The Diner That Changed Everything

The morning sun poured through the windows of the International House of Pancakes on El Camino Real in Sunnyvale, California, casting long rectangles of light across laminate tables worn smooth by decades of coffee cups and butter knives. It was April 1993, and the Bay Area was experiencing one of those perfect spring days that made engineers wonder, briefly, whether they should be indoors at all. But the three men huddled in a corner booth had something far more compelling on their minds than the weather. They had $40,000 between them, a radical idea, and no guarantee that either would be enough to matter.

Jensen Huang sat on one side of the booth, his dark hair still damp from a morning swim—a habit he would maintain for decades, the water a brief sanctuary before the day's demands claimed him entirely. At thirty years old, Huang carried himself with an intensity that seemed to compress the air around him, as if he existed at higher pressure than ordinary mortals. His eyes, quick and restless, processed everything in the room: the tired waitress refilling coffees without being asked, the elderly couple sharing a short stack by the window, the particular way the light fell on the laminated menu. Jensen noticed details the way a predator notices movement—always assessing, always calculating, always prepared.

To his right sat Chris Malachowsky, a lanky Wisconsin native whose easy laugh and Midwestern politeness masked a fiercely intelligent engineer with an almost physical need to build things. Malachowsky had the kind of face that looked permanently surprised, as if the universe had been offering him a series of gifts he hadn't quite expected. He had left a comfortable position at Sega, the Japanese video game company, where he had been working on graphics technology that seemed impossibly advanced for its time. The pay was good. The work was interesting. But something was missing—or rather, something was calling.

Across the table, Curtis Priem completed the trio. Priem had the calm demeanor of someone who had learned to wait for the right moment, and then move decisively when it arrived. He had spent years at Sun Microsystems, watching the computer industry evolve from the outside, studying the patterns of success and failure with the patience of a chess player analyzing an opponent's strategy. Priem understood something that many of his contemporaries missed: the difference between a good idea and a *timely* idea. Timing, he had come to believe, was worth more than brilliance.

The waitress approached with a fresh pot of coffee. "Coffee, hon?"

"Coffee would be great," Jensen said, his voice carrying that peculiar blend of warmth and directness that would become his signature. "And could we get some napkins? Lots of napkins."

The waitress raised an eyebrow but didn't ask questions. She'd seen stranger requests at the IHOP on El Camino Real—this was Sunnyvale, after all, where the boundary between genius and madness was thinner than most people realized. She returned moments later with a stack of napkins and a pot of coffee, deposited them on the table, and retreated with the professional invisibility of good servers everywhere.

Jensen pulled a pen from his pocket—unremarkable, the kind of cheap ballpoint you'd find in any office supply store—and began to draw.

---

To understand what happened next, and why it mattered, we need to step back and consider the world as it existed in the spring of 1993. The personal computer revolution was barely a decade old. IBM had released its first PC in 1981, and the machine that would eventually become the industry standard was still so new that most people didn't own one. The internet existed, technically, but only in the primitive form of academic networks and government systems. The World Wide Web had been invented just two years earlier by Tim Berners-Lee at CERN, and fewer than a hundred websites existed in the entire world.

The technology that would eventually become the smartphone, the tablet, the smart speaker, the electric vehicle, and the AI that would reshape human civilization—all of it existed only as speculation in the minds of dreamers and mad scientists.

The video game industry, which would eventually become one of the most important economic and cultural forces on the planet, was still a niche hobby. Nintendo had released its 8-bit console, the Nintendo Entertainment System, just eight years earlier. Sega was competing fiercely with its Genesis system, offering "blast processing" and a edgier image designed to appeal to older players. The games of 1993—Mortal Kombat, DOOM, Myst—represented the cutting edge of interactive entertainment, and they were rendered in what we would now consider laughably primitive graphics.

A character in a 1993 video game was typically composed of a few thousand colored squares, or pixels. A single human face in a modern video game might contain more visual information than an entire game from that era. The shapes were blocky, the colors limited, the animations jerky. And yet these games represented miracles of engineering, the product of years of incremental improvement in both hardware and software.

The problem, the three men in the IHOP booth understood, was that all of this progress had been happening on the wrong axis.

---

At this point, we need to pause and explain a concept that will recur throughout this book: the difference between two-dimensional and three-dimensional graphics. Understanding this distinction is essential to understanding why NVIDIA would eventually become one of the most valuable companies in human history.

Imagine a piece of paper. You can draw on it—a house, a tree, a person. Everything you draw exists on the flat surface of the paper. It has height and width, but no depth. This is two-dimensional, or 2D, graphics. Every pixel in a 2D image is simply a colored dot placed at a specific location on a flat surface. When you play a classic Super Mario game, for example, Mario exists as a collection of these colored dots arranged to form his iconic shape. As he moves across the screen, the game is essentially showing you different collections of colored dots in different positions. The computer isn't thinking about Mario as a three-dimensional object existing in space; it's just manipulating a flat image.

Three-dimensional graphics, or 3D, works differently. In the real world, objects exist in three dimensions: height, width, and depth. A cube has length, width, and height. A sphere has a radius that extends in all directions from its center. Your own body exists in three dimensions, and you can walk around it, view it from different angles, see how light falls on it differently depending on where you're standing.

The challenge of 3D graphics is that computers, at their most fundamental level, think in two dimensions. A computer screen is flat. Every pixel on your monitor exists at a specific horizontal and vertical coordinate. So how do you take a three-dimensional world—a world with depth—and render it on a two-dimensional surface?

The answer involves mathematics that would have made Euclid weep and Descartes proud. To render a 3D scene on a 2D screen, you need to perform a process called projection. Imagine holding a flashlight behind a three-dimensional object and tracing its shadow on a wall. The shadow is a 2D representation of a 3D object. Computer graphics performs a similar mathematical operation, but instead of a flashlight, it uses virtual cameras and mathematical transformations to calculate how light would travel from virtual light sources through a virtual scene and onto a virtual screen.

The mathematics involved are staggering. For every single pixel on your screen, the computer must calculate which object in the 3D scene is visible at that pixel's position, what color that object should be based on the lighting in the scene, what texture is applied to that object, how far away the object is, and dozens of other