In a recent interview with Colossus, Scott Wu was asked an interesting question: why are so many of the top figures in AI from the competitive programming world?
The interviewer drew a parallel to the early 20th-century "Budapest Martians" — John von Neumann, Leó Szilárd, Edward Teller, and other Hungarian-Jewish geniuses who were born around the same time, went to similar schools, competed in the same math contests, and eventually met at Los Alamos to build the atomic bomb.
The analogy isn't a stretch for today's AI scene. Among Wu's IOI (International Olympiad in Informatics) teammates: Alexandr Wang founded Scale AI; Johnny Ho co-founded Perplexity; Jesse Zhang founded Decagon; Jeffrey Yan founded Hyperliquid. Look further: OpenAI co-founder Greg Brockman placed in the top 24 in the US Math Olympiad and won a silver in the Chemistry Olympiad. OpenAI's head of research, Mark Chen, coached the US IOI team alongside Wu. OpenAI's chief scientist, Jakub Pachocki, was a frequent rival of Wu's at international competitions. Anthropic founder Dario Amodei made the US Physics Olympiad team.
Wu's explanation is simple: competitions filter for two things — intelligence and a refusal to lose. He used the word "salty," saying he wanted to beat everyone from a young age, even if they were years older. People who are smart enough and salty enough will, after a few years of grinding, eventually develop product sense, people skills, and business judgment.
As for why so many of them started companies around the same time, Wu credits Alexandr Wang. The two became best friends on Google Hangouts in middle school, and even wrote a document of startup ideas together. Wang was the first in the group to actually start a company; seeing him move inspired the rest. "A lot of people don't even know that starting a company is an option. We were lucky — we went through this together, watching each other grow," Wu said.
He drew another analogy: starting a company is becoming like poker. The early poker greats relied on intuition and street smarts; now the tables are full of math nerds. AI startups have reached that point — founders with deep technical chops fare better than storytellers, because the core problems are hard technical ones.