Jensen Huang resigned as CEO of Nvidia on November 12, and roughly $1 trillion in market value evaporated with him. The company's capitalization fell from about $5 trillion on November 3 — when it was the most valuable company in the world — to $4 trillion by the morning of his departure. Huang's own net worth had dropped $7.4 billion by that point. The resignation came a day after reports of a second affair, and under mounting pressure from the board.

The affair that started the unraveling involves Alice Huang (no relation), also known as Huang Le, a senior executive at Megaspeed International, a Singapore-based company that became an Nvidia customer. U.S. intelligence officials reportedly suspect her of being a Chinese spy; the Chinese government has denied it. The timing of the relationship's start remains unclear — a gap that matters, given what followed.

What did the chip smuggling allegations involve?

Megaspeed reportedly smuggled Nvidia chips into China, bypassing U.S. sanctions first imposed in 2022 over concerns China could use the chips for military development and AI competition. Alice Huang reportedly helped move $7.5 billion worth of Nvidia chips — around 240,000 banned units, mostly H100 and H200 models — into China over the course of the relationship.

The New York Times first reported the relationship and the spy allegations on October 29. Huang initially described it as an ordinary business relationship. On November 6, the Financial Times reported it was romantic. Two days later, Huang admitted the affair in an email to employees. On November 11, reports emerged of a second affair, with an Nvidia employee — reportedly Kimberly Powell, vice president of health care. He resigned the next day. Notably, neither Nvidia nor Huang has publicly addressed how a sanctions-bypassing channel operated under the CEO's watch for years.

Why does one resignation shake the AI industry?

Nvidia sits at the center of the AI boom as the world's largest supplier of chips for AI data centers, controlling an estimated 80% of that market. A large data center might hold tens of thousands of its GPUs (graphic processing units, the chips that run the calculations behind AI). Training a new AI model can involve up to 100,000 GPUs running simultaneously for months; afterward, the same chips handle "inference" — using the trained model to generate responses. Around 60 million Americans use AI chatbots daily, according to a Pew survey this summer. Nvidia also supplies CPUs (central processing units, which coordinate tasks and run operating systems) and the networking equipment that lets GPUs communicate.

Huang cofounded the company in 1993 with Chris Malachowsky and Curtis Priem, originally making GPUs for video games. A 2006 platform opened the chips to other uses, and in 2012 researchers used them to train AlexNet, a breakthrough AI program for image recognition. By the time ChatGPT launched, Nvidia's technology was fundamental to AI development.

Can anyone replace him?

Probably not quickly. Huang involved himself in decisions most executives delegate, from product construction details to how salespeople spoke with customers. A typical CEO has five or 10 direct reports; Huang had as many as 60, and reportedly sent hundreds of emails a day. He had an unusual level of control — and no succession plan. CFO Collette Kress is interim CEO while the board searches.

What does this change for Nvidia's operations? Strategic decisions on products, manufacturing, and customers could slow. Huang oversaw how the full system was integrated, not just GPUs, and his successor may lack the same product vision. Customer relationships might deteriorate; internal struggles for control are possible. None of this is certain — but none of it is ruled out, either.

Who else is exposed to the fallout?

Nvidia's business model is notably circular: it sells to the same companies it finances. It committed $30 billion to OpenAI's financing round and guaranteed up to $105 billion for an Ohio data center serving OpenAI. It invested $2 billion in AI infrastructure company CoreWeave in January and agreed to purchase up to $6.3 billion of CoreWeave's unused computing capacity, alongside similar cloud-computing purchase deals with other customers. It has partnered with Apollo, BlackRock, and other asset managers to mobilize AI infrastructure financing.

Anthropic is tightly entangled as well. Nvidia committed up to $10 billion to the company in 2025; Anthropic committed to use up to 1 gigawatt of Nvidia computing capacity. Nvidia had reportedly weighed billions in credit guarantees for a new Anthropic data center, plus an investment of up to $10 billion in Anthropic's IPO, which the company had planned to file next week.

All of this propped up demand for Nvidia's own technology, making it harder to gauge how fast AI would have grown on underlying economics alone. Huang personally spearheaded those investments. Whether his successor will finance customers as freely is unknown — and that uncertainty alone could slow the AI industry's planned expansion.