Jen-Hsun “Jensen” Huang was born in Taiwan and emigrated to the US with his family when he was nine. He first moved to Kentucky and eventually to a suburb of Portland, Oregon. Jensen studied electrical engineering at Oregon State and then at Stanford. After his studies, he started as a staff engineer, rapidly became a director at LSI Logic, and then worked for AMD and gained a reputation for being able to solve challenging problems. On his 30th birthday, he decided to start NVIDIA.

When he pitched Don Valentine, Jensen tells us that he had not come up with the “accelerated computing” category or the term “GPU” and couldn’t describe NVIDIA’s customers, market, or market size. Further, he thought he would build for the video game industry since Sequoia funded Electronic Arts, but Don derided the idea because startups can’t depend on other startups. Nonetheless, Don told Jensen that Sequoia would invest a million dollars, and Jensen would probably lose it, but Don would kill him if he did. Jensen’s storytelling was hilarious, but what gave Don so much conviction to invest in Jensen and NVIDIA?

Here is the backstory, according to Michael Moritz. Wilf Corrigan, the founder of LSI logic, had been a longtime jousting partner of both Don Valentine and Pierre Lamont. Wilf was a tough British nut, schooled as an engineer, emigrated to America, worked in the chip industry, and eventually wound up at Fairchild. So when he formed LSI logic at the start of the 1980s, Don had been his first call, and Sequoia became a shareholder. LSI logic may have been the best-returning Sequoia investment before Cisco. When Wilf called to alert Don that Jensen, whom Wilf considered the smartest guy in his company, was leaving to start NVIDIA, Don immediately jumped on the opportunity to meet Jensen.

In this era, many startups developed chipsets for personal computers, NVIDIA, S3, and Neomagic. Neomagic and S3 pulled ahead but are footnotes in technology history. On the other hand, it was difficult for NVIDIA out of the gate. NVIDIA’s first chip was a graphics accelerator, but it missed the market window, and the company was on its heels. Jensen salvaged the business, which was just one of the potential near-death experiences for NVIDIA.

NVIDIA was a small business until some noticed that GPUs worked beyond games. The entertainment industry would embrace GPUs for various special effects in their movies, but it would take computer scientists at Pixar and ILM to showcase the GPUs power in media before it was widely used in special effects. Then, AI researchers noticed that GPUs are extremely good for parallel processing and for the problems they were studying. This cycle has been underway for the last decade.

One of the critical insights Jensen developed was that he no longer wanted to build chips for specific use cases, such as games or even special effects. Instead, he raised the scale of his ambition and wanted to create a generalized computer that accelerated computing for general problems. This approach is very different from building a chip for image recognition to improve ImageNet or a specific implementation of AI transformers to speed up OpenAI’s ChatGPT.

To do this, NVIDIA solves complex problems foundational to electrical engineering and computer science. For example, NVIDIA builds its own research data centers to understand how to optimize its products. In addition, NVIDIA develops its acceleration packages, software applications, and even LLMs to understand how its products are being pushed to the boundaries.

Jensen shared that NVIDIA has accelerated computing by a millionfold in the last decade, and he speaks confidently that another million-fold is possible. We can’t wait to live in that future.