Nvidia founder Jensen Huang has resigned as CEO, with CFO Colette Kress set to take over on an interim basis. The company gave no reason for his departure. The move comes as Nvidia remains a leading supplier of AI chips, while governance questions, a US investigation involving China-related chip exports and customers’ in-house chip programs are putting its leadership and influence under fresh scrutiny.
Huang’s exit follows by only a few weeks his remarks at an event in London that Nvidia was willing to invest in startups. The company’s market value had at one point topped $5.7 trillion, while its business and investments expanded across AI labs, data-center operators and emerging cloud providers. Nvidia is now confronting uncertainty over its founder’s departure after being one of the biggest beneficiaries of the AI boom.
Governance questions and investigations
Before Huang resigned, scrutiny had focused on his relationship with Alice Huang and the controversy surrounding Nvidia customer Megaspeed. Alice Huang was previously an executive at the Singapore-based data-center company, which bought tens of billions of dollars’ worth of Nvidia chips. US intelligence agencies were reportedly concerned that she had links to Chinese intelligence and had sought information about US AI chips and pressed Nvidia to sell advanced chips to China. Nvidia’s board said it had no reason to believe company secrets or national-security information had been compromised.
In a memo to Nvidia’s 42,000 employees, Huang acknowledged being unfaithful to his wife and said he was deeply sorry. The board then opened an independent investigation. US lawmakers called for Huang to answer questions in person, and federal agencies said they were investigating. Separate allegations later emerged that Huang had an affair with an Nvidia executive who reported directly to him. The company’s resignation announcement did not address those reports.
A separate line of inquiry concerns whether Megaspeed helped Chinese companies evade chip-export restrictions. The US Commerce Department’s investigation into the company was reportedly paused before being referred to the Justice Department. An Nvidia spokesperson said the company had contacted officials to answer questions, share information gathered from its business operations and seek guidance. China called the allegations baseless speculation and slander. Nvidia’s announcement did not explain Huang’s departure or provide findings from the investigations.
China business and export-control debate
China has long been an important market for Nvidia. Since 2022, the US has restricted sales of the company’s most advanced AI chips to China; subsequent rule changes also affected less powerful models. Chinese authorities have, in turn, urged domestic AI companies to reduce purchases of US chips. Huang publicly opposed the export restrictions, saying Nvidia’s share of China’s AI-chip market had fallen from 95% to zero. He argued that US companies should compete in China to extend the reach of American technology.
In late 2025, Nvidia received permission to sell a less advanced chip to approved Chinese customers under an arrangement that included a 25% share for the US government. Ed Mills, a policy analyst at Raymond James, described Huang’s lobbying as effective and noted that support had emerged in Washington for allowing China to continue using US technology. The Megaspeed controversy therefore raises questions beyond Huang’s personal relationships, including export controls, customer due diligence and whether companies are doing enough to prevent chip diversion.
Nvidia’s investment network and financing commitments
Nvidia’s role in the AI industry now extends well beyond chip sales. As of the end of July, its investments in major AI model companies were close to $50 billion, while its total equity investments across the AI ecosystem were about twice that amount. The company has also used its investment-grade credit to help AI businesses and emerging cloud data-center providers secure financing. Those commitments total $170 billion, including a $105 billion guarantee for an OpenAI facility in Ohio. Nvidia has also discussed with a group of Wall Street lenders a plan worth as much as $500 billion to turn AI chips into tradable securities.
Critics say the arrangements could create a funding loop: Nvidia invests in AI companies and data-center operators, which then use the money to buy Nvidia chips. Such a structure could make underlying end demand harder to assess and deepen the industry’s reliance on a single supplier and its financing support. At the same time, Nvidia chips remain widely used, with supply sold out and orders booked through the following year. Scrutiny of governance and financing does not, by itself, mean demand for the company’s core products has disappeared.
Francine McKenna, an investigative journalist and expert in corporate governance and accounting, believes Nvidia can continue operating without Huang’s personal draw. Jay Goldberg, an analyst at Seaport Research Partners, cautioned that Huang’s willingness to take risks and keep pushing after projects failed was part of what set Nvidia apart. If that decision-making style fades with his departure, Goldberg said, the company could become a more conventional chipmaker. The contrast highlights what succession means for investors: the question is not only who takes over the top job, but whether Nvidia can maintain its pace of investment and product decisions.
Customers build alternatives to Nvidia chips
Nvidia also faces efforts by major customers to develop alternatives. Google uses its own TPUs to train Gemini; Amazon is advancing its Trainium chips; and OpenAI is designing chips with Broadcom. Anthropic is increasing its use of Google TPUs, AMD GPUs and Amazon Trainium, while continuing to use Nvidia products. As AI coding tools make it easier to move workloads between chip platforms, the technical cost of switching suppliers may also fall.
National “sovereign AI” programs are adding to the push for a more diverse chip supply. Countries including the UK are seeking to develop domestic chips and exploring ways to combine different types of processors in the same data center. The shift matters to Nvidia: large cloud providers and hyperscalers account for about half of its data-center revenue, while data-center sales make up more than 90% of the company’s total revenue.
Huang had promoted Nvidia’s offerings in Europe, arguing that the region could run a range of models, including France’s open-source Mistral models, on Nvidia chips. The company has also continued to support open-source AI, seeking to expand the developer ecosystem that uses its GPUs and CUDA software. Whether that strategy can continue to support customer purchases after Huang’s departure will be watched alongside chip availability, corporate investment and the incoming management team’s decisions.