Elad Gil says Anthropic, OpenAI and SpaceX each went from near zero to a trillion-dollar market cap in about five years, a pace he calls unprecedented.
Elad Gil says most companies should consider a sale in a twelve-to-eighteen-month window and revisit the question every six months, since AI moves roughly three to four years per calendar year.
Elad Gil argues that tech thrived under light regulation and warns that AI safety burdens could slow progress the way FDA risk-only rules slowed drug development.
Few More Trillion-Dollar AI Companies Are Coming Soon
AI is creating giant companies faster than any prior wave, yet the hosts think the market is still misjudged on size, speed and founder ambition. Reaching a trillion dollars takes $50 to $100bn of revenue, so few more will arrive soon, while many $100bn companies remain to be built. Compute, timing of exits and light regulation decide who captures the value.
Takeaways
Trillion-dollar speed is unprecedented, but few more are near. Three companies, SpaceX among them, went from near zero to a trillion dollars in about five years. Elad Gil expects few more in three to five years, since a trillion needs $50 to $100bn of revenue.
Investors misjudge both market size and speed. Sarah Guo says AI services are still priced per seat instead of by outcomes, and coding consumption could grow 100x. Elad agrees coding is bigger than assumed but says investors conflate market size with how fast companies reach it.
Top founders may be aiming too small. Elad says some strong founders chase niches or move to hardware out of fear of the labs, while Harvey, OpenEvidence, Decagon, Sierra and Cognition entered large markets. Sarah is more often disappointed that founders are less ambitious than they could be.
Most companies should consider selling within twelve to eighteen months. Elad says Anthropic and OpenAI should not sell near term, but most others should weigh a sale when they are worth the most. Revisit it every six months, since AI years run three to four times faster. Sarah warns private markets can stay irrational.
Compute, not algorithms, limits progress and favors big labs. Elad says code may be solved within about six months and light recursive self improvement could arrive by the end of next year. Compute reinforces an oligopoly, new architectures get copied fast, and tokens should go by return on invested tokens.
California's billionaire tax is pushing founders out. Sarah says regulators may not have thought through execution and many founders are already leaving. Elad says the bill's authors want people gone. A Texas hardware corridor, from SpaceX and Anduril, grew largely because of regulation.
AI should stay lightly regulated, as tech did. Elad cites Janssen on regulatory capture in pharma and warns AI safety burdens could slow progress the same way. He notes nuclear supplies about 70% of France's power but 18% in the US, and says safety rules hurt energy.
Predictions from the hosts
Valuations keep rising over the next year or two, and Elad expects the same founders to face the sell-or-hold question again.
Talk of a California exit tax grows around 2028, in Elad's view.
Tap a name to go there · tap a line to see why they are linked
On this episode 2
Show 1
Sarah predicts a slow migration toward Miami, and Elad says it will take time.
Former big tech engineers become sought after by firms such as GE and PG&E, since small teams like Minecraft's five to ten people can build very valuable products.