From King to Exile to King Again: The Inside Story of Sam Altman’s Whiplash Week

Despite Altman’s apparent sense of ease, tensions had been simmering at his company over the enigmatic CEO’s laissez-faire approach to safety, interest in commercializing OpenAI’s software, and tangled web of personal investments in other AI companies. Even as they built one of the world’s most valuable startups, OpenAI remained beholden to a nonprofit board of directors that had the power to pump the brakes if it felt the organization was straying from its mission to “ensure that artificial general intelligence benefits all of humanity.”

Unbeknownst to anyone outside OpenAI’s top ranks, members of that six-person board were growing worried that the CEO was clipping the safety guardrails as he barreled toward global dominance in AI. As far as some on the board were concerned, Altman was himself teetering on the edge of an abyss.

Just a few hours after the APEC panel, Altman received a text from Ilya Sutskever, a chief researcher, co-founder and board member at OpenAI who was responsible for limiting societal harms from its AI. Sutskever invited him to a video meeting with the board the following afternoon. The coup had begun.

What followed was a dizzying spiral of power plays and shifting alliances that seemed more appropriate for the set of “Game of Thrones'' than for the sleek, plant-filled offices of an AI startup. Altman’s surprise firing set off a chain of events—including the shock resignation of OpenAI’s president, the installation of two interim CEOs, a staffwide mutiny, backroom deals, threats of litigation, counter-offensives and counter-counter-offensives—that culminated with a late Tuesday night deal to reinstate Altman as CEO and revamp the company’s board of directors.

This is the inside story of how it all went down, according to current and former OpenAI employees, investors and internal company documents viewed by The Information.

---

Friday Afternoon Massacre: Friday, Nov. 17

Altman spent Friday morning as he often did, working the phones. In the weeks prior, the peripatetic founder had traveled to the Middle East to raise money for a new venture. That morning he held discussions with semiconductor executives about these early efforts to design new chips that would lower costs for large language model companies like OpenAI, as The Information later reported.

One of those calls was with Rene Haas, CEO of SoftBank-backed Arm, a crucial semiconductor firm that designs a large portion of the world’s chips. Right now, chipmaker Nvidia controls most of the market for processors that power data centers to run AI training software. Building a viable alternative would likely take years. It couldn’t be learned whether Altman was representing OpenAI or a separate venture in the discussions, but the implication was clear: Altman wanted to wrest more power from a leading AI chipmaker.

At noon, Altman joined a Google Meet videoconference call at the prompting of Sutskever. The whole board—minus President Greg Brockman—was on the call. There, to Altman’s apparent shock, Sutskever fired the CEO and told him the board would be releasing a public statement about his departure shortly.

At 12:19 p.m., Sutskever texted Brockman asking to chat, according to a timeline Brockman shared later that day. Four minutes later, he joined a video call with the remaining members of the board—Sutskever, Quora founder Adam D’Angelo, tech entrepreneur Tasha McCauley and Helen Toner, the AI researcher who had recently incurred Altman’s wrath after publishing an academic article seemingly questioning OpenAI’s approach to safety. Sutskever told Brockman he had been removed from his position as chair of the board but would remain as president of the company, Brockman said.

Around this time, representatives from OpenAI reached out to Microsoft, which had invested more than $13 billion in the startup, about the transition. The news blindsided senior leaders at the company, including Microsoft CEO Satya Nadella. They ultimately learned about Altman’s ouster only five to 10 minutes before it was made public.

Around 12:30 p.m., OpenAI published a curiously worded statement announcing Altman’s removal for not being “consistently candid in his communications with the board.” The board also announced Brockman’s demotion and Chief Technology Officer Mira Murati’s transition to interim CEO.

The news rocked the entire senior leadership of OpenAI, who had not been briefed on Altman’s departure in advance. Sutskever announced the shake-up in an email to the company around the same time as the blog post. “Change can be scary,” he wrote, adding that there would be an all-company meeting later that afternoon to provide more context on the board’s decision.

At 2 p.m., the remaining OpenAI leaders attempted to calm shell-shocked staffers during a tense virtual all-hands meeting. Interim CEO Murati, a longtime product leader at the company, reassured employees that its relationship with Microsoft—its biggest outside source of capital and computing power—was stable and that the company had expressed “utmost confidence” in OpenAI following the shake-up.

During the meeting, multiple employees pressed Sutskever on the impetus for Altman’s firing, and asked whether it amounted to a “coup” or a “hostile takeover,” according to a transcript of the meeting viewed by The Information. “You can call it this way,” said Sutskever. “And I can understand why you chose this word, but I disagree with this.” He said that the ouster was necessary to  preserve OpenAI’s mission, but he agreed that carrying out “backroom removals” of high-profile executives wasn’t a great look.

Shortly after the companywide all-hands, Sutskever convened a meeting with OpenAI’s research team. He expressed surprise that the company’s researchers would question the board’s decision, according to a person familiar with his comments.

In an attempt to reassure staffers, Sutskever specifically touted Brockman and research director Jakub Pachocki as key assets OpenAI still had on hand, said the person familiar with the discussion. But within a few hours, both men had resigned from the company, joined by two other senior researchers, Aleksander Mądry and Szymon Sidor, The Information reported.

By this point, speculation about Altman’s departure had reached a fever pitch online, and prominent entrepreneurs, venture capitalists and OpenAI investors—including former Google CEO Eric Schmidt, Airbnb CEO Brian Chesky, Sequoia Capital general partner Alfred Lin and SV Angel founder Ron Conway—issued public statements in support of Altman and Brockman.

As rumors began to swirl, some financial firms who were OpenAI customers grew concerned that if Altman had been ousted over data privacy problems, that could harm their business, The Information reported.

The company’s competitors could smell the blood in the water. Employees at Amazon Web Services and Anthropic—an OpenAI competitor that has received billions of dollars of investments from AWS—began strategizing Friday afternoon about how best to poach OpenAI’s customers. The cloud computing giant quickly assembled a team to respond to panicked OpenAI customers, coordinating its efforts with Anthropic.

Ultimately, over 100 OpenAI customers contacted Anthropic over the weekend as concerns grew over the state of the company sans Altman, The Information reported. Others considered defecting to Cohere, which has partnered with Google Cloud, or to Microsoft’s Azure service.

---

Into the Fog: Saturday, Nov. 18

By the early hours of Saturday morning, it had become clear to many within the company that things were getting out of hand. OpenAI Chief Operating Officer Brad Lightcap attempted to douse the fires of speculation in a companywide memo.

He wrote that the decision was based on a “breakdown in communication between Sam and the board” rather than “malfeasance or anything related to [OpenAI’s] financial, business, safety, or security/privacy practices,” according to a copy of the memo seen by The Information. Meanwhile, the board of directors had gone dark.

Quietly, they began looking for a replacement. On Saturday, members of the board approached Anthropic about a potential merger of the two companies following the removal of Altman, as The Information would later report. According to a source familiar with the discussions, the board had hoped to convince Anthropic CEO Dario Amodei to assume Altman’s role, but Amodei, who had previously worked at OpenAI, quickly declined. After all, Friday’s developments represented a massive opportunity for the second-place competition.

Following his departure, Altman had begun pitching investors on a new AI venture, which Brockman was expected to join, The Information reported. But by Saturday afternoon, an outpouring of support for the ousted CEO from employees and investors shifted the conversation to a topic that had previously seemed unthinkable: Altman’s return.

A group of major investors in OpenAI, including Tiger Global Management, worked behind the scenes to get Altman reinstated. On Saturday evening, Vinod Khosla posted on X that his firm, Khosla Ventures, one of OpenAI’s earliest investors, wanted Altman back at the helm of the company, “but will back him in whatever he does next.”

Sequoia Capital, another OpenAI investor, pushed for Microsoft executives to help engineer Altman’s return on Saturday, though the firm also indicated to Altman that it would support him if he decided to launch a new venture, The Information reported. Investment firm Thrive Capital, which had agreed to lead a tender offer valuing OpenAI at $86 billion prior to Altman’s firing, also spoke with Microsoft about his reinstatement. These early discussions made one thing clear: If Altman was to return, the board would have to be overhauled.

As day turned to night and rain flooded the streets of San Francisco, close confidants flocked to Altman’s $27 million mansion in Russian Hill. Unlike the surrounding buildings—bay-windowed Victorians painted in bright colors—Altman’s starkly modern home, tucked away behind a maze of plants and trees, is hardly visible from the sidewalk. The group of staffers and friends convened at the impromptu war room to learn more about the company’s future and Altman’s fate, an employee leaving the residence told The Information.

The tumult had particularly rattled employees. Over the course of Saturday, some OpenAI staffers who had planned to participate in the company’s previously announced tender offer had grown concerned that the potentially life-changing payday was in jeopardy due to recent events.

OpenAI’s competitors, meanwhile, smelled blood in the water. As The Information reported, rivals like Cohere, Adept and Google’s DeepMind spent the weekend eagerly fielding résumés from AI researchers, scientists and engineers who had helped OpenAI maintain its competitive edge.

Those who remained were acutely aware of the risk of hemorrhaging such valuable talent. Murati began pleading with employees not to make any drastic moves, according to someone with knowledge of the situation, promising that she had one more hail Mary up her sleeve.

Meanwhile, there were other high-level efforts to boost morale. Chief Strategy Officer Jason Kwon sent a note to staff on Saturday night that underscored how drastically the mood had shifted within the company in just 24 hours. He wrote that he was “optimistic” OpenAI could bring back Altman, Brockman and the other key employees who had departed in response to Altman’s firing, The Information reported.

Around 8 p.m., people started to descend the steps outside Altman’s home. Most left quickly, jumping into Ubers waiting outside.

At 8:47 p.m., Altman wrote on X, “i love the openai team so much.” Dozens of OpenAI employees reposted his words with different-colored hearts. The frenzied armchair analysis continued online, as journalists and X users attempted to decipher the hidden meaning of the emojis amid the rapidly developing situation.

Around 9:30 p.m., an OpenAI employee arrived at Altman’s home and was buzzed inside, but he left shortly thereafter. He was told Altman was going to sleep.

---

The Counter-Offensive: Sunday, Nov. 19

On Sunday morning, all was quiet near Altman’s home. An Amazon delivery driver attempted to drop off a package, but when no one answered the door, he lobbed it over the driveway gate. Someone else made trips back and forth from the house to a car, getting rid of what looked like empty pizza boxes and returning with grocery bags and boxes of Topo Chico mineral water.

Murati and other OpenAI leaders had invited Altman and Brockman to the company’s San Francisco headquarters Sunday morning as part of the push for their rehiring. Murati told staff that her team was “the first to ask the board to reinstate OpenAI to its previous state.”

By this point, Nadella was personally assisting Murati with negotiations to bring Altman back on board. These discussions also made it clear that Microsoft, OpenAI’s largest investor, was jockeying for a role on the newly reshuffled board of directors, either in the form of a board seat or as a board observer without voting power.

Around lunchtime, employees started pulling up to the OpenAI offices at 1960 Bryant Street on bicycles and in Teslas and self-driving cars. Will Hurd, a former Republican Congressman who resigned from OpenAI’s board in July 2023, was spotted on the premises, as was research chief Bob McGrew, one of many executives known to be pushing for Altman’s and Brockman’s return.

At 12:45 p.m., a delivery person arrived at company headquarters toting bags of drinks from Boba Guys. A second boba tea delivery appeared minutes later. Around 1:30 p.m., a third (and seemingly final) round of bobas was delivered to the office. Sufficiently plied with tapioca balls, the group continued its work. Meanwhile, Altman posted a sardonic selfie on X, posing with a guest badge at the OpenAI offices and writing, “first and last time i ever wear one of these.”

Despite the high spirits, the negotiations that Altman and Brockman had been summoned for failed. On Sunday evening, the board announced that it had named former Twitch CEO and co-founder Emmett Shear interim CEO of OpenAI, and demoted Murati to her previous position of CTO. The announcement dashed the hopes of many employees that a deal might be struck to reinstate Altman that day. In an internal meeting, Sutskever told staff he and the rest of the board stood by their decision as the “only path” to defend the company’s mission of creating AI that would benefit all of humanity.

Around 9 p.m., most of the OpenAI staffers who’d come in that day departed from the company headquarters, looking defeated. An hour later, an exhausted Sutskever also left the building.

In the days that followed, it would emerge that Shear had been far from the board’s first choice. As The Information reported, the board had earlier approached others about the open CEO role: Amodei, co-founder and CEO of OpenAI’s chief rival, Anthropic; Nat Friedman, former CEO of Microsoft-owned GitHub and a prolific AI investor; and Alex Wang, co-founder and CEO of Scale AI. All three swiftly rejected the offers.

According to The Verge, some employees boycotted an emergency all-hands meeting on Sunday that was intended to serve as an introduction to the new interim CEO Shear. They also allegedly responded to news on Slack of his hiring with a “fuck you” emoji. That night, OpenAI locked some employees out of their company computers.

Then, as much of the East Coast slept, arrived one final plot twist: Around midnight, Microsoft CEO Nadella announced on X that Altman and Brockman would join his company to “lead a new advanced AI research team.” The post also promised that Nadella and his colleagues “remain committed to our partnership with OpenAI and have confidence in our product roadmap.”

What had begun as an overthrow of one man, Altman, had turned into a coup for one of tech’s most powerful companies.

---

Rallying the Troops: Monday, Nov. 20

November 20 was supposed to kick off a week of paid vacation for the entire staff of OpenAI in recognition of the grueling year they’d spent shipping updates to their hit product, ChatGPT. Some employees had already boarded flights for the Thanksgiving holiday, but they wouldn’t be able to exhale just yet. Coming off a sleepless weekend, they found a fresh round of chaos waiting for them.

After midnight, an open letter circulated among OpenAI’s staff, forcefully denouncing the board’s decision to fire Altman and fundamentally questioning the members’ fitness to serve. “We, the undersigned, may choose to resign from OpenAI and join the newly announced Microsoft subsidiary run by Sam Altman and Greg Brockman,” the letter stated. It continued, “We will take this step imminently, unless all current board members resign, and the board appoints two new lead independent directors, such as Bret Taylor and Will Hurd, and reinstates Sam Altman and Greg Brockman.”

Around 2 a.m., dozens of OpenAI employees—including top executives like Murati and Lightcap—began posting the same cryptic message on X: “OpenAI is nothing without its people.” Altman reposted several of them with heart emojis.

In a shocking turn of events, Sutskever—one of the chief orchestrators of the so-called coup—appeared to reverse course in the middle of the night. At 5:15 a.m., Sutskever posted on X, “I deeply regret my participation in the board’s actions. I never intended to harm OpenAI. I love everything we’ve built together and I will do everything I can to reunite the company.” Later that morning, Anna Makanju, head of global affairs at OpenAI, felt compelled to inform colleagues that Sutskever had confirmed to her his signature on the ultimatum letter was not a prank.

His change of heart had come after intense deliberations with OpenAI employees, as well as an emotional conversation with Greg Brockman’s wife, Anna, according to a person familiar with the talks. Anna pleaded tearfully with him to change his mind, they said.

Yet the rest of the board was still standing by its decision to hire Shear. Some OpenAI investors had already begun exploring legal recourse against the board, fearing that its actions would cause the company to collapse and billions of dollars in value to vanish. Khosla took to X to encourage his followers to send Shear direct messages pressuring him to resign.

As the board, management and investors squabbled, OpenAI’s staff found solidarity within their ranks. By Monday afternoon, over 700 of the company’s approximately 770 employees had signed the open letter, threatening to quit unless the board resigned and reinstated Altman. Just a handful of holdouts, including effective altruism–aligned members of Sutskever’s Superalignment team, abstained from signing.

On Monday afternoon, minutes after Microsoft finished off the trading day with shares at a new record price of $377.44, Nadella appeared on CNBC’s “Closing Bell: Overtime.” He made it clear that while the door remained open for Altman, Brockman and others from OpenAI, he would also be amenable to their staying at OpenAI. “I’m open to both options,” he said.

Late Monday night, Bloomberg reported that Makanju had sent a note to staff hoping to buy some more time. Makanju and the other members of OpenAI management who’d assumed the role of mediators, were in touch with Altman, Shear and the board, she wrote, but “they are not prepared to give us a final response this evening.” She did, however, reassure staffers that “our number one goal remains to reunify OpenAI.” She added: “We have a plan that we are working towards.”

---

The Return of the King: Tuesday, Nov. 21

By Tuesday, the events of the week had officially gone from tech insider intrigue to a topic of widespread conversation. But for the parties directly involved, it was starting to feel like Groundhog Day. Rumors continued to swirl that investors were lining up lawsuits to file against the OpenAI board. Even Shear was allegedly threatening to quit as interim CEO “if the board [couldn’t] clearly communicate to him its reasoning for Altman’s sudden firing,” according to Bloomberg.

Reports did, however, emerge that the board was finally engaging with Altman after having gone dark on Saturday. On Altman’s side of the table was his close friend and adviser, Airbnb CEO Brian Chesky. Meanwhile former Salesforce co-CEO Bret Taylor stepped in as a neutral mediator. By this point, Adam D’Angelo was thought to be the toughest holdout and the leader of the board’s resistance, and Shear represented him in discussions.

Somehow, amidst all the chaos, the OpenAI team managed to ship a new feature that afternoon: ChatGPT with voice for all free users. Brockman, who was effectively on leave from the company, cheerfully posted the news on X. Some who logged into ChatGPT in the following hours encountered an error page stating that the software had exceeded user capacity.

Just when some people were starting to feel like resolution was in sight, Elon Musk arrived to add fuel to the fire. On Tuesday afternoon, the former co-founder of OpenAI posted a letter on X that had already been circulating on online message boards, addressed to OpenAI’s board of directors from an anonymous group of employees. Though its veracity is yet unconfirmed, the letter appeared to offer some concrete claims about Altman’s and Brockman’s misbehavior. The letter called on Shear to expand his investigation to address “Sam Altman’s actions since August 2018, when OpenAI began transitioning from a non-profit to a for-profit entity.”

Around 9 p.m., leadership sent out the same email they’d been sending to staff every night, letting them know they could go to bed without any updates. They were wrong.

At 10:01 p.m. OpenAI posted on X, “We have reached an agreement in principle for Sam Altman to return to OpenAI as CEO.” They also announced a new initial board with former Salesforce co-CEO Taylor as chair, former secretary of the treasury Larry Summers and D'Angelo taking board seats. Toner and McCauley were off the board. On X, users advised the former board members to close their DMs.

Following the news, hundreds of employees—including Altman and Brockman—gathered at the OpenAI offices in San Francisco, hugging and drinking. Brockman gave a speech toasting to the cathartic end of a rollercoaster week, and later posted a team selfie on X with the words, "We are so back." Sutskever was notably absent from the festivities.

At one point, the building’s fire alarm went off, sending everyone spilling out into the courtyard where celebrations continued as two fire trucks roared onto the scene. Firefighters made their way through the throngs of revelers and into the building, where they confirmed that there wasn’t a fire at all—just a lot of smoke from a fog machine that had been set up in the office.

They gave the all clear and the party raged on, as if nothing had happened.