Sam Altman spoke at Y Combinator's Startup School 2026 about why right now is the best point in time to build something new.
What I liked is that the talk is genuinely motivating overall. None of that "AI is going to take all our jobs" stuff. Instead, he's saying the ceiling on what one person or a small team can pull off with the current AI stack has gone to the moon.
I pulled out the most interesting ideas from the video:
- **⚡️ Agents are rewriting the rules.** What used to take 3 months of nonstop coding 20 years ago now gets built by AI agents in 7 minutes. Taste and fluency with the tools now beat years of traditional industry experience.
- **🚀 The alpha is in the next six months.** Altman's prediction: model progress over the next 6 months will feel like all the progress of the last 2 years combined.
- **Tokenomics and the demand for intelligence.** 6.5 years ago, the world record holder was burning 100,000 tokens a month. Now 100k is just the global average, and the leaders are torching hundreds of billions. Altman predicts that in another 6.5 years, the average user will be spending 500 billion tokens a month.
- **🛡 Startups as a defense against dystopia.** Altman sees new products as a defense against the concentration of power. Otherwise there's a risk of ending up in a scenario where AI hands everyone material abundance but builds the perfect surveillance state — permanently stripping people of freedom and agency.
- **💡 Advice for founders.** Look for ideas that break consensus. When OpenAI launched, most experts called them idiots and accused them of setting up for an "AI winter." Ignore the armchair critics, build your own crew, and go do the thing that looks impossible.