
Tech • AI • Robotics • Game
Nvidia remains dominant in artificial intelligence chips, but Jensen Huang’s abrupt resignation has exposed how much of the company’s client allocation, financing support and political access depended on one person.
Nvidia said on November 12 that founder Jensen Huang had stepped down as chairman, chief executive and director with immediate effect. The stock fell a little over 4%, but larger losses hit companies that buy Nvidia chips and rent them out, including declines of about 9% for CoreWeave and 16% for Nebius. Interim leadership passed to chief financial officer Colette Kress while the board searches for a successor.
US authorities are examining ties involving Megaspeed, a Singapore-based Nvidia customer that installs advanced systems in Southeast Asian data centers and rents the computing power remotely, including to Chinese groups. Trade data cited in recent reporting showed more than 17,000 GB300 processors imported into Malaysia in November 2025 alone, worth nearly $800 million. The US Department of Commerce has been investigating since 2024, and the matter was referred to the Justice Department, though officials have also said no conclusions on espionage had been reached and no charges had been filed.
For three years, demand for AI processors has outstripped supply, giving Nvidia unusual control over customer selection. The company favored newer cloud-computing groups such as CoreWeave, Nebius and Asian operators including Megaspeed, helping create alternatives to giant customers like Microsoft, Google, Amazon and Meta, all of which are developing in-house chips. That approach protected Nvidia from becoming dependent on a handful of buyers capable of eventually bypassing it.
Nvidia’s structure reportedly had more than 60 people reporting directly to Huang, a setup designed for speed in a fast-moving market. Investors interpreted his departure as a sign that the informal decision-making behind chip allocation may now slow or change. That matters most to customers whose businesses relied on preferential access rather than to Nvidia itself, whose underlying demand remains strong.
Since July 1, Nvidia has committed in some cases to re-rent unused computing capacity from cloud partners at pre-agreed prices, effectively backstopping lenders. That support helped underpin a financing platform of about $500 billion, with Nvidia able to guarantee up to $125 billion, while CoreWeave alone had roughly $35 billion in debt at the end of June. Investors now fear that already signed commitments will stand, but future guarantees may be harder to secure under an interim chief executive facing investigations, lawsuits and congressional scrutiny.
Huang was also central to Nvidia’s relations with Washington and Beijing. After a meeting with Donald Trump in July 2025, Nvidia secured special licenses to sell the restricted H20 chip in China. The White House later allowed the more powerful H200 under a revenue-sharing arrangement with the US government. Those permissions were widely seen as built on long-standing personal trust, something no successor can easily replicate.
Despite the leadership shock, Nvidia’s business remains formidable. The company keeps more than 70% gross margin on each chip sale, supported by CUDA, the software ecosystem on which much of the AI industry depends. That leaves Nvidia with what many in the market view as a toll-booth position for large-scale AI over the next two to three years, even if confidence in side agreements and guarantees has weakened.
Nvidia’s biggest customers have spent years reducing dependence. Google runs major workloads on its TPUs, Amazon has Trainium, Microsoft has Maia, and Meta is developing its own chips. OpenAI has signed custom-chip deals with Broadcom for more than 10 gigawatts and with AMD for another 6 gigawatts. In China, regulators have also limited the domestic entry of Nvidia’s H20 despite shortages, apparently preferring local scarcity to deeper dependence on American technology.
The reaction to Huang’s fall has also highlighted resentment around an AI boom that has created vast wealth while offering uncertain benefits to much of the public. In recent local votes in the US, opposition to data centers has intensified, with municipalities moving to block projects and some officials losing office over support for them. That backlash suggests the industry’s political challenge is no longer only export controls or financing, but also whether ordinary users and communities feel they share in AI’s gains.
Nvidia is unlikely to lose its core dominance immediately, but Jensen Huang’s exit has shaken the web of personal trust that helped allocate scarce chips, unlock financing and secure political approvals. The deeper question is whether the company can preserve that influence long enough to stay ahead of customers and governments already building ways around it.
Ask a question