Sam Altman credits two of Silicon Valley’s most influential figures with teaching him one of the most important lessons of his career: the best investment opportunities are almost never the ones everyone else is chasing. Speaking on the Invest Like the Best podcast, the OpenAI CEO said both Paul Graham and Peter Thiel drove home the idea in different ways that truly spectacular returns come from betting on things that look unpopular or even outlandish in the moment. You can do fine following trends and catching waves early, Altman explained, but to do spectacularly well you almost always have to do what nobody else is doing.
Altman’s relationship with Graham runs especially deep. Graham cofounded Y Combinator in 2005, and Altman later took the helm of the legendary startup incubator. He said Graham taught him that investing in founders is like being a flight instructor. Founders don’t need someone lecturing them from a whiteboard. They need someone sitting beside them pointing out what worked, what was missed, and where things are going wrong. That hands-on guidance, channeling what Altman called the Brownian motion energy of an early-stage company, is what actually helps.
Those lessons came rushing back recently as OpenAI raced to secure computing power after realizing GPT-4 had put them on an exponential curve of model improvement. Altman said they started calling cloud providers, chip fabricators, and energy companies, only to be told repeatedly that they were crazy. Nobody had ever seen an industry try to scale that fast, and potential partners warned it was reckless and bound to bust. It felt exactly like fundraising for an early-stage startup, he recalled. Most people tell you no, but all you need is one or two yeses. Eventually those yeses came, and OpenAI was able to keep building.
