TAIPEI, July 8 — The calculus of artificial intelligence dominance has a new vulnerability: the hardware itself. Jensen Huang, the co-founder and CEO of NVIDIA, is confronting a problem that signals just how high the broader stakes have become in the global AI arms race. The company’s H100 GPU, a $30,000 component essential for training the most advanced large language models, has become a prime target for thieves.
The H100 GPU, launched in 2022, is the engine powering the AI revolution. It is the chip that trains systems like OpenAI’s GPT-4.
Under Huang’s leadership, NVIDIA’s market capitalization surged past $2 trillion by 2024, a reflection of the insatiable demand for this hardware. In 2024 alone, NVIDIA shipped 2 million H100 units. Yet supply constraints persist, and that scarcity has created a black-market incentive.
The U.S. Department of Energy’s March 2025 report documented the scale of the problem: 500 H100 GPUs were stolen from data center construction sites in 2024. The thefts, valued at $15 million, represent a direct assault on the infrastructure that underpins the AI sector.
Huang has addressed the issue. In NVIDIA’s 2024 annual report, he stated that the company is working with law enforcement to track stolen chips. Huang’s personal trajectory provides context for the company’s strategic response.
Born in 1963 in Tainan, Taiwan, he moved to the United States as a child. He co-founded NVIDIA in 1993 and has served as its CEO ever since. His background in electrical engineering — a B.S. from Oregon State University and an M.S. from Stanford University — shaped the technical foundation that allowed the company to pivot from graphics cards to AI accelerators.
The IEEE Founders Medal in 2020 recognized his contributions to the field. The theft of the H100 GPUs is not an isolated incident but a symptom of a larger geopolitical reality.
The hardware is the bottleneck. With supply constrained and demand from AI companies relentless, every chip that disappears from a construction site represents a delay in training cycles, a setback in model deployment, and a potential advantage lost to competitors — whether corporate or state-sponsored. The U.S. Department of Energy’s report underscores the vulnerability of the physical supply chain.
The thefts occurred at data center construction sites. The H100 GPU retails for $30,000 each, but its true value lies in what it enables: the training of models that can process language, generate code, and analyze data at scale.
Losing 500 units is not just a $15 million loss; it is a loss of computational capacity that could have been used to advance research or military applications. Huang’s public commitment to working with law enforcement reflects an understanding that the problem extends beyond corporate security. The U.S. Department of Energy’s report, issued in March 2025, provided the first official accounting of the scale of the thefts.
The figure of 500 GPUs stolen is based on reported incidents at data center construction sites. NVIDIA’s response has been methodical. The company is working with law enforcement to track stolen chips.
Huang, now in his third decade at the helm of NVIDIA, has navigated the company through multiple technological shifts. The transition from graphics processing to AI computing was the most consequential.
Under his leadership, the H100 became the industry standard for training large language models. The demand is so high that even shipping 2 million units in 2024 could not satisfy it. The thefts represent a new kind of challenge.
They are not a technical problem to be solved with a better algorithm or a faster chip. They are a security problem, rooted in the physical world.
And they require a response that combines corporate resources with government authority. The broader stakes are clear. The hardware that enables AI capabilities is now a target.
Huang’s commitment to securing the supply chain is not just a business necessity. What to watch next is whether the thefts escalate or decline. If the tracking and law enforcement collaboration prove effective, the number of stolen chips could drop.
If not, the vulnerability may force changes in how data centers are constructed and secured. The H100 may be a marvel of engineering, but its value has made it a liability.
Huang’s response will determine whether that liability is contained.
{“@context”: “https://schema.org”, “@type”: “NewsArticle”, “headline”: “NVIDIA CEO Jensen Huang’s H100 GPUs Prime Targets for AI Thieves”, “description”: “TAIPEI, July 8 — The calculus of artificial intelligence dominance has a new vulnerability: the hardware itself. Jensen Huang, the co-founder and CEO of NVIDIA,…”, “datePublished”: “2026-07-08T09:30:00+00:00”, “author”: {“@type”: “Person”, “name”: “Noah King”}, “publisher”: {“@type”: “Organization”, “name”: “InfoPulse Today”, “url”: “https://newsfeed.infopulsetoday.com/”}}
