Sam Altman's most consequential decision was not building better AI, but releasing it to the world instead of locking it away. In a two-hour conversation with Stripe CEO Patrick Collison at Stripe Sessions, Altman outlined the contrarian moves that transformed OpenAI from a research lab into the infrastructure backbone of the AI era. While most founders are still debating whether to use AI, OpenAI is signing 20-year contracts that assume AI will reshape every industry.

Why Did OpenAI Release ChatGPT When the Industry Said Not To?

The dominant view inside AI research before ChatGPT launched was clear: keep the technology locked inside the lab. Control who touches it. Frame the restriction as a safety measure. But Altman saw a paradox in this logic. "It is extremely important that we avoid that kind of power concentration and that we build this for the world," Altman explained. Locking technology inside a research cohort, he reasoned, created exactly the power concentration the field claimed to prevent. Iterative public deployment was the contrarian call, and it was a closer decision than most people realize. Every useful AI product built by founders today traces directly back to this single choice.

What Does It Feel Like to Believe in a Technology No One Else Can See?

OpenAI finished training GPT-4 eight months before launch. During those eight months, everyone inside the company used it daily. Zero external signal. Zero market validation. Just conviction and a question with no good answer yet. Altman described the psychological weight of this period with unusual candor: "We'd walk the halls sometimes. Are we engaging in collective psychosis? Have we gotten totally whipped each other into this frenzy? And there was no feedback to keep us in check or sane from the outside world". This is what genuine technological discontinuity feels like from the inside, not triumphant but vertiginous. Founders who have experienced a real threshold moment will recognize this immediately and should prepare for it before they are inside it.

Altman

The craziest OpenAI story Altman chose to tell was not the board crisis or the competition with competitors. It was eight months of living inside a reality the rest of the world had no access to, with no external validation that the bet would pay off.

How to Build an AI Company Without a Technical Co-Founder

The most important founding filter just shifted for the first time in two decades. For years, venture capitalists made fun of the "idea guy" who could not code. Altman now explicitly wants to fund founders who understand their users deeply and cannot write a single line of code. The model is the guitar player; the founder is the songwriter. This shift opens an arbitrage opportunity for a large cohort of operators, clinicians, lawyers, and domain experts who were previously unfundable.

• Deep User Understanding: You must know your customers better than anyone else in the market, including their pain points, workflows, and unmet needs that AI can address.
• Problem Clarity: You have to know exactly what you are solving and why it matters, with specificity that goes beyond vague statements about "efficiency."
• Distribution Instinct: The build is easier than getting people to use it; founders need a clear sense of how their product reaches customers and why they will adopt it.

Most of these previously unfundable founders do not know they just became fundable. That is an arbitrage with a short window. Study what top VCs look for in 2026 before that window closes.

Why CEOs Must Automate Their Own Work First

Tobi Lutke, CEO of Shopify, was the first CEO Altman saw do this correctly. Lutke did not launch incentive programs or gamified pilots. Instead, he built AI automations himself, held the standard personally, and told the team: this is how we operate now. "It was not like a token leaderboard. It was not some other kind of gamified hackable thing. It was just like the CEO of the company said, we are now going to put AI into everything we do," Altman noted. OpenAI is now testing a program where they embed a full-time employee directly with a CEO to automate that CEO's own workflows first. The fractal effect follows naturally; the behavior propagates when the leader goes first and stalls when the leader delegates.

What Science Breakthroughs Could AI Accelerate in the Next Decade?

This is the most consequential claim Altman made in two hours, and it received the least discussion in the room. "If we can start doing a decade of science of what it would have taken us in the old world in a year, the compounding effect there and what we'll be able to do and discover will just be extremely great," Altman stated. Most investor attention is on consumer apps and AI agents. The civilizational leverage is in science, and most portfolios are making only one of the two bets on the table.

• Biology at Foundation Model Level: The analogy to language models is direct; the training data exists, and the architecture already works on protein structure prediction and drug discovery.
• Complex Disease Cures: AI hypothesis generation changes the search space entirely for conditions that have resisted drug development for decades, potentially unlocking treatments for cancer, Alzheimer's, and other intractable diseases.
• Materials Science: Altman calls this massively underrated; new materials unlock physical limits in energy, computation, and manufacturing simultaneously, with implications for everything from batteries to semiconductors.

Why OpenAI Wants to Be Infrastructure, Not a Consumer App

Altman admires the Stripe model: pure infrastructure, revenue aligned with customer success. "I'd be happy for us to be a forever low margin as long as we can be huge and growing fast business, and I would like us to supply kind of an intelligence meter," Altman explained. Stripe gets bigger when its customers get bigger. That is the structure OpenAI wants. Switching costs in AI tools are lower than most people assume; lock-in through capability is the only defensible position long-term. As models improve across the industry, margins compress for everyone. Volume and customer alignment become the durable position. Founders building on OpenAI should take this stated intent seriously. An infrastructure provider that competes with its customers destroys the trust that makes the infrastructure worth using.

What Management Skill Made OpenAI's Breakthroughs Possible?

Altman did not identify his own most important contribution. Someone writing a book on OpenAI did. A biographer observed: "You figured out how to get a lot of people who all thought they were the only capable or most capable person, and everything had to go their way, to work together long enough to figure out the breakthroughs. That was the magic of OpenAI". For founders managing elite technical talent, the forcing function is shared conviction strong enough that the pain of working together is worth it. Lose the belief and the pain ends the company. The structure that made this work included concentrated resources on one direction, not spread thin across competing bets; shared conviction on the mechanism, with everyone believing scaling laws were real before the market confirmed it; and a goal worth the cost, where the mission made the pain of working together functional instead of fatal.

Someone

OpenAI is now signing 20-year infrastructure contracts with major customers, a bet that assumes AI will remain central to computing for decades. This long-term commitment signals confidence that the company's infrastructure will remain essential as the technology evolves, and it locks in revenue streams that justify massive capital investments in compute and research.