The Computer That Dreamed of Pictures

The fluorescent lights of the Denny's on Story Road hummed with the particular frequency of all-night diners everywhere—that persistent, almost subliminal drone that has soundtrack to a thousand late-night conversations about ideas too big to wait until morning. It was April 1993, and three men sat in a corner booth, nursing cups of coffee that had long since gone cold, their heads bent together over a laminated menu they had no intention of ordering from. Outside, East San Jose stretched out in the particular darkness of a place that had seen better days, its strip malls and car dealerships dreaming their own dreams of commerce that might never arrive. But inside that booth, something was happening that would, in the fullness of time, reshape not just an industry but the entire technological landscape of the twenty-first century.

Jensen Huang was not yet thirty years old, but he carried himself with the gravity of a man who had already made peace with the possibility of failure. His dark hair, worn slightly longer than convention dictated, fell across a forehead that furrowed easily into expressions of intense concentration. When he spoke—and he spoke often, in rapid bursts of energy—it was with the conviction of someone who had spent years watching from the sidelines while lesser men made fortunes building things that didn't matter. Born in Taiwan in 1963, Jensen had immigrated to the United States at age five, his family joining the wave of Taiwanese professionals who saw in America not just opportunity but necessity. His father, a chemical engineer by training, had watched his own prospects diminish as Taiwan's economy shifted, and he resolved that his children would have the chances he had been denied. Jensen had absorbed this ambition like nutrition, letting it become part of his own metabolism. He had studied electrical engineering at Oregon State University, then earned his master's degree at Stanford—the very institution where, seventeen years earlier, two graduate students named Larry Page and Sergey Brin were still a decade away from their own conversation about organizing the world's information. After Stanford, Jensen had gone to work at LSI Logic, a chip design company, and then at AMD, where he had seen firsthand how the semiconductor industry worked. He had also seen its limitations.

Chris Malachowsky sat across from Jensen, his frame slumped against the vinyl booth in a posture that suggested either exhaustion or profound contemplation—or perhaps both. Where Jensen burned with an almost aggressive intensity, Malachowsky radiated a quieter, more contemplative energy. He had grown up in Wisconsin, the son of an electrical engineer father who had taught him to take apart radios and television sets before he was old enough to drive. By the time he graduated from the University of Florida, Malachowsky had developed an almost intuitive understanding of how electrons moved through silicon, how the invisible dance of charge carriers could be choreographed into computation. He had worked alongside Jensen at LSI Logic, and it was there that the two had discovered a shared frustration with the pace of innovation in the semiconductor industry. The big chip companies—Intel, AMD, IBM—moved like battleships, their massive organizational inertia preventing them from making the quick turns that the market seemed to demand. The two men had talked often about starting something of their own, but talk is cheap in Silicon Valley, and most conversations that begin in coffee shops end up as nothing more than stories people tell at reunions.

The third man, Curtis Priem, completed the unlikely trinity. Where Jensen and Malachowsky had met in the crucible of the semiconductor industry, Priem came from a different world entirely. He had graduated from Rensselaer Polytechnic Institute and then gone to work at Sun Microsystems, where he had spent years working on the graphics systems that powered Sun's workstations. At Sun, Priem had developed a deep appreciation for the mathematics underlying computer graphics—the complex transformations that turned abstract numerical data into the vivid images that appeared on screens. He had also developed something else: a conviction that the techniques being used in expensive workstations could, someday, be brought down to the level of ordinary personal computers. This was not a mainstream view. In 1993, the personal computer was still primarily a tool for productivity, a digital typewriter with delusions of grandeur. People used PCs to write documents, to manage spreadsheets, to organize their finances. The idea that these beige boxes might become entertainment devices, capable of generating the kind of visual experiences previously reserved for arcade games and military flight simulators, seemed far-fetched to many. But Priem had seen something that others had missed. He had watched the trajectory of semiconductor technology—the relentless doubling of transistors on a chip every two years, a phenomenon that would come to be known as Moore's Law—and he had extrapolated. If that trend continued, the graphics capabilities that currently required a $50,000 workstation would eventually fit on a chip that cost a few hundred dollars. When that happened, the personal computer would transform into something entirely new.

It was this insight that had brought the three men to Denny's on that April night. They had been meeting informally for months, sketching out ideas on napkins and envelopes, debating the technical specifications of a graphics chip that did not yet exist. The conversations had started cautiously, each man testing the waters, trying to gauge whether the others were truly committed or merely indulging in the kind of idle speculation that pervades every happy hour in every tech hub from San Jose to Bangalore. But gradually, as the months passed and the conversations grew more detailed, a conviction had formed among them. They believed they saw a window opening—a brief moment when the pieces might align to create something genuinely new. The personal computer was about to become a consumer entertainment device, and the missing piece was realistic graphics. Not the crude, blocky images that characterized early PC games, but the smooth, textured, lifelike visuals that had until recently been the exclusive province of military contractors and Hollywood visual effects houses. Someone was going to build the chip that made this possible. Why not them?

The decision to actually start a company is never as dramatic as the stories make it seem. There is no single moment of epiphany, no lightning bolt of inspiration that suddenly illuminates the path forward. Instead, there is a gradual accumulation of small steps, each building on the last, until the leap from employee to entrepreneur feels less like a choice than a inevitability. For Jensen, Chris, and Curtis, the final step came in the form of a conversation with a venture capitalist named Don Valentine.

Valentine was a legend in Silicon Valley, though his legend was of a particular kind—the kind earned not through the technical brilliance of the products he helped create but through the relentless, sometimes ruthless pragmatism of his investment strategy. He had founded Sequoia Capital in 1972, and over the two decades since, he had built it into one of the most successful venture capital firms in the world. His portfolio included Apple, Cisco, and Oracle—companies that had collectively reshaped the technology industry. Valentine was known for many things, but perhaps most of all, he was known for his intensity. He had a way of looking at entrepreneurs that made them feel as though he were reading their innermost thoughts, probing for the weaknesses that would inevitably cause their ventures to fail. He was not a warm man, not in the conventional sense. He did not pat backs or offer meaningless encouragements. When he met with entrepreneurs, he challenged them, tested