Power has a ceiling

Altman sits near the top of the AI industry, but he says an early Silicon Valley lesson showed him how little being in charge actually buys you.

At YC he could only govern

He ran Y Combinator from 2014 to 2019 and could not tell portfolio startups what to do. As an investor, not a boss, he had to persuade each company on its own terms.

“At best you can govern, you don't get to, like, rule.”

Same job, bigger stakes

OpenAI now absorbs pressure from governments, competitors, investors and the wider AI ecosystem, all arguing about how fast the technology should move and whose interests win.

“We have to negotiate tense things with governments and with our ecosystem.”

Be predictable on purpose

Altman's answer is to make OpenAI a reliable partner: act so outsiders can anticipate what the company will do, while weighing interests that pull in opposite directions.

“We won't get every decision right.”

A structure built to say no

He defended OpenAI's complicated corporate setup as the thing that lets it make calls that are not obviously good for the business or its shareholders when the mission demands it.

Not the moment to IPO

Going public now would be ill-advised given AI safety concerns, Altman said. If slowing development became necessary, he would tell investors it cuts against their financial interest.

“For the good of humanity, we're going to do that.”

Rivals in one room

He expects leading AI labs to reach some collective agreement on safety as models get stronger, and floated Trump and Xi sharing a Nobel Peace Prize over joint US-China AI standards.

“I think that will happen.”

What he'd bet on next

Bio and materials science are about to be transformed, Altman said, along with cybersecurity, new energy, and software written on the fly. His YC lesson is about to get a much bigger test.