Long before Nvidia was a household name on trading screens, it was an idea sketched out over coffee in a Denny's booth in East San Jose. In 1993, Jensen Huang, then 30 and working at chip designer LSI Logic, met there with two engineers from Sun Microsystems, Chris Malachowsky and Curtis Priem. Their bet: personal computers would one day need dedicated chips to draw 3D graphics, and video games would be the market that paid for it.

The setting was fitting. Huang, born in Taiwan and sent to the United States as a child, had worked at Denny's as a teenager — washing dishes and busing tables. He went on to study electrical engineering at Oregon State and later Stanford, and designed microprocessors at AMD and LSI Logic before deciding to start his own company.

The near-death years

Nvidia's first product, the NV1, shipped in 1995 and bet on a rendering approach the rest of the industry did not adopt. When Microsoft's DirectX standard went another direction, Nvidia was left with a chip that didn't fit the market and a dwindling bank account. Huang has said the company's unofficial motto for years was: "Our company is thirty days from going out of business."

The company cut staff, refocused, and staked what was left on a single chip: the RIVA 128, released in 1997. It sold strongly and bought Nvidia time. Two years later, in January 1999, Nvidia went public, and that same year launched the GeForce 256 — which it marketed as the world's first "GPU."

"Our company is thirty days from going out of business." — Jensen Huang, describing Nvidia's early years

The bet nobody asked for

The most consequential decision came in 2006 with CUDA, a platform that let developers program Nvidia's graphics chips for general-purpose computing. For years it looked like an expensive science project. Then in 2012, a neural network called AlexNet — trained on two Nvidia gaming GPUs — won the ImageNet competition by a wide margin, and researchers realized GPUs were the ideal engine for deep learning.

Nvidia had spent six years building the software ecosystem nobody else had. When the generative-AI wave arrived a decade later, it was the default supplier. In 2023 its market value passed $1 trillion, and in July 2025 it became the first company to pass $4 trillion.

What founders can take from it

  • Survive first. Nvidia's best ideas only mattered because the company outlived its early mistakes.
  • Build the platform, not just the product. CUDA was a decade-long investment in developers, and it became the moat.
  • Pick markets that pay for the R&D. Gaming funded the chips that later powered science and AI.