THE AFTERMATH: Jensen Huang's resignation this morning ends Nvidia's 33-year founder era, but it doesn't end the story. There's a search for a successor with no obvious candidate, an espionage case now touching four governments, a congressional hearing, and a growing roster of people offering their own picks for who should run the world's most important AI chipmaker.
Nvidia moved quickly to project continuity. In an internal memo, longtime director Harvey Jones told employees he would serve as the company's first-ever board chair. Jones, a former Synopsys CEO, has sat on Nvidia's board since its founding in 1993. The fact that Nvidia never had a separate chair until today says a lot about how much authority sat with one person.
CFO Colette Kress, at Nvidia since 2013, will run the company on an interim basis while keeping her finance role. CNBC reports she is gathering the executive team at the Santa Clara HQ to talk leadership and continuity. Her first real test: Nvidia is set to report quarterly results in just a few days, on November 17.
Meanwhile, investors are not waiting and the immediate reaction has been that Nvidia shares opened down more than 13%, briefly dropping the company below a $4 trillion market cap, before partially recovering by midday. Nvidia is still on track for six straight weekly losses, its longest losing streak since 2012. Apple has reclaimed the title of world's most valuable company. Wall Street and everyone else is split on what comes next.
No succession plan
Bloomberg reports that Nvidia has hired an outside firm to find candidates, but replacing Huang won't be easy. For three decades, Huang combined deep technical knowledge with tight operational control, running the company with a famously flat structure and dozens of direct reports.
Kress has become the obvious continuity pick, with the expectation that she will hold the seat for about a year before an outsider takes over. Among insiders, the name that comes up most is Ian Buck, who runs Nvidia's hyperscale and HPC business and is closely tied to CUDA's origins.
Ben Bajarin called him the most logical choice, though Bloomberg notes he lacks the operating experience typically expected of a CEO. Analysts have also floated chief scientist Bill Dally, CTO Michael Kagan, and 32-year veteran Jeff Fisher. Huang's children, Madison and Spencer, both hold management roles at the company, but even observers who mention them question whether either is ready.
The outside list is a who's who of chip leadership: Arm CEO and former Nvidia executive Rene Haas, Intel's Lip-Bu Tan, Apple hardware chief Johny Srouji, and AMD's Lisa Su, who happens to be Huang's first cousin once removed and runs Nvidia's most credible rival. Some analysts have even suggested a co-CEO setup, pairing a technical leader with an operator.
President Trump has also got a favorite. He declared that only Tim Cook could fill Huang's shoes, adding: "A Legend replaces a Legend!"
Elon Musk, meanwhile, is recruiting. Hours before the resignation, while reports of a possible ouster were circulating, a user on X asked Musk to hire Huang. Musk replied: "That would be an honor." He also endorsed a post arguing Nvidia's board should wait for its investigation to conclude before acting.
Is it also game over for GeForce?
That list should worry anyone who buys graphics cards. Look closely and you'll notice most of those names built their careers in data centers, finance, mobile chips, or CPUs. None of them co-founded a company in 1993 to make 3D graphics for PC games.
Huang did, and for all of Nvidia's pivot to AI, he never fully let go of that. Even after rebranding Nvidia as an AI infrastructure company, he kept reminding audiences that GeForce is what brought CUDA into the world, and that the AI boom wouldn't exist without it. At GTC in March, he called the modern Nvidia "the house that GeForce made."
Nvidia's gaming business hit a record $16 billion in the company's last fiscal year, but it was still a rounding error next to the data center. Then this year, Nvidia stopped reporting gaming separately and folded it into a new "Edge Computing" bucket alongside workstations, consoles, robotics, cars, and telecom gear. In the most recent quarter, that entire group brought in $7.2 billion, versus $89 billion from the data center. That's a mere 7.5% of revenue for everything Nvidia sells outside a data center and the earnings release didn't mention GeForce or gaming once.
Under Huang, GeForce arguably survived in part on sentiment, the founder's attachment to the business that built the company.
Gamers have been feeling the shift for a couple of years now. As AI swallowed the world's memory supply, Nvidia reportedly cut GeForce production by as much as 40% in the first half of the year. Kress herself (now interim CEO) warned investors that supply constraints would weigh on gaming going forward. It was widely known that no new gaming GPUs would come out this year, which would be the first such gap in roughly three decades.
Under Huang, GeForce arguably survived in part on sentiment, the founder's attachment to the business that built the company. A successor without that history will look at the same spreadsheet and see a comparatively unattractive consumer business competing with data center chips for the same scarce memory and TSMC wafers. From there, the options range from benign neglect to a spin-off or outright sale.
Then again, GeForce is still the entry point to CUDA for many developers and students. Consumer and data center GPUs share architectures, so much of the R&D carries over.
Nvidia also supplies the chip inside Nintendo's Switch 2. Untangling all of that would be messy. A more realistic risk isn't a sale but a slow fade: fewer launches, longer gaps between generations, and graphics card pricing dictated by how much memory is left after the data center business has taken its share. (If the AI bubble goes bust, that's an entirely different conversation.)
Whoever takes the job, how much attention Nvidia gives GeForce over the next couple of years will be telling.
The spy story keeps getting stranger
The Wall Street Journal is reporting a most surreal detail about the whole spy saga (yet): Huang was introduced to Alice Huang by Kang Wei, a Chinese businessman and former star fund manager he'd known for about a decade. Kang was also Alice Huang's husband, according to people who know the couple.
Huang and Kang reportedly first crossed paths in the 2010s, when Kang's firm Shanghai Hexi was investing in video game companies and Nvidia was mostly a gaming chip business. For roughly two years, Kang appeared to accept his wife's relationship with the Nvidia CEO, at times joking to associates that she worked as Huang's assistant.
The Journal describes the couple as serial dealmakers who moved from games to pharmaceuticals to cloud computing. In 2023, they renamed a Singapore company Megaspeed International, with Alice Huang as a director. Using Alibaba as a promised anchor customer, Megaspeed secured access to chips as an Nvidia partner.
According to the WSJ's sources, Alice Huang and Kang Wei pitched investors on new data center projects, promising to triple their money in three years and expressing confidence they could get Nvidia chips into China.
People familiar with the matter say most of the servers went to Malaysian data centers serving Alibaba, but some were shipped into China. Megaspeed says it followed all export control rules, but the WSJ has sources who say the couple pitched investors on new data center projects, promising to triple their money in three years and expressing confidence they could get Nvidia chips into China.
The New York Times is also reporting today that Singapore's government has just disclosed it held Alice Huang's passport from June 2025 to February over potential false declarations tied to Megaspeed. She was eventually released for lack of evidence, given a stern warning, and banned from re-entering the country.
What happens now with the "central bank of AI"
Beyond Nvidia's own stock, the GPU maker became more than a chip supplier under Jensen: it is an investor, financier, and loan guarantor to a huge slice of the AI industry, with obligations CNBC estimates at around $400 billion.
To name a few such data points: shares of Nvidia-backed neoclouds dropped today, with CoreWeave down about 8%. Big commitments include the $12.9 billion Hugging Face acquisition, which is under review, and a $10 billion share purchase tied to Anthropic's IPO next week.
But none of this changes Nvidia's product roadmap. Vera Rubin is shipping, demand remains strong, and while AMD stock has climbed ~30% in the last few days, there's no way AMD could magically absorb Nvidia's customers even if they all switched tomorrow.
AMD isn't the only alternative, either. Google is now selling its TPUs to outside customers like Anthropic and Meta instead of only renting them through its cloud, and Amazon has signed up Anthropic and OpenAI for gigawatts of Trainium chips. Intel remains the question mark. Nvidia even owns a stake in it after a $5 billion investment last year, and its CEO is on some analysts' list to run Nvidia.
For 33 years, Nvidia's succession plan was essentially Jensen Huang. It now needs a real one.
