Overview
Nvidia designs the chips that most of the world's artificial-intelligence systems are trained and run on. It does not own factories; it designs graphics processing units (GPUs), networking hardware and software, and has them made by contract manufacturers, chiefly TSMC. In the fiscal year to 25 January 2026 it reported revenue of $215.9 billion, of which $193.7 billion came from data centers, according to its fiscal 2026 results. In the quarter to 25 October 2026 it reported $115.8 billion in a single three-month period.
November 2026 also brought the biggest change in its history. Co-founder Jensen Huang, chief executive for 33 years, resigned, and the company is now run on an interim basis by its finance chief.
History
Huang, Chris Malachowsky and Curtis Priem founded Nvidia in April 1993, sketching the plan in a Denny's in East San Jose. Their bet was that personal computers would need dedicated chips for 3D graphics, with video games supplying the demand. The first product, the NV1, went against the graphics standard Microsoft was setting and flopped. By 1996 the company had cut deeply and was close to running out of money. A frank conversation with Sega, which agreed to honour a payment even though Nvidia was abandoning the design it had ordered, bought time to ship the RIVA 128 in 1997.
Nvidia listed on the stock market in January 1999 and later that year launched the GeForce 256, which it branded the first "GPU". In 2006 it released CUDA, software that let programmers use GPUs for general calculations rather than only graphics. For years investors saw CUDA as an expensive side project. That changed in 2012, when the AlexNet image-recognition model, trained on Nvidia cards, showed how well GPUs suited neural networks.
The data center became the company's centre of gravity from the mid-2010s. The $6.9 billion purchase of Mellanox, completed in 2020, added the high-speed networking that links thousands of GPUs together. A planned takeover of the chip designer Arm was abandoned in 2022 after regulators objected. The launch of ChatGPT in late 2022 turned steady growth into a scramble for Nvidia hardware. The company's market value passed $1 trillion in 2023 and $5 trillion in October 2025.
How the business works
Nvidia sells computing systems, not just chips. A modern AI cluster combines its GPUs, its own CPUs, networking switches and cables from the Mellanox line, reference server designs and a deep layer of software. The buyers are mainly the cloud giants (Microsoft, Amazon, Google, Oracle), AI labs, large companies and, increasingly, governments.
The moat is the software. CUDA and the libraries built on it have been refined for almost two decades, and millions of developers write code for it. Switching to a rival chip means rewriting and retuning that work, so customers tend to stay even when competitors offer cheaper hardware. That lock-in supports gross margins of over 70%. In the third quarter of fiscal 2027 Nvidia made GAAP net income of $65.3 billion on $115.8 billion of revenue, according to its results filing.
The risks are concentration and politics. A handful of hyperscale customers account for a large share of sales, and several of them are designing their own AI chips. US export controls limit what Nvidia can sell to China; its current guidance assumes no China data-center revenue at all.
Ownership and leadership
Nvidia is listed on Nasdaq as NVDA and is owned by its shareholders, mostly institutions. Huang remains one of its largest individual shareholders. On 11 November 2026 the board accepted his resignation as president, chief executive and director, effective immediately, and appointed chief financial officer Colette Kress as interim CEO and Harvey C. Jones as the board's chair, according to the company's SEC filing. The filing gives no reason. The board is searching for a permanent chief executive and, on 17 November, said it did not intend to reinstate Huang.
Controversies
Crypto disclosures (PROVEN, settled). In 2022 the SEC found that Nvidia had failed to disclose how much cryptocurrency mining was driving its gaming sales in 2017–2018. Nvidia paid a $5.5 million penalty without admitting or denying the findings.
Megaspeed and Huang's resignation (ALLEGED). The New York Times reported in October 2026 that US intelligence believed Alice Huang, a former executive at Singapore-based Megaspeed, had been recruited by Chinese intelligence to get close to Jensen Huang, who later acknowledged an affair with her. Megaspeed, a large Nvidia customer, is under investigation over the alleged diversion of restricted chips to China; it denies this, and Nvidia says it found no evidence of diversion and stopped sales to Megaspeed in June 2025. Beijing denies she is a spy. A shareholder class action, Kistenmacher v. NVIDIA, alleges investors were misled. None of these claims has been proven, and Nvidia's board has said it has no reason to believe company secrets or national security were compromised.
Antitrust (ALLEGED). In September 2025 China's market regulator announced a preliminary finding that Nvidia breached its anti-monopoly law in connection with the Mellanox deal. No final decision or penalty has been published.
Why it matters
Nvidia is the clearest modern example of a company that spent years building for a market it could not precisely predict, then owned that market when it arrived. It is also now a test of whether a founder-shaped company can keep its edge without its founder. The record quarter reported under Kress suggests the business is, for now, bigger than any one person.
Read the founder's story in Jensen Huang: The Thirty-Year Overnight Success, our analysis of the Jensen Huang Nvidia scandal, and what the AI build-out means for consumers in who pays for AI data centers' electricity.