At 19, Scott Phoenix decided every other problem was a waste of time. If you could build human-level AI, everything else would follow. Most teenagers with that realization buy books. Phoenix started building.
He's 43 now. The first startup - a gaming news website - launched when he was 16. The second was peer-to-peer file sharing. By the time he hit Y Combinator in 2008, he'd already figured out that incrementalism was for people who didn't understand exponentials.
Vicarious, the company he co-founded with neuroscientist Dileep George in 2010, wasn't trying to make better chatbots or smarter recommendations. They were reverse-engineering human cognition. The pitch was audacious enough to get Elon Musk's attention. And Mark Zuckerberg's. And Jeff Bezos's. $250 million from people who don't write checks for maybes.
The Neuroscience Bet
Most AI companies in 2010 were chasing narrow applications. Phoenix and George were reading neuroscience papers and building computational models of the visual cortex. While competitors optimized click-through rates, Vicarious published in Science.
The company spent 12 years in build mode. No pivot to enterprise SaaS. No pivot to consumer apps. Just the expensive, slow work of trying to crack general intelligence. Published research in NeurIPS and ICML. Filed patents. Stayed in the lab.
If you actually figure out how to build the first human-level AI, then you've solved all the other problems.
That's not founder-speak. That's the operating thesis. Phoenix wasn't building a company. He was building the last company anyone would need to build.
The Marathon Rubicon Detour
Before the AI moonshot, there was award-winning digital art. And open-source game development - Marathon Rubicon, a total conversion mod that's still played today. Phoenix has always been the kind of technical founder who can't just code. He has to make it beautiful.
The Founders Fund stint as Entrepreneur in Residence gave him the network. Y Combinator gave him the credibility. But the gaming background gave him something more useful: an intuition for systems that feel intelligent without being explicitly programmed. NPCs that seem to think. Enemies that adapt. He'd been thinking about artificial minds since before it was lucrative.
The Alphabet Exit
When Intrinsic acquired Vicarious in 2022, the terms stayed private. Phoenix spent a year as Chief Product and Revenue Officer at the Alphabet robotics division, then left. Not burned out. Not cashed out and retired. He joined Fifty Years as a partner.
Fifty Years invests in the kind of companies Phoenix understands - scientists building hard things that might not work. PsiQuantum (quantum computing). Sila Nano (battery materials). Tenstorrent (AI chips). Zipline (delivery drones). Science Corporation (brain-computer interfaces). These aren't SaaS companies with 40% margins. They're bets on atoms, not bits.
Phoenix's personal investments read like a syllabus for post-AGI infrastructure. If artificial general intelligence changes everything, you need new materials, new compute architectures, new ways to interface with machines. He's not waiting to see what happens. He's funding the transition.