#16. Sam Altman: Building The Most Valuable Startup in History

26:01The non-profit structure is working and Google is still winning the talent war. Google is still hiring the best people and throwing money at the problem. This is the moment where everything could fall apart. Here's where Sam makes the bet that defines OpenAI. So it's March 2019. OpenAI announces a new structure, OpenAI LP. It's what's called a capped profit entity. So here's what this means. It's a hybrid, still controlled by the nonprofit, but now able to raise capital from investors, able to offer equity-like returns, capped at 100 times the initial investment. So think about what he's doing here.

26:35He's threading the needle, he's keeping the mission, but getting the resources he needs. Now this is quite controversial because some of the early supporters feel betrayed. The whole point was that OpenAI was supposed to be a non-profit, mission-driven, not corrupted by money. But now it looks like they're becoming just another tech company chasing venture capital. But Sam frames it differently. He says, we need to be able to attract capital and talent to accomplish our mission. If we can't compete, we can't build safe AGI. And if we don't build it, someone else will, and they might not care about safety.

27:11It's compelling logic. You can disagree with it, but it's compelling. Four months later in July 2019, OpenAI announces a billion-dollar partnership with Microsoft. And in doing so, Microsoft becomes OpenAI's exclusive cloud provider. In exchange, Microsoft gets to integrate OpenAI's models into its products. This is the turning point. Sam has just transformed OpenAI from a scrappy nonprofit into a company that can raise real money, offer real equity, and compete with Google. The capped profit structure is his way of keeping the mission while getting the resources he needs. If OpenAI works, it's not just big.

27:49It's not just Airbnb for X. It's civilization scale. It's the most important thing happening in technology. And if it doesn't work, he's still young enough to do something else. You can always go back. That's an incredible amount of confidence to have and enormous pressure for one person to deal with. So over the next year, something happens. In May 2020, OpenAI releases GPT-3. So to give some context here, GPT-3 has 175 billion parameters, orders of magnitude larger than anything before it. It can generate text that's almost indistinguishable from human writing. It can translate languages, write code, summarize documents, explain concepts.

28:29It's shockingly good. But Sam later says that when they released GPT-3, most of the world didn't care. Most people didn't understand what had happened. They had no idea. But Silicon Valley, Silicon Valley got it. Startup founders were like, oh, this is AGI. In reality, it's not AGI, but it's pointing at something real. So fast forward to June of that year, OpenAI releases GPT-3 via an API. at first, the only companies that build real businesses with it are doing AI power copywriting. It's cool, but it's not transformative yet. But here's what Sam notices, and it's crucial. People love playing with GPT-3 in the OpenAI playground.

29:07They love talking to it, even though it's not designed for conversation. They're spending hours just typing things and seeing what it says back. That single observation that people want to talk to AI, they want a conversation, that observation becomes the seed for chat GPT. But we're going to get into that in the next inflection point. For now, here's what he's done. He's turned OpenAI from a non-profit research lab into a real company with a billion-dollar partnership with Microsoft, with a structure that can raise capital, with a product, GPT-3, that's captured the imagination of researchers and developers.

29:40He's positioned OpenAI to compete with Google and Facebook, not as an academic lab, but as a serious AI company. But most importantly, Sam has survived the transition. The non-profit mission hasn't been totally abandoned. It's just been redefined. The original believers might be skeptical, but the new structure allows OpenAI to actually build something.

30:08Okay, time for the iconic moment when Sam made the world stand still. It's late 2022. OpenAI has released GPT-3, then GPT-3.5. They've released DAL-E, an image generation model. They're making progress. They're publishing papers. They're getting respect in the research community. But here's the thing. They're still mostly known just inside Silicon Valley cycles, not to the general public, not to regular people. To most people in the world, OpenAI is basically invisible. Inside OpenAI, though, there's a debate happening. They've built these incredible, powerful language models. Models that can do things that seemed impossible just a few years earlier.

30:46But most people can't access them. The API is for developers. The playground is for tinkerers. It's for people who know how to code. So the question remains, how do you make this accessible for everyone? Some engineers are skeptical. Building a simple chat interface feels too basic, too obvious, not worth the time. It's just a wrapper around GPT-3. There's no innovation here. There's no breakthrough. proof. But Sam has this intuition and intuitions are interesting because they often come from observation. He saw the playground data. He saw how people were using it. People weren't using it the way it was designed.

31:20They're having conversations with it, typing prompts, reading responses, spending hours just talking to the AI. So Sam thinks about this. He acknowledges that people don't necessarily want the API. They want to be able to talk to AI. They don't want to write code. they want a conversation that's what they actually want so sam makes a call we're building a chat bot it's going to be free it's going to be dead simple no bells and whistles just the text box where you type and the ai responds so it's crazy to think about this but the team builds chat gpt in a matter of weeks it's basically gpt 3.5 with a chat interface and some fine tuning to make it more helpful and less prone to generating harmful content that's it that's the entire innovation so now here's the big moment november 30th 2022 chat gpt goes live and here's what's interesting there's no huge marketing campaign no glitzy keynote just a simple blog post and a tweet we've launched chat gpt try it here that's it so getting into sam's head for a moment what's happening here is that sam is looking to change the entire conversation about ai he knows that most people don't understand what these models can do they basically think that ai is siri which is obviously bad, or self-driving cars, which are far away, or something that might happen someday, something abstract or theoretical.

32:40The reality is that most people don't know that AI can write essays, debug code, explain complex topics, and be creative. They don't know any of that is here, that this is right now. But there's something deeper happening in Sam's mind, and it's more strategic than this. He realizes that open AI needs to own the consumer relationship. If they build a product that people use directly for an abundance of use cases, then they're a platform. They're not just enabling others to build companies on it. They are the lifeblood of each and every organization. So Sam is betting that a free, simple chatbot will be the thing that finally makes AI click for regular people.

33:17Not developers, not researchers, regular people. Everyday people who have no idea what machine learning is. That's a big bet, and it's a bold one. But here's the problem. A lot of Sam's own team don't believe in this. Greg Brockman, OpenAI's president and Sam's closest ally, later admits something. He says, most engineers, including me, oppose building a simple chatbot. We thought it was worth the time. So keep in mind, most of the engineers building this, including his own president, they don't believe in this. They're thinking, why on earth would we spend engineering resources on this when we could be working on GPT-4 or DALI-2 or actual innovations?

33:55Plus there's risk. Real risk. If chat GPT launches and no one cares, it's embarrassing. OpenAI has built all this hype around GPT-3, supposedly the most powerful model ever. And now they're releasing what's essentially a simplified version in a chat interface. If people don't get it, the narrative becomes OpenAI is out of ideas. And there's another risk, a different kind of risk. If you make a powerful language model accessible to everyone with no barriers, people are going to use it for bad stuff. generating spam, writing phishing emails, creating disinformation and spreading lies. And this makes the safety team nervous.

34:31They're worried about what happens when this gets into the wild. But Sam overrules them. And he says that we're launching it. So back to that famous date, November 30th, 2022, and ChatGPT goes live. Within five days, it hits a million users. One million users in five days. To put that into perspective, Instagram took two and a half years to hit a million users. TikTok took nine months. Facebook took 10 months. And ChatGPT does it in five days. By January 2023, just two months after launch, ChatGPT has 100 million monthly active users. It's the fastest growing consumer application in history. Again, that's worth repeating.

35:18The fastest growing consumer application in history. The internet explodes. People are sharing screenshots of ChatGPT writing poetry, solving math problems, explaining quantum physics like you're five years old. Students are using it to write essays. Programmers are using it to debug code. And therapists are worried it's going to replace them. This is not some slow, gradual adoption. It's viral. It's one of those paradigm shifts where the internet discovers something and collectively loses its mind. Now, here's where Sam does something that no one else is willing to do. He makes it free and frictionless.

35:53No signup required, at least initially. No API key, no credit card. Just go to the website and start typing. That's it. That's all you have to do. This is the opposite of what most tech companies do. Most companies would have immediately launched this as a premium product. Gated access, invite only, make people beg to try it. But Sam says no. We're going to fully open this up. We want everyone to try it. We want it to spread. and the virality is insane. Every person who uses ChatGPT tells their friends. Those friends tell more friends. It's pure word of mouth. OpenAI spends basically nothing on marketing because the product markets itself.

36:32But here's the other thing Sam does that's crucial and this is genius level thinking. He frames ChatGPT not as a finished product but as a research preview. The landing page literally says, this is a research preview. It will make mistakes, help us improve it. Now that is genius framing because it lowers expectations. People aren't upset when it gets things wrong because it's a research preview. It's allowed to be imperfect, but it also creates engagement. People want to help improve it. They're reporting bugs, suggesting features, sharing feedback. OpenAI is getting millions of data points on how people actually use AI.

37:07It's crowdsourced product development and it's brilliant. So fast forward a few months, it's now February 2023 and ChatGPT has over 100 million users. By March 2023, OpenAI launches GPT-4, an even more powerful model, and in parallel announces ChatGPT+. OpenAI priced at$20 a month, faster response times, and early access to new features. And this subscription is an instant success. Within a year, ChatGPT has over 15 million paying subscribers. That's over$300 million in ARR just from subscriptions. But the bigger result is this. ChatGPT changes the narrative around AI. Before ChatGPT, AI was this abstract thing that might happen someday.

37:50But after ChatGPT, AI is a thing that everyone's using right now. Not future tense anymore. It's happening right now. And all of a sudden, Sam becomes the face of the AI revolution. He testifies before Congress. He meets with world leaders. He's on the cover of Time magazine. He's one of the most influential people in technology. This launch was Sam's bet that experience beats performance. GPT 3.5 wasn't even the best model at the time. It was just good enough, wrapped in an interface that anyone could use. And it worked. Fast forward a little bit more, by mid-2024, ChatGPT has 400 million weekly active users.

38:32By mid-2025, it's 800 million. It's one of the 10 most visited websites in the world. And here's the crazy part. Sam's engineers were right. It was just GPT 3.5 with a chat interface. but what the engineers did get wrong was that this matters sometimes the interface is everything and that's a lesson that goes way beyond ai that's a lesson about how the world actually works but this journey wasn't a smooth one for sam far from it so let's dig into why that is

39:11all right final inflection point it's november 16 2023 open ai is at its absolute peak chad gpt is the fastest growing app ever the company's valued at 80 billion dollars sam has become one of the most recognizable faces in technology everything looks perfect everything looks like is working, everything looks like success. But behind the scenes, there's tension. The board of OpenAI is worried, and I mean genuinely worried. They think Sam is moving too fast. They think he's prioritizing growth over safety. There are concerns that he's not being fully transparent about what's happening inside the company.

39:48He's making decisions without them that he's become too powerful. Ilya Sutskova, OpenAI's co-founder and chief scientist, has been documenting Sam's behavior for over a year. He's written a 52-page memo accusing Sam of a consistent pattern of lying, undermining his execs, and pitting his execs against one another. He says that Sam told him and Mira Morati, the CTO, conflicting things about how the company would be run. Different stories, different promises. Suskova talks to the other board members, Helen Tona, Tasha McCauley, Adam D 'Angelo. They're all nervous. They think Sam is too powerful, too focused on commercialization, and not focused enough on safety.

40:29On Thursday, November 16th, Sam gets a text from Ilya asking him to join on a Google Meet the next day. And at that time, Sam has no other reason but to think that this is a routine meeting. So it's now Friday, November 17th, midday. Sam joins the Google Meet and the board tells him, you're fired. Effective immediately. You're off the board and Mira Morati is the new interim CEO. So Sam naturally asks why. and they say you are not consistently candid in your communications with the board. That's it. No specifics, no details, just you're out. Greg Brotman, OpenAI's president and Sam's closest ally gets a separate message.

41:07You're off the board too, but you can stay as president reporting to Mira. So naturally Greg says, screw that, I'm resigning. Within an hour, OpenAI posts a blog announcing Sam's departure and the tech world explodes. Twitter is on fire. No one understands what's just happened. So here is where the board makes a massive miscalculation. They thought they could fire Sam cleanly, install Mirror, move on, get back to business. But they didn't anticipate three things. Three things that would cascade and destroy their plan. First, Microsoft is furious. They've just invested$13 billion into OpenAI. $13 billion.

41:46They bet their entire AI strategy on this partnership. And they get the news at the same time as everyone else. Through Twitter. Satya Nadella, Microsoft CEO, finds out that his company's$13 billion investment just had its leader removed, and he wasn't even consulted. So by Sunday, something happens. Over 500 of OpenAI's 770 employees sign a letter, threatening to quit and join Microsoft unless Sam is reinstated. 500 out of 770. That's nearly 65 % of the company. Most people who get fired go home. They sulk, they call a lawyer, They disappear from the public eye. But Sam doesn't do that. Within hours of being fired, he starts negotiating his return.

42:29And again, here's where Sam's incredible resourcefulness kicks in at a level most people have never seen. So it's Friday night. Sam's in shock, but he's already talking to people about what's next. Investors are calling him. Founders are calling him. The ecosystem is gathering around him. So now it's Saturday. Sam and the board start negotiating his return. But the board is divided. Some people want him back. Some don't. The discussions go late into the night. Nobody is sleeping. So now it's Sunday. The negotiations break down. The board appoints someone called Emmett Shear, the former CEO of Twitch, as the new interim CEO, replacing Mira.

43:05This feels like a betrayal because Mira was supposed to be the temporary. Now she's out too. Nobody trusts the process. It's now Sunday night, late Sunday night. And Satya Nadella announces that Microsoft is hiring Sam and Greg to lead a new advanced AI research team at Microsoft. This is Nadella's power play. If OpenAI won't give Sam back, Microsoft will take him and Greg too and the whole thing fractures. Now we're at Monday morning and the employee revolt goes nuclear. Over 500 employees sign the letter and that letter is brutal. It says, your actions have made obvious that you are incapable of overseeing OpenAI.

43:43We, the undersigned, may choose to resign from OpenAI and join the newly announced Microsoft subsidiary. And then something extraordinary happens. Ilya Sutskov, Ilya, who'd led the coup to fire Sam, who wrote the 52-page memo accusing him of lying, Ilya flips. He tweets, I deeply regret my participation in the board's actions. I never intended to harm OpenAI. I'll do everything I can to reunite the company. So here's what's important to understand. Sam doesn't win this through force. He doesn't have more power than the board. He doesn't outmaneuver them strategically in that moment. He won because he'd spent years, 15 years, building relationships.

44:22Every employee at OpenAI who signed that letter owed Sam something. He'd hired them, mentored them, protected them. The network he built at Y Combinator, the reputation he built at Loop, the goodwill he built over 15 years, it all came due in one weekend. And on Tuesday, the board caves. Sam is reinstated as CEO. The board is reshuffled. Brett Taylor, former co-CEO of Salesforce, becomes chairman. Larry Summers joins. And there is only one original member who stays. Sam is back after five days. But he's not just back. He's stronger. The old board is gone. The employees are loyal. They've just proved it.

45:06Microsoft is even more invested. And Sam has just survived something that would have destroyed most people. Here's what this moment revealed. And here's the insight. Sam's superpower, it isn't technical skill or vision. It's not that he's the best coder or the smartest person in the room. His superpower is that he makes himself indispensable. When they tried to remove him, the entire company nearly collapsed. That's not luck. That's 15 years of relationship building paying off. That's understanding that power comes from people, not titles. That's making sure that everyone around you knows what you've done for them.

45:45And here's the other thing. Sam learned something from this. In a direct quote from an interview with Lex Friedman, he says, it definitely changed how I think about default trust. I was an extremely trusting person. This was so shocking that it's changed me. He admits he's more cautious now, more paranoid. He trusted the board and the board tried to remove him without warning. But he also says something else, something that surprised people. He says a direct quote, The dominant emotion of that weekend was love, not hate. Despite everything, despite the betrayal, despite the coup attempt, the dominant emotion was love.

46:23He felt supported by the people who mattered, the employees, the investors, the founders he'd helped over the years. The network came through. By early 2024, OpenAI is back to business. ChatGPT continues to grow. GPT-4 gets even better. And Sam, he's untouchable now. He survived a coup, got rehired by popular demand, came out more powerful than before. And here's what I want you to understand about all this. When Paul Graham said that Sam has such force of will that he's going to get what he wants, this is what he means. This moment, this weekend. It's not that Sam is aggressive or manipulative.

47:01he understands how systems work he understands that organizations are made of people and he spent 15 years building deep relationships with people when the moment of truth came when someone tried to take him down all those relationships activated at once that is power that's real power and that's a lesson that goes way beyond open ai

47:31Okay, so we've walked through five massive turning points in Sam Altman's life. Looped, Y Combinator, OpenAI, ChatGPT, the firing and reinstatement. So now let's step back and talk about what's really going on here. What are the patterns, what makes his decision making and strategic thinking so different from everyone else? I see three big things. Sam is relentlessly resourceful to the point where it becomes his unfair advantage. Think about the loop story. Trying 30 different paths into a mobile carrier until one finally works. An exec literally tells him, I'm meeting with you because I want you to stop bothering us.

48:09And Sam takes that as a win because he understands something most people don't. Persistence is an underrated superpower. Or look at the OpenAI firing. Within 48 hours, he has Microsoft offering him a job. 500 employees threatening to quit. The board negotiating his return. Most people try once, maybe twice. They get rejected or they get tired. But Sam keeps trying until the problem bends to his will. Most people stop after the first ignored email. But Sam doesn't. Most people stop after the first no. But Sam sees that as a data point, not a death sentence. He tries 30 different parts, 50 if he has to.

48:47So the takeaway here for founders listening is that it's not about being the smartest person in the room or the most talented. It's about being the most persistent in finding a way through. Second, Sam plays a different game than everyone else. He's optimizing for leverage, not immediate outcomes. Look at his decision to become president of Y Combinator. Most people in his position would have thought, I should just go start another company and go for a massive exit. But Sam saw something different. He saw that being president of YC gave him access to hundreds of founders every single year. It gave him the ability to shape the next generation of startups.

49:23It gave him a platform to build relationships at scale. that's leverage not immediate money not immediate prestige but leverage the ability to make things happen the ability to influence hundreds of companies and the ability to compound relationships over time same calculation with open ai most people would have stayed at y combinator it was the most powerful position in startup land guaranteed influence guaranteed access guaranteed deal flow but sam saw the open ai if it worked would be civilization scale not billion dollar startup scale, civilization scale. So he left guaranteed influence for uncertain, but potentially infinite leverage.

50:02Here's the question Sam seems to be asking himself at every inflection point. What is the highest leverage move I can make? Not what's the safest move, not what's going to make me look the best right now, but what position do I want to be in five years from now? And what move today gets me closest to that? That's a very different optimization function than what most people use. Third, and this might be the most important one, Sam understands that relationships are the ultimate compounding asset. Every single inflection point in his story comes down to relationships. He gets into Y Combinator because Paul Graham likes him.

50:36He becomes president of YC because Graham trusts him. He survives the open AI firing because 500 employees chose him over the board. But here's what's subtle and it's what most people miss. Sam didn't build these relationships by being transactional, he built them by being genuinely helpful, by taking care of people, by being generous with advice and introductions. He understands at a deep level that the relationships you invest in are the dividends you get paid later. As he writes in one of his essays, one of the best ways to build a network is to develop a reputation for really taking care of the people who work with you.

51:12That's exactly what he did. And when it mattered, when the board tried to fire him, those relationships were the thing that saved him. Not his technical skills, not his intelligence, not his vision for AI, but the fact that people wanted him to win because he'd invested in them. He'd helped them. He'd build goodwill that compounded over 15 years.

51:40So here's what keeps coming back to me about Sam Wattman's story. It's not about being the best. It's not about being the smartest or the most technically skilled. It's about being the person who stays in the game long enough for the game to change in your favor. Sam dropped out of Stanford to build a company that sold for$43 million. It's good, but not a great outcome when you're talking about the standards of elite entrepreneurs. Most people would have looked at that and said, that's a pretty decent outcome. But Sam looked at it differently. He said, I've now built relationships with the best people in Silicon Valley.

52:12I've learned how to operate at scale. well, I've earned credibility. Now I'm in the game. He became president of Y Combinator, not because he'd run the most successful startup, but because Paul Graham saw something in him, a relentless resourcefulness, a willingness to do whatever it took to make things work. He turned open AI from a nonprofit into a real company, not because he had the best technical vision, but because he understood leverage better than anyone else. He knew the nonprofit structure wouldn't allow them to compete. So he changed the structure. He launched ITT not because his engineers thought it was a good idea.

52:47They didn't. He launched it because he had this intuition. People want to talk to AI and he was right. And he survived getting fired because he'd spent 15 years building relationships, creating goodwill and making himself indispensable. That's the pattern. Sam doesn't win by being the best. He wins by being relentlessly resourceful, by optimizing for leverage instead of immediate outcomes and by building relationships that compound over time to build the most valuable startup in history. Thank you for joining us on Inflection Moments. If today's story sparked a new perspective or challenged your thinking, be sure to share it with someone you know loves this stuff as much as you and I do.

53:27Maybe it's a college buddy, your water cooler buddy, or maybe even someone in the family group chat. If you enjoyed this deep dive, make sure to leave a five-star review and subscribe to our channels so you can be the first one to hear what we've got coming next. And if you're interested in insights, ideas, and lessons from some of the world's greatest entrepreneurs, sign up for our newsletter. The link is in the show notes. Until next time, keep building and talk soon.