Jensen Huang resigned as Nvidia’s CEO on Thursday morning, ending a 33-year run atop the company he cofounded. The immediate trigger was an investigation into his relationship with an alleged Chinese intelligence asset, but the fallout extends far beyond one executive’s personal conduct. Nvidia’s market value has dropped over $1 trillion in less than a month since the scandal broke, and the company’s future direction on everything from open source AI to geopolitical strategy is now uncertain.

Colette Kress, Nvidia’s CFO since 2013, is running the company on an interim basis. She helped build Nvidia’s financial engine during its transformation from a gaming chip maker into a $2 trillion AI infrastructure company. She knows the numbers cold. What she hasn’t done is set strategic direction for an industry that Huang shaped almost single-handedly over the past three years.

President Trump weighed in hours after the resignation, posting on Truth Social that Tim Cook should take over. Cook stepped down as Apple’s CEO in September and moved to an executive chairman role, which apparently makes him available in Trump’s mind. There’s no indication Cook has any interest, and Nvidia’s board hasn’t commented on Trump’s suggestion. But the fact that a sitting president is publicly lobbying for his preferred CEO at America’s most strategically important chip company tells you how big this has gotten.

The open source question

Huang’s exit raises immediate questions about Nvidia’s $12.9 billion acquisition of Hugging Face, which closed just four months ago. Huang personally drove that deal as part of a broader push to position Nvidia as a champion of open source AI. The company released several open-weight models this year and Huang became one of the loudest voices against AI export restrictions.

Hugging Face CEO Clément Delangue said he’s committed to continuing that fight, but he’s not the one who controls Nvidia anymore. The company’s board could easily decide that supporting open source models that run on competitors’ hardware isn’t a great use of $12.9 billion. Or that weighing in on export policy is more trouble than it’s worth when you’re already under Congressional investigation.

Democrats on the House Oversight Committee launched their second investigation into Nvidia since the scandal broke, this one focused on the company’s China sales and what they called Huang’s “scheme” with Trump. The first investigation started after reports surfaced that Huang had been selling advanced chips to Chinese customers through murky licensing arrangements. None of that has been proven, but it doesn’t need to be proven to make Nvidia’s new leadership want to stay far away from geopolitical controversy.

What the industry loses

Huang wasn’t just Nvidia’s CEO. He set the direction for AI infrastructure across the industry. He decided to bet on transformer models years before ChatGPT made them obvious. He pushed CUDA as an AI development platform when it could have remained a gaming technology. He convinced hyperscalers to spend $50 billion a year on his chips by showing them what was possible.

You can see the uncertainty in how Silicon Valley responded to his resignation. Many tech leaders and Nvidia employees are still defending him, according to reports, apparently convinced this was all overblown. Some seem to be in denial that he’s actually gone. That reaction makes sense when you consider that Huang made a lot of people very rich and convinced them his vision for AI was inevitable.

But vision doesn’t matter much when you’re caught in a counterintelligence investigation. Reuters reported hours after the resignation that Huang had a second affair with a direct report, adding a more conventional corporate governance problem to the espionage concerns. Nvidia’s board didn’t have much choice.

The money

The market’s reaction has been brutal. Nvidia lost over $1 trillion in market value since The New York Times first reported that Huang was “close” to a suspected Chinese spy. Some of that would have corrected anyway after the company’s massive run-up. But this is different. Investors aren’t worried about a cyclical downturn in chip demand. They’re worried that Nvidia’s strategic advantages might have been compromised and that the company just lost the one person who understood where AI infrastructure was heading.

The timing is terrible. Nebius just announced plans to spend $40 billion building AI data centers, funded largely through customer prepayments. SoftBank lost over $3 billion preordering natural gas turbines to power AI infrastructure. These massive capital commitments assumed that Nvidia’s roadmap was solid and that Huang would keep delivering better chips on a predictable schedule. That assumption is now in question.

Kress can run the operations. She can manage the analyst calls and keep the supply chain moving. But can she tell customers what Nvidia’s chip architecture will look like in 2028? Can she decide whether to prioritize AI training chips or inference chips as the market shifts? Can she negotiate with TSMC when fab capacity gets tight? Those were Huang’s calls, and he made them based on instincts he developed over three decades.

Maybe Kress is great at all of that. Maybe Nvidia’s bench is deep enough that the next permanent CEO will be just as visionary. Or maybe the company is about to learn that certain advantages can’t be replaced when the person holding them gets forced out over an affair with a spy.