It’s been over 32 years since Jensen Huang, Chris Malachowsky, and Curtis Priem started a new company together. It even took them a while to come up with a name, but they eventually settled on Nvidia. Yes, that little, independent, underdog company. The one worth $5 trillion back in October. That level of fortune gives one plenty of opportunities to spend hours reminiscing about the past, present, and future, which is precisely what Jensen Huang and Joe Rogan recently did.
Specifically, for almost two and a half hours, Huang chatted on the Joe Rogan Experience about all things AI, GPU, CUDA, consciousness, and what have you. There’s an awful lot to delve into, but I was particularly drawn to the section where Rogan remarks on Nvidia’s massive AI growth and capabilities: “How funny would it be that it [AI] is birthed out of the desire for computer graphics for video games? It’s kind of crazy when you think about it that way.”
This prompted Jensen to start going over Nvidia’s beginning, when it was a tiny group of engineers in 1993: “We were trying to create this new computing approach. The question is, what’s the killer app? We wanted to create a new type of computing architecture, a new type of computer that can solve problems that normal computers can’t solve.”
A noble goal, one might say. However, Huang hits the issue with this idea right on the head. “Well, the applications that existed in the industry in 1993 are applications that normal computers can solve, because if the normal computers can’t solve them, why would the application exist?
“And so we had a mission statement for a company that has no chance of success, but I didn’t know that in 1993; it just sounded like a good idea, right?”
(Image credit: Nvidia)
If you’re not familiar with Nvidia’s early history, the application it eventually settled upon was 3D graphics rendering. More importantly, the kind of rendering that Sega was doing with its arcade units, e.g. Virtua Fighter. I thoroughly recommend that you grab a copy of The Nvidia Way: Jensen Huang and the making of a tech giant, by Tae Kim, as it’s richly detailed and goes into great depth about how Nvidia did more than just gain inspiration from Sega.
Here’s a key snippet: “Amid the business of the trade show, the Nvidia team managed to secure an introduction to representatives from the Japanese video-game and console maker Sega. Impressed with the NV1 demonstration, Sega agreed to begin working with Nvidia as it planned its next console. On December 11, 1994, Jensen and Curtis Priem flew to Tokyo to suggest a chip-development deal to Sega management.”
Huang explains further in the podcast: “$5 million was a mountain of money to Sega at the time, and so I told him [Sega’s CEO] that if you invested that $5 million in us, it is most likely to be lost, but if you didn’t invest that money, we’d be out of business, and we would have no chance.
“So what he decided was, Jensen was a young man he liked, that’s it.”
The NV1 was Nvidia’s first-ever 3D acceleration processor, and in the Joe Rogan podcast, Huang explains that it was very nearly the only processor that Nvidia would design: “We lucked into the Sega partnership. We started taking off, started building our game console, and about a couple [of] years into it, we discovered our first technology didn’t work.
(Image credit: LGR / YouTube)
“The architecture concepts were sound, but the way we were doing computer graphics was exactly backwards. The inverse texture mapping. We were doing forward texture mapping instead of triangles. We did curve[d] surfaces. So other people did it flat.
“Other technology, the technology that ultimately won, the technology we use today has Z buffers. It automatically sorted. We had an architecture with no Z buffers, the application had to sort it. So we chose a bunch of technology approaches, three major technology choices. All three choices were wrong.”
Sega’s console, the Dreamcast, never used the NV1. Instead, the company went with UK-based VideoLogic (now called Imagination Technologies) and its PowerVR graphics architecture. It looked like Nvidia was done for, especially as Sega told it that it wouldn’t be using the proposed NV2 for its next console.
(Image credit: Future)
Tae Kim explains how Nvidia cleverly got its way out of the situation: “Jensen had deftly worked into the initial contract a clause for a $1 million payment from Sega if Nvidia was able to produce a working prototype of a chip that could be installed onto a self-contained motherboard that was about the same size as the older Sega Genesis/Mega Drive motherboard.
“After about a year spent on the project, [Nvidia] was able to get an NV2 prototype working within Sega’s specifications. The milestone triggered the $1 million payout, money that was a key lifeline during a time of crisis.”
(Image credit: Future)
And key it was, because the payout (and a massive reduction in staff) generated enough funds to research and develop the NV3, also known as the Riva 128. That graphics chip was fast, capable, and although initial drivers were a bit iffy (what drivers weren’t in 1997?), sales of the processor were good enough to secure Nvidia’s future.
“We bet the farm on video games,” said Huang to Rogan. “We narrowly focused our problem statement so I could reject all of the other complexities, and we shrunk it down into this one little focus, and then we supercharged it for gamers.” Given that Nvidia now controls 92% of the discrete GPU market, that’s a bet that paid off by any measure you can think of.
Of course, Nvidia is all about AI these days, which is what the bulk of the JRE podcast is really about. But I like going back to the 1990s in my thoughts, remembering those first graphics cards I bought: Rage Pro, Riva TNT, Voodoo. Ah, good times.
