The scandal is loud. The key-person risk is louder.
The Nvidia story looks, at first glance, like a scandal about Jensen Huang. It has everything a market drama needs: a founder-CEO, allegations involving extramarital relationships, a woman linked to a Singapore-based company under export-control scrutiny, angry investors, a petition for reinstatement, and even public comments from Donald Trump. On Monday, Trump said he “would certainly not mind” seeing Huang return, while Nvidia’s board was reported as not currently considering bringing him back .
But the most important question is not whether Huang can stage a comeback. It is whether Nvidia’s AI empire was too dependent on him in the first place.
That distinction matters because Nvidia is not an ordinary chip supplier. It is the central hardware platform of the generative-AI boom, a lender of credibility to young infrastructure companies, a political actor in Washington’s chip war with China, and a financial backstop for parts of the “neocloud” ecosystem. When Huang resigned last week, the market did not simply mark down a CEO scandal; it marked down a network.
Nvidia shares fell 4.1% on Monday, and the company has lost $735 billion in market value since the scandal around Huang’s conduct and China ties erupted, according to AP market reporting . The wider AI trade also weakened: the S&P 500 lost 1.9%, the Nasdaq dropped 2.5%, and CoreWeave, which buys Nvidia chips and rents computing access over the cloud, sank 6.1% . That is the market’s way of saying the issue extends beyond one executive’s behavior.
The board can remove a founder. It cannot instantly replace his role.
Huang’s resignation exposed a governance reality that had been easy to ignore during Nvidia’s ascent. Unlike some founder-led technology companies, Nvidia does not operate under a dual-class share structure that gives the founder supervoting control. Axios noted that Nvidia has a single class of common stock, while Huang held just 3.58% of shares as of March . In other words, the board could do what boards are supposed to do: accept or force the departure of a chief executive.
That is healthy governance in theory. In practice, it creates a second problem. A board can remove a CEO in a day, but it cannot reproduce three decades of accumulated trust, technical authority, customer access, geopolitical relationships, and market theater.
Huang was more than a chief executive. He was Nvidia’s salesman-in-chief to hyperscalers, sovereign AI projects, data-center developers and Wall Street. He was the person who could explain GPU roadmaps to engineers, reassure customers waiting for scarce chips, persuade financiers that massive AI infrastructure projects were bankable, and lobby governments that Nvidia’s interests aligned with national competitiveness.
That is why the “Stand with Jensen” campaign gained traction so quickly. AP reported that the petition attracted current and former Nvidia staff, investors and Silicon Valley supporters, including Elon Musk and Epic Games CEO Tim Sweeney . CBS reported that the petition had gathered more than 180,000 signatures, including more than 9,000 people identified by organizers as Nvidia employees . The numbers may not decide anything, but they reveal the fear: for many supporters, Huang was not just replaceable management. He was part of the product.
The real balance-sheet question: who supports the AI buildout now?
The deeper vulnerability is Nvidia’s role as financier and guarantor of the AI boom. The company’s chips are the foundation of AI training and inference, but its influence goes further. Nvidia has helped customers and partners fund data centers, expand capacity and convince lenders that AI demand will justify extraordinary capital expenditure.
Axios Macro reported that the Federal Reserve, for the first time, devoted part of its financial-stability assessment to AI financing, warning that the largest U.S. banks have nearly $1 trillion in exposure to technology companies, data-center developers and other businesses tied to AI infrastructure . The same report said companies borrowed more than $680 billion for data centers and advanced computing infrastructure in the first 10 months of 2026, about five times the amount raised in all of 2023 . It also highlighted more than $3 trillion in future lease payments, contracts and guarantees among seven large technology companies, including Nvidia, Microsoft and Oracle .
That is where Huang’s departure becomes a systemic issue. If Nvidia’s new leadership is more cautious, it may still sell every advanced GPU it can manufacture. But it may be less willing to use Nvidia’s balance sheet, guarantees, investments or informal influence to keep the infrastructure machine running at the same pace.
Axios reported that Nvidia has invested billions in AI companies that buy its chips, provided guarantees to help developers build data centers, and worked with Wall Street firms on a $500 billion AI infrastructure financing initiative that had not yet produced a deal . If that posture changes, the first companies to feel it are not necessarily Nvidia’s largest cloud customers. They are the more leveraged AI infrastructure players whose business model depends on buying enormous quantities of Nvidia hardware and renting it out.
That is why CoreWeave’s decline matters. It is not merely a sympathy move. It is a warning that the market is revaluing companies built around access to Nvidia chips, Nvidia confidence and Nvidia-linked financing.
Washington is now part of the succession plan.
The other reason this is bigger than the scandal is that Nvidia’s next CEO will inherit a geopolitical job. On Monday, the Trump administration issued new rules restricting remote access to advanced U.S. computing chips and servers, aimed at preventing China and other restricted countries from using data centers in third countries to get around export controls . AP reported that the rule adds end-user identification requirements for overseas data centers serving countries including China, Russia, Iran and Cuba, and adds licensing requirements for chip sales to several countries viewed as higher diversion risks, including Indonesia, Malaysia, Singapore, Thailand and Brazil .
That rule lands directly in the middle of Nvidia’s crisis. AP reported that Huang resigned after scrutiny of ties to a Chinese executive formerly at Singapore-based Megaspeed International, which is suspected of diverting advanced Nvidia chips to China through Southeast Asian data centers . Whether or not the most sensational allegations are ultimately proven, Nvidia’s leadership question is now tied to export controls, cloud access, congressional scrutiny and U.S.-China policy.
That makes a simple “bring Jensen back” narrative much harder. The board is not just answering to employees and shareholders. It is answering to regulators, lawmakers, intelligence concerns, customers and foreign governments. Even Trump’s support does not erase that pressure; AP reported that Treasury Secretary Scott Bessent had called for a “dramatic shake-up” and said Nvidia’s next CEO should come from outside the company .
Nvidia is still dominant, but dominance is not the same as resilience.
None of this means Nvidia is suddenly weak in technology. Its chips remain essential to AI data centers. Its software ecosystem, developer relationships and product roadmap still give it a formidable moat. The company’s upcoming earnings call, with interim CEO Colette Kress expected to address Wall Street, is now a test of whether investors can separate operational strength from leadership uncertainty .
The market’s fear is more precise: Nvidia may be dominant, yet less resilient than investors believed. The company has spent years convincing the world that the AI future runs through its GPUs. The past week has forced a different question: how much of that future also ran through one man?
The answer will determine whether this episode remains a scandal, or becomes a reset for the entire AI infrastructure trade. If Nvidia can show customers that chip roadmaps, delivery commitments, financing discipline and Washington relationships survive Huang’s exit, the sell-off may look overdone. If not, the scandal will have merely revealed the real problem: the AI economy’s most important company was built with too much human single-point-of-failure risk.
Even Skynet, after all, would need a team.

Comments
Be the first to comment.