Altman once remarked onstage that the best book he’d read the previous year in 2018 was The Mind of Napoleon, a more than three-hundred-page compilation of quotes from Napoleon Bonaparte, the French military leader who led a coup to seize control of the French government, installed himself as France’s emperor, and subsequently sought to conquer Europe.

“Obviously deeply flawed human, but man, impressive,” Altman said. “What kinds of insights did he have?” asked Tyler Cowen, an economics professor at George Mason University, who was hosting the event.

“His incredible understanding of human psychology,” replied Altman, who was weeks away from switching to OpenAI full time and still president of YC. “That is something we see among many of our best founders.”

Altman then recounted a specific passage that had struck him most. It was Napoleon’s reflections on the motto of the French revolution (what would become the country’s national motto)—“Liberté, egalité, fraternité”—and how it could be reinterpreted and wielded to consolidate his own power. It was ultimately under that banner that Napoleon did the opposite: He restricted freedom, dismissed fraternity—a philosophy based in unity and solidarity—and granted equality only to French men, not women, while reintroducing colonial slavery in an effort to reconstruct a French empire.

“So he talked about how you build a system…where you can kind of control the people,” Altman reflected. “I was like, ‘Wow. I’m glad he does not run the United States ’cause that is a dude who understands something deep that I did not and clearly was able to use it for power.’ ” Six years after my initial skepticism about OpenAI’s altruism, I’ve come to firmly believe that OpenAI’s mission—to ensure AGI benefits all of humanity— may have begun as a sincere stroke of idealism, but it has since become a uniquely potent formula for consolidating resources and constructing an empire-esque power structure. It is a formula with three ingredients:

First, the mission centralizes talent by rallying them around a grand ambition, exactly in the way John McCarthy did with his coining of the phrase artificial intelligence. “The most successful founders do not set out to create companies,” Altman reflected on his blog in 2013. “They are on a mission to create something closer to a religion, and at some point it turns out that forming a company is the easiest way to do so.” Second, the mission centralizes capital and other resources while eliminating roadblocks, regulation, and dissent. Innovation, modernity, progress—what wouldn’t we pay to achieve them? This is all the more true in the face of the scary, misaligned corporate and state competitors that supposedly exist. “Who will control the future of AI?” wrote Altman in a July 2024 op-ed for The Washington Post amid the aftershocks of the Omnicrisis. “Will it be one in which the United States and allied nations advance a global AI that spreads the technology’s benefits and opens access to it, or an authoritarian one, in which nations or movements that don’t share our values use AI to cement and expand their power?”

Most consequentially, the mission remains so vague that it can be interpreted and reinterpreted—just as Napoleon did to the French Revolution’s motto—to direct the centralization of talent, capital, and resources however the centralizer wants. What is beneficial? What is AGI? “I think it’s a ridiculous and meaningless term,” Altman told The New York Times just two days before the board fired him. “So I apologize that I keep using it.”

In this last ingredient, the creep of OpenAI has been nothing short of remarkable. In 2015, its mission meant being a nonprofit “unconstrained by a need to generate financial return” and open-sourcing research, as OpenAI wrote in its launch announcement. In 2016, it meant “everyone should benefit from the fruits of AI after its [sic] built, but it’s totally OK to not share the science,” as Sutskever wrote to Altman, Brockman, and Musk. In 2018 and 2019, it meant the creation of a capped profit structure “to marshal substantial resources” while avoiding “a competitive race without time for adequate safety precautions,” as OpenAI wrote in its charter. In 2020, it meant walling off the model and building an “API as a strategy for openness and benefit sharing,” as Altman wrote in response to my first profile. In 2022, it meant “iterative deployment” and racing as fast as possible to deploy ChatGPT. And in 2024, Altman wrote on his blog after the GPT-4o release: “A key part of our mission is to put very capable AI tools in the hands of people for free (or at a great price).”

Even during OpenAI’s Omnicrisis, Altman was beginning to rewrite his definitions once more.

During the all-hands on May 15, 2024, after Sutskever’s and Leike’s departures, Altman stressed that OpenAI would soon reach new levels of AI capabilities that would require the company to rethink and reorganize. “We’re now going to assume we’re, like, entering the AGI era,” he said. In the name of its mission, OpenAI would need to close itself off further and to double down on its global lobbying and public messaging. “There’s a lot of stuff we are not currently ready for,” he said. “The standards for security, the policy plans that we have to have, and also the convening of governments that will need to happen to get this ready; a plan, a story, a future that people can see themselves in when it comes to the socioeconomic impact of this.”

At the same time, in the name of its mission, OpenAI would not slow down commercially. “It does not mean we’re not going to ship great products. It does not mean we’re not going to keep doing great research. It does not mean we’re not going to do all sorts of partnerships and other cool things.” At the end of that month, media reports would surface that OpenAI had secured a major deal to bring its models to Apple’s products; both companies would confirm the news two weeks later.

In fact, Altman noted in the all-hands, OpenAI and Microsoft were renegotiating their partnership to ensure that commercialization continued to happen. “I am flying up to Seattle right after this to talk about that. It’s going to have to evolve,” he said. “When we originally set up the Microsoft deal, we came up with this thing called the sufficient AGI clause,” a clause that determined the moment when OpenAI would stop sharing its IP with Microsoft. “We all think differently now,” he added. There would no longer be a clean cutoff point for when OpenAI reached AGI. “We think it’s going to be a continual thing.” The two companies would continue to partner and release ever-advancing technologies—at a great price.

It was a bizarre and incoherent strategy that only made sense under one reading: OpenAI would do whatever it needed, and interpret and reinterpret its mission accordingly, to entrench its dominance.

Behind the scenes, Altman was also laying the groundwork to entrench his own control. The board crisis had made clear that OpenAI’s structure—a nonprofit governing a for-profit—had made his ouster as good as inevitable. The setup had not only enshrined the company’s two countervailing forces—the Boomers and the Doomers—both vying for control of AI development but had also given the board the broad power to fire him based on their own, and not his, interpretation of whether or not he was best serving the mission.

Such a conflict was bound to happen again if the structure stayed in place. On May 28, less than a week after executives sought to bring back Sutskever, an employee posted again in the Slack channel #i-have-a-question. “I don’t know whether/how to ask this,” the question began: Deep in the weeds of OpenAI’s latest shareholder agreement were new details that seemed to allow for the dissolution of the nonprofit. “How solid is the non-profit?” the employee wrote. “Is the plan to remain governed by a non-profit?”

The next day, The Information reported that this did not seem to be the plan. On Altman’s list of top priorities for the year was a restructuring of the organization to look more like a typical company. The following month, the publication confirmed more details. Altman was considering a few different scenarios: one could be transitioning OpenAI to a traditional for-profit; the other would be transitioning it to a for-profit public benefit corporation like Anthropic and xAI. Both scenarios would retain the existence of the nonprofit as a separate entity but dismantle its board’s control over the company’s business. Under this new structure, investors were also pressuring Altman to take equity in the company to align his incentives more directly with their own.

---

Over the next few months, as OpenAI developed plans for the transition, the two forces at play during the Omnicrisis continued. Doomers escalated their public pressure on the company. On June 4, The New York Times profiled Daniel Kokotajlo and a new campaign he launched calling for advanced AI companies to commit to greater transparency and whistleblower protections that preserve the right of employees to warn the public about risks that they saw within the company. In an open letter detailing their demands, Kokotajlo was joined by twelve other signatories, ten of whom were from various eras of OpenAI’s Safety clan. A month later, The Washington Post reported that a group from OpenAI had also filed a complaint with the SEC, alleging that the company had violated federal whistleblower protections with its overly broad exit agreements. Later that month, five US senators would send a letter to Altman with questions demanding greater clarity on the various allegations from Leike, Kokotajlo, and others, as well as Piper’s reporting over OpenAI’s disregard of AI safety and suppression of employee criticism.

Meanwhile, the chaos among leadership and frustrations at Altman continued unabated. Combined with the ongoing Boomer-Doomer tussle, it was leading to repeated changes in the company’s reporting structure. Brockman stopped reporting to Murati and reported instead to Altman; Aleksander Mądry was reassigned shortly after the senators’ letter from heading the Preparedness team to a smaller role in research. OpenAI was also bringing in more seasoned executives, including Sarah Friar, the former CEO of the neighborhood social media platform Nextdoor, to be chief financial officer, and Kevin Weil, a former product leader at Facebook, Instagram, and Twitter, to be chief product officer.

Soon enough, the company would lose a string of its most tenured executives. First to go was John Schulman, who announced his departure on August 5, 2024, noting his desire “to deepen my focus on AI alignment” and his decision to do so at Anthropic. On the same day, Brockman announced that he was taking a sabbatical through the end of the year, framing the leave of absence, in part a culmination of employee grievances with his leadership, as a needed break after nine years of sprinting at the company.

The following month, on September 25, the other executive who had voiced serious concerns about Altman to the board, Mira Murati, abruptly announced she was also leaving. “My six-and-a-half years with the OpenAI team have been an extraordinary privilege,” she wrote, thanking Altman and Brockman and noting how much she cherished and would continue to root for the company. “I’m stepping away because I want to create the time and space to do my own exploration.” Within hours, two more key leaders issued their own departure statements: Chief Research Officer Bob McGrew and VP Barret Zoph, who had co-led the post-training team with Schulman. All three emphasized to colleagues and the public that the timing felt right to leave OpenAI on a high note, after it had reached another major milestone: the shipping of OpenAI’s latest model, Strawberry, in mid-September under the company’s new naming convention, o1, building upon one of Sutskever’s final contributions to the company.

---

In truth, the timing was terrible. After the Omnicrisis, the competition facing OpenAI had only accelerated. Musk was expanding his computing capacity at an alarming pace to build xAI. Anthropic’s latest version of Claude was pulling customers away from ChatGPT. Sutskever had officially formed his new rival company, Safe Superintelligence, and had only just announced a starting $1 billion in funding.

At the same time, after over a year of work, OpenAI was still struggling to attain the desired performance for Orion to justify its release. The company was beginning to stare down the barrel of an uncomfortable prospect: Its tried-and-true formula of scaling no longer seemed to be enough to work; to advance its AI systems further, it likely needed fundamentally new research ideas. This was far easier said than done in general, but even more so after OpenAI had spent years orienting its hiring and team organization around exploiting existing research rather than exploring uncharted science.

Two days before Murati’s announcement, Altman had published his most bombastic blog post yet amid OpenAI’s latest fundraise. The post was titled “The Intelligence Age,” with its breathless promises about the “unimaginable” prosperity to come. “How did we get to the doorstep of the next leap in prosperity?” Altman wrote. “In 15 words: deep learning worked, got predictably better with scale, and we dedicated increasing resources to it.”

During an all-hands, Murati explained to employees the abruptness of her announcement. “I wanted the news of my departure to come to all of you from me first, and not to hear it from your managers, from anyone else, and let alone from the press.” With all of the scrutiny on OpenAI, she had seen no other way to do so without springing it as a surprise on everyone. Upon learning of Murati’s decision, McGrew then decided it was also time to go, he said. “I realized that I actually accomplished most of the key things that I wanted when I came here.”

As with Sutskever’s departure, OpenAI sought to smooth over the trio of exits. “Mira, Bob, and Barret made these decisions independently of each other and amicably,” Altman wrote in an internal note that he then tweeted, “but the timing of Mira’s decision was such that it made sense to now do this all at once, so that we can work together for a smooth handover to the next generation of leadership.”

In lieu of McGrew, Mark Chen, one of the research leads who presented next to Murati and Zoph at the 4o demo event, would step up as the new senior VP of research, to lead the division alongside Pachocki. Another longtime researcher at OpenAI, Joshua Achiam, would step into a new role, head of mission alignment. Liam Fedus, one of the Googlers who had arrived with Zoph, would soon take over his work leading post-training. For the time being, Altman said, OpenAI would not seek another CTO to replace Murati.

With all of the leadership and planned structural changes, the company was revealing its one constant: It was and still would be Sam Altman’s empire of AI.

---

In early October 2024, OpenAI’s newest funding round closed to the tune of $6.6 billion, the largest VC round in history, valuing the company at $157 billion. It included a hitch: Investors could demand their money back if the company did not convert into a for-profit in two years.

Through the rest of 2024, OpenAI’s hemorrhaging of key staff continued: Luke Metz, the third Googler who had joined with Zoph and Fedus; Miles Brundage, head of policy research; Lilian Weng, who had inherited Dave Willner’s trust and safety work and had been newly promoted to a VP of research leading safety; and Alec Radford, the original researcher who set OpenAI down the path of GPT models.

Anthropic began an ad campaign for Claude with a cheeky message on its billboards in San Francisco: “The one without all the drama.” Brockman returned early from his sabbatical amid the talent exodus. Annie sent Sam a legal letter, notifying him of her intent to sue. Musk, allied now with newly reelected president Donald Trump, cranked up his lawsuit, objecting to OpenAI’s anticipated for-profit conversion and releasing more early emails of OpenAI’s founding, including those that recounted Altman speaking badly about Brockman and Sutskever behind their backs (“Admitted that he lost a lot of trust with Greg and Ilya through this process. Felt their messaging was inconsistent and felt childish at times.”) and showing an aversion to transparency (“Felt like it distracted the team.”). In a surprise allegiance, Zuckerberg, who had long feuded with Musk, backed him up; in a letter to the California attorney general, Meta similarly urged a block on OpenAI’s conversion. “OpenAI’s conduct could have seismic implications for Silicon Valley,” Meta wrote. The conversion could set a dangerous precedent for many more startups to designate themselves as nonprofits, granting them and their investors government tax write-offs until they turned profitable.

Late in the year, nestled in the holiday news dump, OpenAI formally announced the plans for its new structure. It would transition into a for-profit public benefit corporation, and the nonprofit would persist as a separate entity with shares in the for-profit. This structure, the announcement argued, was the best way to equip both for-profit and nonprofit with the right resources to carry out their respective objectives while serving the mission. “We once again need to raise more capital than we’d imagined,” it said. “The world is moving to build out a new infrastructure of energy, land use, chips, datacenters, data, AI models, and AI systems for the 21st century economy. We seek to evolve in order to take the next step in our mission, helping to build the AGI economy and ensuring it benefits humanity.”

At the start of the new year, Altman was back to grandstanding. “We are now confident we know how to build AGI as we have traditionally understood it,” he wrote in a new blog post on January 6, 2025. “We are beginning to turn our aim beyond that, to superintelligence in the true sense of the word.”