Seven lessons from Sam Altman for founders

Sam Altman recently called 2026 the best time in history to start a company. His YC AI Startup School talk had seven points worth keeping.

1. Build what others dismiss. OpenAI spent years looking unrealistic from the outside, which gave the team time to work without serious competition.
2. Avoid crowded consensus. When everyone is building the same thing, the opportunity is usually already being priced in. The bigger openings tend to come from capabilities that appeared only in the past year or two.
3. A small group of believers is enough. Altman said that when perhaps 50 people seriously believed AGI was possible, many of them were already at OpenAI. You need a strong co-founder and a few people with conviction, not mass approval.
4. Help people before you need anything. Altman met Greg Brockman years before OpenAI after helping Stripe with an early hire. Important partnerships often begin long before anyone calls them strategic.
5. Faster tools should increase ambition. If agents reduce three months of work to a few days, use that advantage to attempt something larger, not to produce another minor feature faster.
6. Keep the destination clear and the first steps flexible. OpenAI began as a research organisation without a conventional product roadmap. The roadmap emerged through experiments, failures and real-world use.
7. The timing matters. Model capabilities are improving quickly while the cost of building, testing and launching keeps falling. Problems that were too expensive for a small team last year are now within reach.

My addition: cheaper execution does not replace customer understanding. Founders still need to know whose problem they are solving, why it matters and whether anyone will pay.

That is where the real advantage starts.