The story in full
In 1996, a three-year-old chip company in California was running out of money. Its first product had flopped, and the chip it was building for a video game giant was heading in the wrong direction. Its founder later said that, whichever way he turned, the company would be out of business. For years afterwards he opened staff presentations with the same line: "Our company is thirty days from going out of business."
Thirty years later, that company is worth around $5 trillion and its chips power the AI boom. This is the history of Nvidia.
Three engineers and a Denny's
The story starts in a Denny's diner on Berryessa Road in East San Jose. Three engineers met there to plan a new company: Chris Malachowsky and Curtis Priem from Sun Microsystems, and Jensen Huang from the chip company LSI Logic. Huang knew the place well: he had washed dishes and waited tables at Denny's when he was younger.
The pitch was simple. Personal computers were about to get 3D graphics for gaming and multimedia, and that would need a new kind of chip. According to Priem, Huang refused to join unless the business could make $50 million in sales a year. Nvidia was founded on 5 April 1993 with about $40,000 in the bank. Investors including Sequoia Capital followed, and Nvidia raised about $20 million in venture funding.
The chip that flopped
Nvidia's first chip, the NV1, arrived in 1995 and was sold in shops as the Diamond Edge 3D card. It tried to do everything: graphics, sound, even ports for Sega Saturn controllers. But it drew 3D images using curved shapes instead of the triangles almost everyone else was moving to. When Microsoft built its DirectX graphics system for Windows around triangles, the NV1 was left on the wrong side of the industry. It didn't sell well.
Thirty days from going out of business
Nvidia still had one big deal: Sega had hired it to develop the graphics chip for the console that became the Dreamcast. But that chip, the NV2, used the same curved-shape approach. After about a year, Huang concluded it wasn't going to work. If Nvidia finished it, the company would fall behind the industry; if it stopped, it would lose the money keeping it alive.
So Huang told Sega the truth. He asked Sega's American boss, Shoichiro Irimajiri, to let Nvidia out of the contract and to pay anyway. Irimajiri said yes, and Sega put about $5 million into Nvidia. "His understanding and generosity gave us six months to live," Huang said. Nvidia also laid off about half of its staff. (Accounts differ on whether Sega paid the contract in full or bought shares.)
One shot
With six months of money left, Nvidia bet the company on one new chip, the RIVA 128. This time it drew with triangles and worked with DirectX. A new chip normally went through several rounds of fixes; Nvidia couldn't afford a second. "We only got one shot," Huang recalls. So the team worked backwards, writing the software in advance and testing the design on an emulator before the chip physically existed. The RIVA 128 came out in 1997, and Nvidia sold more than a million in its first four months.
Inventing the GPU
On 22 January 1999 Nvidia floated on the Nasdaq at $12 a share. That August it announced the GeForce 256 and marketed it as "the world's first GPU", or graphics processing unit (a claim that is still debated). For the next decade Nvidia was best known for graphics cards for PC gamers.
The bet nobody asked for
To draw a game, a GPU does thousands of small sums at the same time, and some scientists realised they could use that for other work. In November 2006 Nvidia launched CUDA, which let programmers write ordinary code that ran directly on the graphics chip. For years it looked like a bet with no obvious pay-off.
Then in 2012, Alex Krizhevsky, Ilya Sutskever and Geoffrey Hinton at the University of Toronto entered the ImageNet image-recognition contest. Their system, AlexNet, was trained in five to six days on two Nvidia GTX 580 gaming cards and beat the field by a huge margin. The AI researchers had found their engine, and it was a graphics card.
The AI machine
In August 2016 Huang personally delivered the first DGX-1 AI computer to a young San Francisco lab called OpenAI. In November 2022 OpenAI launched ChatGPT, built on Microsoft supercomputers that strung together tens of thousands of Nvidia chips. Suddenly every big tech company wanted as many Nvidia chips as it could get. On 30 May 2023 Nvidia became the first chip company worth $1 trillion, and on 29 October 2025 the first company ever worth $5 trillion.
Where Nvidia is now
Remember the $50 million a year Huang wanted at Denny's? In the year to January 2026, Nvidia's revenue was almost $216 billion. Size brings scrutiny: its roughly $40 billion deal to buy Arm was called off in February 2022 after a US Federal Trade Commission lawsuit, and in September 2025 China's market regulator said a preliminary investigation found Nvidia had broken its anti-monopoly law (the regulator's claim, not a final ruling). Jensen Huang is still chief executive, more than thirty years after that meeting at Denny's.
Key dates
- 5 April 1993: Nvidia is founded.
- 1995: the NV1 launches and flops.
- 1997: the RIVA 128 sells more than a million in four months.
- 22 January 1999: Nvidia floats on the Nasdaq.
- November 2006: CUDA launches.
- 2012: AlexNet is trained on two Nvidia gaming cards.
- 29 October 2025: Nvidia becomes the first company worth $5 trillion.
Three lessons
- Admitting you're wrong early can save you. Nvidia survived because Huang told Sega the truth.
- When you only get one shot, plan backwards from the finish line.
- The biggest wins can come from markets you didn't build for. Nvidia made chips for gamers and ended up powering AI.
Accurate as of 4 November 2026. Figures such as valuations and revenue move over time. This is an independent editorial history, not affiliated with or endorsed by any company mentioned. The full source list is in the YouTube description and pinned comment.


