Huang’s exit leaves Nvidia’s AI economics under sharper scrutiny

By Tom Whitfield

Nvidia’s 63-year-old founder and CEO Jensen Huang resigned Thursday, and its stock fell 5.6%. The board named vice president and chief financial officer Colette Kress interim CEO. The exit followed an extramarital affair with Alice Huang, a former executive at Singapore-based data center company MegaSpeed International, and a member of Huang’s top staff.

U.S. intelligence briefed President Donald Trump in October 2025 about concerns that Alice Huang was a spy for China’s Ministry of State Security. MegaSpeed and connected companies acquired billions of dollars of AI equipment containing Nvidia chips in 2023 and 2024. Jensen Huang and Alice Huang are not related.

Huang co-founded Nvidia in 1993 and led it since, through its rise as the world’s most valuable company during the AI boom. His departure creates governance uncertainty at the leading AI-chip producer, but analysts expect little operational disruption. The lasting risk runs through credibility around AI spending and valuations.

After the November 3 U.S. midterms, Nvidia and AI data-center stocks fell sharply before rebounding. Thursday’s 5.6% decline came after Wednesday’s partial recovery.

The board’s move may test enterprise confidence in AI spending and valuations. It may give buyers more room to choose another supplier.

What does Huang’s exit change for enterprise buyers?

“Little, for now,” R “Ray” Wang, founder of Constellation Research, said. “They’ve got a product roadmap. They’ve got products that are going to come out, and so they have to execute on the roadmap.”

He said Nvidia will probably remain financially sound, although Huang could face serious problems with the U.S. government over the alleged spying relationship. Investors also no longer need Nvidia to thrive for the wider AI trade to rise.

“Nvidia is still the leader in the space,” Wang said. “It gives people the freedom to go choose someone else if they have the product.”

Gold Associates president Jack Gold expects more leverage through AMD, Intel and smaller rivals. “If you want to buy Nvidia systems in the future, they may not be the most available and there will likely be more competition in this space from AMD, Intel and others,” he said.

Gold also expects more competition in networking and some movement away from Nvidia’s ubiquitous CUDA development platform. “It means that the numerous partnership agreements in place with Nvidia, while not in any immediate jeopardy, might change as companies become more cautious about the uncertainty,” he said.

Nvidia continues to fund Anthropic, OpenAI and major neocloud vendor CoreWeave. It also backs infrastructure operators, including 49-year-old Oracle, which use that support to buy Nvidia hardware and rent it into the AI market.

Their data centers contain Nvidia equipment, while Anthropic and OpenAI have committed to buying access to it. Hyperscalers and cloud providers keep buying Nvidia GPUs, the preferred hardware for generative AI.

David Nicholson, chief advisor at Futurum Group, said the tumult does not change Nvidia’s position as a major investor in AI vendors, cloud providers and neocloud companies. But he described the market as an “intertwined and incestuous” circle with Nvidia at its centre.

“The cycle works only while AI users keep buying,” Nicholson said. “It all hinges upon whether the end users, enterprise customers … are going to continue to do so.”

AI spending enters price discovery

Kashyap Kompella, founder of RPA2AI Research, framed the loss as a credibility problem at a moment of rising skepticism. “The AI industry has lost one of its most credible public advocates at a time when skepticism about AI is growing,” he said.

Kompella added: “Huang was very effective at connecting technological innovation to a broader economic narrative. His departure creates a vacuum just as the industry needs to strengthen its social license to operate.”

Kompella also placed the exit inside the U.S. chip export-control debate with China, America’s chief AI rival. “Huang’s departure also removes an influential voice from the export-control debate, where national-security concerns and long-term technological competitiveness are increasingly in tension,” he said.

Anthropic’s IPO, expected next week by most accounts, carries a lower-than-expected valuation suggesting an AI infrastructure pullback. Investors will test whether explosive revenue growth covers compute costs and long-term capital commitments.

“Anthropic’s IPO could become a defining price-discovery moment,” Kompella said. “Investors will look beyond explosive revenue growth to scrutinize sustainable profitability. Compute costs, infrastructure commitments and ongoing capital requirements will come under greater scrutiny.”

Investors have also lowered expectations for OpenAI’s IPO, expected next year. Enterprises struggle to justify returns from staggeringly expensive AI systems. Nicholson said Nvidia’s turmoil could prompt some customers to retreat, though he disputed claims about the chipmaker’s valuation.

“A lot of people are going to point to what’s going on with Nvidia right now as being the primary mover of Nvidia’s valuation, and I would just say that’s nonsense,” he said.

Open-source AI models from China and, increasingly, the West can undercut proprietary models from Anthropic, OpenAI and Google. That creates an awkward irony for Huang and Nvidia, both advocates of open source as a democratic, transparent approach.

Nvidia recently acquired the Hugging Face developer platform, extending that reach. Enterprise buyers can test lower-cost models against Nvidia services before renewing infrastructure commitments.

Nvidia’s product engine may keep running. The capital cycle now depends on customers believing returns will last.