% of Nvidia’s data center GPU output may have found its way to Chinese buyers through various indirect channels since the export controls took effect.
Nvidia’s stock, which had been on a historic run fueled by AI infrastructure spending, dipped modestly on the news of the indictment but recovered quickly. Investors appear to view the case as a law enforcement matter rather than a direct threat to the company’s business. But the longer-term risk is regulatory. If the government concludes that voluntary compliance measures are insufficient, mandatory tracking requirements — perhaps even chip-level serialization and geolocation monitoring — could be imposed on manufacturers and distributors. Such measures would add cost and complexity to an already strained supply chain.
For now, Wally Liaw faces the full weight of the federal government’s prosecution machinery. His initial court appearance is scheduled for April 2025 in San Jose. If convicted on all counts, he could face up to 50 years in prison, though sentences in export control cases typically run much shorter. The government has also moved to seize assets allegedly connected to the scheme, including bank accounts in Singapore and real estate in California and Taiwan.
The case is a marker. Not the first GPU smuggling prosecution, and certainly not the last. But its scale — $2.5 billion, a company co-founder, a network spanning four countries — makes it the most significant test yet of whether the United States can enforce the technology controls it has staked so much of its China strategy on. The answer, so far, is mixed at best.
WebProNews is an iEntry Publication