The Open AI Architects: Meet the People Behind the Staggering IPO Coming to Wall Street

PART 2 OF 3

SPECIAL REPORT: THE OPENAI IPO

They started it as a nonprofit to save the world. Then they fought over it, fired each other, walked out, sued each other, and built competing companies. This is the story of the people who created the most consequential technology company of our time — and what each of them is worth now.

PART ONE: THE FOUNDING — A BILLION-DOLLAR NONPROFIT IN A VICTORIAN MANSION

The Room Where It Happened

In December 2015, eleven people sat around a table in San Francisco and agreed to build the most powerful technology in human history — as a charity. Not a company with a public mission statement. An actual 501(c)(3) nonprofit, legally prohibited from distributing profits to its founders, with a board of independent directors, and a founding document that committed the organization to the benefit of 'all of humanity' rather than its shareholders, because its founders believed there were no shareholders.

The pledges made that evening totaled $1 billion — a sum that, at the time, exceeded the endowments of most American universities. Elon Musk pledged the largest share, eventually contributing approximately $44 million in cash before his departure in 2018. Sam Altman pledged $1 million. Greg Brockman and Ilya Sutskever left paying positions at established companies to take no salary for the first year. The building they worked in was a rented Victorian in the Mission District. The coffee was from a local café. The computational infrastructure was, in the early months, borrowed.

What drove these eleven people to that table was not primarily the prospect of wealth — though wealth, in quantities none of them could have fully imagined that December, is precisely what the OpenAI IPO now promises. What drove them was fear. Fear of what was being built inside the research labs of Google, Meta, and Microsoft without any independent oversight, without any genuine commitment to the public good, and without any mechanism to pump the brakes if the technology began to develop in ways its builders did not intend. The founders of OpenAI believed, genuinely and urgently, that artificial general intelligence was coming — and that the world was not ready.

"They didn't leave for money. They left because they were afraid of what was being built without them — and of what would happen if nobody with a conscience was in the room."

The Original Eleven

The eleven founding members of OpenAI were: Elon Musk, the entrepreneur and engineer who would simultaneously be the company's most generous initial donor and its most consequential eventual antagonist; Sam Altman, then the president of Y Combinator; Ilya Sutskever, a Ukrainian-born AI researcher who had co-authored the AlexNet paper that launched the deep learning revolution and was at the time a senior researcher at Google Brain; Greg Brockman, the former chief technology officer of Stripe; and seven additional researchers — Trevor Blackwell, Vicki Cheung, Andrej Karpathy, Durk Kingma, John Schulman, Pamela Vagata, and Wojciech Zaremba — whose combined technical contributions would shape the trajectory of artificial intelligence research for the decade that followed.

Of the eleven, three remain in active leadership roles at OpenAI as it approaches its public offering: Altman, Brockman, and Zaremba. The other eight have departed to found competitors, to conduct independent research, to manage companies of their own, or to argue publicly about the right way to build artificial general intelligence. The exits have been consequential, the debates have been fierce, and the fortunes — on both sides — are extraordinary. Together, the eleven founders of OpenAI will collectively be among the most significant wealth creators in the history of Silicon Valley, even the ones who left.

 PART TWO: THE CURRENT LEADERSHIP — THE THREE WHO STAYED

Sam Altman — Chief Executive Officer

Sam Altman was born on April 22, 1985, in Chicago, Illinois, and grew up in St. Louis, Missouri, in a household that encouraged intellectual ambition in the specific, practical form of early access to computers. By the age of eight, he had taught himself to code. By his early teens, he had developed a clear sense that technology was not merely an interest but a vocation. His mother is a dermatologist. His father was in real estate. The combination of a rigorous scientific household and an entrepreneurially minded one produced, in Altman, an unusual hybrid: someone equally comfortable with research literature and capitalization tables.

He enrolled at Stanford University to study computer science, lasted two years, and dropped out in 2005 to found Loopt, a location-based social networking company. The timing was slightly too early for mobile-first social networking — the smartphone revolution that would have made Loopt a phenomenon came two years after he launched it — but the company attracted venture capital, survived for seven years, and was acquired in 2012 by Green Dot Corporation for approximately $43 million. Altman walked away with a modest outcome and an education in what it takes to build a technology company from nothing.

His real education came next, at Y Combinator. Altman joined YC as a partner in 2011 and became its president in 2014, at the age of twenty-eight, succeeding Paul Graham, who had built the institution into the dominant force in early-stage startup investing. In his five years running YC, Altman expanded the program's batch sizes, internationalized its reach, launched YC Research as an independent research arm, and personally invested in and advised a generation of companies — Stripe, Airbnb, Reddit, and Dropbox among them. When he left YC to run OpenAI full-time in 2019, he carried with him relationships, institutional knowledge, and a reputation as one of the sharpest strategic minds in Silicon Valley.

The paradox of Sam Altman's position as OpenAI prepares for its IPO is the most discussed anomaly in American corporate finance. He runs the most valuable private technology company in the history of American enterprise. He has done so for seven years, navigating a boardroom coup, a restructuring, multiple regulatory inquiries, a historic patent lawsuit, and the transformation of AI from academic curiosity to global phenomenon. His annual salary is $76,001. He owns no equity in the company.

That last fact — now being remedied as part of the PBC conversion — has produced a unique situation. Altman's disclosed net worth, estimated at approximately $2 to $3.3 billion by various outlets, derives entirely from investments he made before OpenAI existed: a Reddit stake (approximately 8.7 percent of shares outstanding at the time of the 2024 IPO), positions in Stripe, Airbnb, and other YC alumni companies, and a $1.7 billion stake in Helion Energy, a nuclear fusion startup. He has none of the paper fortune that his peers — from Dario Amodei at Anthropic to Elon Musk at SpaceX — carry into their companies' public offerings. The equity grant expected to be formalized in the S-1 will, for the first time, place Altman on the capitalization table of the company that his leadership has made worth nearly a trillion dollars.

"The CEO of the most valuable private technology company in American history earns $76,001 a year and owns none of it. The S-1 will tell us how they finally fix that."

Greg Brockman — President and Co-Founder

Greg Brockman grew up in Thompson, North Dakota, a small agricultural town whose population hovers around one thousand people. His father was an engineer. His mother taught elementary school. The environment was not conventionally associated with the production of Silicon Valley luminaries, but it produced in Brockman exactly the qualities that define the best engineers: meticulous attention to how things work, comfort with complexity, and the patient willingness to rebuild something from scratch when it fails.

Brockman was admitted to Harvard University, where he studied mathematics, and then transferred to MIT — following the pull of a more technically rigorous environment — before dropping out of MIT in 2010, at age twenty-two, to join Stripe as its third employee. His time at Stripe, where he eventually became chief technology officer and helped build the company's engineering culture and product infrastructure during its critical growth phase, made him one of the most respected engineering executives of his generation. When Altman and Musk came to him in 2015 about founding OpenAI, Brockman had the credibility to attract world-class researchers to a nonprofit with no product and no revenue. That credibility was itself a form of capital.

At OpenAI, Brockman has served as the organizational glue between the research and engineering functions — the person who translated Ilya Sutskever's research vision into something that an engineering organization could actually build, test, ship, and scale. He was among the first to be fired by the board in November 2023 — resigning immediately upon Altman's ouster in solidarity — and among the first to be reinstated when the board collapsed. He took an extended sabbatical in 2024, returning in November with a post on X: 'Back to building OpenAI.' The sabbatical has not been fully explained publicly; its circumstances will likely be a topic of analyst and journalist inquiry during the IPO roadshow.

Brockman's equity stake in OpenAI, while not publicly disclosed with precision, is expected to be a founding co-founder position of significant size. At the $852 billion valuation, even a one percent stake represents $8.52 billion in paper wealth. His actual position is believed to be materially larger than that, though the precise figure will be disclosed for the first time in the S-1.

Wojciech Zaremba — Co-Founder and Research Leader

Wojciech Zaremba was born in Kluczbork, Poland, and studied computer science at the University of Warsaw before pursuing doctoral research at New York University and spending time at Google Brain and Facebook AI Research. He is, among the three remaining founding co-founders of OpenAI, the least publicly visible — which is consistent with the culture of serious AI researchers who prefer papers to press and benchmarks to brand management. His contributions to OpenAI's research have been substantial, particularly in the areas of reinforcement learning and the development of the technical infrastructure underlying the API products that generate much of the company's enterprise revenue. His equity position, like Brockman's, has not been disclosed publicly but is expected to represent a meaningful founding-era stake in the S-1.

PART THREE: THE DEPARTED — AND WHERE THEY WENT

Elon Musk — Co-Founder and Co-Chairman (2015–2018)

The largest early donor to OpenAI, and the most complicated chapter in its history, is a man who no longer works there. Elon Musk's relationship with OpenAI traces a nearly perfect arc of the American technology drama: visionary patron, strategic disagreement, bitter departure, public war, federal lawsuit, jury verdict, and appeal.

Musk arrived at the founding with both the most money and the most insistence on structural control. The concerns he brought to the table in 2015 were genuine: he believed, more viscerally than almost anyone in public life, that advanced AI posed existential risks to humanity, and he wanted OpenAI to be the institution that would ensure those risks were managed. As the company grew and as its research began producing results that made the frontier of AI feel both closer and more consequential, Musk's desire to direct that research — and to do so as CEO or chairman of an institution he was funding — began to conflict with the other founders' views of governance.

By 2018, the conflict was irresolvable. Musk proposed taking control of OpenAI, a proposal that Altman and the other founders rejected. Shortly thereafter, Musk resigned from the board, citing a need to avoid conflicts of interest as Tesla developed its own AI capabilities for its Autopilot and Full Self-Driving programs. The public account was polite. The private account, as presented in the May 2026 trial, was more contentious: Altman testified that Musk had made demands for control that the board could not accept, and that when he left he 'left the nonprofit for dead,' withdrawing his promised future funding.

In 2023, Musk founded xAI, a direct competitor to OpenAI, and launched Grok — a large language model integrated into the X platform that he had acquired for $44 billion the prior year. xAI raised $42 billion since inception, achieved a $230 billion valuation before its absorption into SpaceX in February 2026, and now sits as a division of the combined SpaceX-xAI entity that is currently pursuing its own historic public offering. In 2024, Musk filed the federal lawsuit that was dismissed in May 2026. His financial position — built primarily on his equity in Tesla, SpaceX, and xAI — makes him the wealthiest person in the history of human civilization, with a net worth that various estimates place between $200 billion and $350 billion, depending on the current price of Tesla shares and the valuation of his private holdings. He has announced his intention to appeal the jury verdict in his case against OpenAI.

Ilya Sutskever — Co-Founder and Chief Scientist (2015–2024)

If there is a figure in artificial intelligence whose reputation approaches the mythic — who is simultaneously the most celebrated and the most enigmatic of the founding generation of AI researchers — it is Ilya Sutskever. Born in 1986 in Nizhny Novgorod, Russia, Sutskever moved with his family to Jerusalem when he was five years old, and from there to Canada, where he studied mathematics and eventually pursued graduate work in computer science at the University of Toronto under Geoffrey Hinton. The collaboration with Hinton produced the AlexNet paper of 2012 — the ImageNet competition result that won by such a large margin that it convinced the entire field that deep learning was the future of artificial intelligence. That paper, and the research it represented, is arguably the foundational document of the modern AI era.

Sutskever joined Google Brain after completing his graduate work, spent two years as one of the most productive researchers in the history of the field, and then left to co-found OpenAI with Altman, Musk, Brockman, and the rest in December 2015. As chief scientist, he was the intellectual center of gravity of OpenAI's research program — the person whose instincts about what to build next, and why, shaped the direction of an institution that would ultimately change the world. He co-invented Reinforcement Learning from Human Feedback, the technique that made ChatGPT useful to ordinary people. He drove the development of GPT-2 and GPT-3. He was, by any external assessment, among the two or three most important researchers at OpenAI.

He was also, in November 2023, one of the four board members who voted to fire Sam Altman — a decision he publicly stated, within days, that he 'deeply regretted.' The following months at OpenAI, with a new board and a restructured leadership team, were apparently not sufficient to restore Sutskever's sense of alignment with the company's direction. In May 2024, he announced his departure. His stated reason: a desire to pursue 'personally meaningful work.' The unstated reason, as interpreted by every AI researcher and journalist who covers the field, was a conviction that OpenAI's safety priorities had been subordinated to its commercial ones — a conviction shared by the colleagues who left OpenAI to found Anthropic three years earlier.

In June 2024, Sutskever co-founded Safe Superintelligence Inc. with Daniel Gross, the former head of Apple's AI efforts, and Daniel Levy, a researcher and investor. SSI's stated mission — to build safe superintelligence as the company's first and only product — is, in its ambition and its austerity, a direct challenge to both OpenAI and Anthropic. The company has raised $6 billion at a $32 billion valuation despite having no commercial product, approximately twenty employees, and a CEO who has declined to discuss publicly what they are actually building. SSI rejected an acquisition offer from Meta Platforms in mid-2025. Sutskever's equity stake in SSI, at the current $32 billion valuation, is believed to represent a multi-billion-dollar position. His equity in OpenAI, as a founding member, will be disclosed in the S-1.

"Sutskever co-invented the technique that made ChatGPT work. Then he voted to fire the CEO. Then he said he regretted it. Then he left to build something even more powerful, more carefully. The AI field has never had a drama quite like this."

Andrej Karpathy — Co-Founder (2015–2017, 2023)

Andrej Karpathy was born in Bratislava, Slovakia, grew up in Toronto, and studied computer science at the University of Toronto — overlapping, in the small world of deep learning, with Sutskever's graduate years, though they did not work directly together until OpenAI. Karpathy completed his PhD at Stanford under Fei-Fei Li, one of the pioneers of computer vision, and was among the early hires at OpenAI when it was founded in 2015. He spent two years at the company before leaving to lead Tesla's Autopilot AI team — a role in which he spent four years and produced some of the most ambitious real-world applications of deep learning for autonomous systems in the history of the field.

He returned briefly to OpenAI in February 2023, during the post-ChatGPT frenzy when every AI research organization was competing for talent with an intensity not seen since the dot-com era. He left again in July 2023, citing a desire to pursue independent work — specifically, he has devoted a significant portion of his post-OpenAI time to producing educational content about deep learning that has earned him an extraordinary following among students and practitioners worldwide. His YouTube tutorials on training large language models from scratch have been watched by millions of people seeking to understand the technology that has transformed their professional lives. His net worth, from OpenAI founding equity and Tesla compensation, is estimated to be substantial but has not been publicly disclosed.

John Schulman — Co-Founder (2015–2024)

John Schulman's name is not as widely known to the general public as Altman's or Musk's, but within the AI research community his contributions occupy a central place. Schulman co-developed Proximal Policy Optimization — PPO — one of the most widely used reinforcement learning algorithms in the field, and was a principal architect of Reinforcement Learning from Human Feedback, the technique that taught AI models to be helpful by learning from human preference judgments rather than raw data alone. RLHF is, in the most direct technical sense, why ChatGPT is ChatGPT: why it responds to questions in ways that feel natural and useful rather than statistically likely, why it declines to produce harmful content, why it acknowledges uncertainty. Schulman's work made the product possible.

He left OpenAI in August 2024 and joined Anthropic — a move that was, by any standard, the most commercially significant talent transfer in the history of the AI industry. It was not a simple job change. It was a defection, in the sense that the phrase is used in research communities: a researcher who had built the foundational techniques of his former employer's most valuable products crossing the line to the most direct competitor. He was explicit about his reasons, stating publicly that his decision was 'rooted in a desire to deepen my focus on AI alignment.' The implication — that OpenAI had not provided sufficient focus on alignment — was not lost on anyone in the industry. His equity stake from his OpenAI founding position will be disclosed in the S-1. His role at Anthropic, where he joins former OpenAI colleagues Dario Amodei, Daniela Amodei, Tom Brown, and others, places him at the institution where the most consequential personnel convergence in the history of AI safety research has occurred.

Mira Murati — Chief Technology Officer (2018–2024)

Mira Murati is not a co-founder of OpenAI — she joined the company in 2018, three years after its founding, as VP of Applied AI and Partnerships. But her influence on what OpenAI became is inseparable from the company's story, and her departure in September 2024 removed from its leadership team one of the most respected executives in the history of Silicon Valley.

Murati was born in Vlorë, Albania, and immigrated to Canada at the age of sixteen, where she attended the Pearson United World College of the Pacific in British Columbia on a scholarship. She earned an engineering degree from Dartmouth College and worked at Goldman Sachs, Leap Motion, and Tesla before joining OpenAI. Her rise within the company was remarkable: from VP of Applied AI to CTO in 2022 to acting CEO during the five chaotic days of November 2023, when the board's implosion left the company without a functioning chief executive and Murati held it together long enough for Altman's reinstatement to be negotiated.

She announced her departure in September 2024 with a brief statement saying she wanted 'to create the time and space to do my own exploration.' She left alongside two other senior executives: Bob McGrew, the chief research officer, and Barret Zoph, the VP of research. The simultaneous departure of three senior leaders in the research and technology leadership was significant enough that Altman acknowledged publicly that the timing had been coordinated for operational continuity. Murati's next chapter has not been publicly announced, but her reputation — and her equity stake from six and a half years at a company now approaching one trillion dollars — positions her as one of the most formidable individuals in the AI industry.

Durk Kingma — Co-Founder

Durk Kingma co-founded OpenAI in 2015 and is best known in the research community for co-creating the Variational Autoencoder — a foundational technique in generative AI — and for co-authoring influential work on the Adam optimizer, one of the most widely used training algorithms in deep learning. After leaving OpenAI, Kingma joined Google Brain and later Google DeepMind, where he has continued his research on generative models. His original OpenAI equity stake, from a founding position in a company now approaching a trillion dollars in value, represents a paper fortune whose magnitude will be disclosed in the S-1.

Trevor Blackwell, Vicki Cheung, and Pamela Vagata — Early Founders

The remaining founding members — Trevor Blackwell, Vicki Cheung, and Pamela Vagata — each played roles in the early days of OpenAI before pursuing other paths. Blackwell, a programmer and roboticist who had co-founded Anybots and worked with Paul Graham at Y Combinator, was among the first engineers in OpenAI's initial research environment. Vicki Cheung, a researcher with a background in probabilistic programming, and Pamela Vagata, a machine learning engineer who had worked at Facebook AI Research, both contributed to early experiments and departed within the company's first years. Their founding equity stakes, whatever their current dollar values, will be among the disclosures of the S-1.

 PART FOUR: THE BOARD — WHO GOVERNS THE MOST POWERFUL AI COMPANY IN THE WORLD

Bret Taylor — Chairman

Bret Taylor is, in the assessment of virtually everyone who has worked with him in corporate governance, the right person to chair the OpenAI board at this moment in the company's history. A computer scientist and entrepreneur who co-created Google Maps, served as CTO of Facebook, and co-founded Salesforce Quip before becoming Salesforce's co-CEO, Taylor has been at the center of consequential technology decisions for two decades. He was chairman of Twitter's board during the period of Elon Musk's acquisition — a transaction managed under extraordinary legal pressure and public scrutiny, which provides a specific kind of preparation for the pressures of leading the board of OpenAI. He was recruited to chair the reconstituted OpenAI board following the November 2023 boardroom crisis, and has been the primary architect of the governance framework through which the company has navigated the PBC conversion.

Adam D'Angelo

Adam D'Angelo, the CEO of Quora and a co-founder who served as Facebook's first CTO, is one of the two board members who survived the November 2023 crisis that replaced the rest of the board. His presence on the reconstituted board has been the subject of criticism — he was among those who voted to fire Altman, and his continuation on the board was seen by some observers as inconsistent with the accountability the reinstatement drama implied. His perspective on AI, shaped by Quora's own experiments with AI-generated content, makes him a substantive voice on product and research questions.

General Paul M. Nakasone

Paul Nakasone retired from the United States Army as a four-star general after serving as the director of the National Security Agency and commander of United States Cyber Command from 2018 to 2023. His addition to the OpenAI board reflects the company's increasing engagement with national security themes — including the Stargate project, the 'Classified Stargate' proposal, and the broader framing of AI development as a geopolitical competition with China. A former NSA director brings, to any board, both substantive expertise in the intersection of technology and national security and a set of relationships within the intelligence and defense communities that are useful for a company seeking government contracts and government goodwill.

Dr. Zico Kolter — Safety and Security Committee Chair

Zico Kolter's presence on the board is, in structural terms, the most distinctive of any director. As the chair of the Safety and Security Committee, Kolter — a computer scientist at Carnegie Mellon University who specializes in the intersection of machine learning and robust optimization — is the director responsible for overseeing model safety at OpenAI. The California Attorney General's agreement that cleared the PBC conversion specifically required that the chair of the Safety Committee serve on the Foundation board but not on the PBC board, ensuring a degree of independence in safety oversight. Kolter's role, and the resources the company devotes to his committee's work, will be subjects of intense scrutiny during the IPO roadshow and long afterward.

 PART FIVE: THE WEALTH — WHAT THE IPO MEANS FOR EVERYONE IN THE ROOM

The Most Consequential Cap Table in Technology History

When OpenAI's shares begin trading — on whatever morning the market opens for the company that ChatGPT built — the resulting wealth creation will be without precedent in the history of American technology. The SpaceX IPO may raise more capital. Anthropic's listing may come at a higher per-share premium relative to revenue. But the number of people whose lives will be materially transformed by OpenAI's public offering, and the concentration of that transformation in a relatively small community of researchers, engineers, and investors, makes this one of the most consequential single wealth events in Silicon Valley history.

Microsoft's stake alone — currently valued at approximately $228 billion on a $852 billion post-money basis — represents a return on $13 billion that exceeds, in absolute dollar terms, any investment return in the history of technology venture capital. SoftBank's position, worth approximately $99 billion, would alone be the largest single investment return in SoftBank's history, exceeding the legendary Alibaba bet. Early angel investors — Reid Hoffman, Peter Thiel, and others who wrote checks when AGI was still a punchline — have achieved approximately 140 times their initial investment.

The founding team's aggregate position is estimated to be in the range of 10 to 20 percent of the company's fully diluted equity, with Brockman, Zaremba, and the early departed founders holding stakes that, at $852 billion, represent tens of billions of dollars in aggregate. The exact allocation will be disclosed in the S-1. What will not be easily quantified is the cost at which those positions were acquired — not in dollars, but in years of work on technology whose consequences none of them, when they sat around that table in December 2015, could fully foresee.

"The eleven people who founded OpenAI as a nonprofit to save humanity from AI are about to become some of the wealthiest people on earth. The irony has not been lost on any of them."

Sam Altman's Expected Grant: The Final Chapter of the Most Unusual CEO Compensation Story in Corporate History

The question of what Sam Altman's equity grant will look like has been, for two years, the most-discussed unknown in Silicon Valley. His decision to run OpenAI without equity — a choice he made deliberately, citing a desire to avoid conflicts of interest and a stated intention to donate the bulk of any eventual windfall — created a situation without precedent: the CEO of a nearly trillion-dollar company who had no direct financial stake in its success or failure. The IPO has required that anomaly to be resolved. The S-1 will disclose the terms of any grant.

Sources familiar with the process expect the grant to be structured as a combination of direct equity and performance-based awards, potentially tied to revenue milestones or market capitalization targets. Given the company's trajectory and the precedents set by comparable grants at other frontier technology companies, the package is expected to be substantial — potentially placing Altman, in paper wealth terms, among the wealthiest individuals in the United States upon the IPO. The specific structure, size, and vesting conditions will be among the most widely reported details of the prospectus.

The People Who Left — and What They're Building Now

The departures from OpenAI over the past several years have produced, inadvertently, one of the most interesting portfolios of AI safety and research talent in the world. Ilya Sutskever is building Safe Superintelligence, valued at $32 billion, focused entirely on safe AGI with no intermediate commercial products. John Schulman is at Anthropic, applying his RLHF expertise to Constitutional AI, a safety framework that Anthropic believes is more robustly aligned than OpenAI's approach. Andrej Karpathy is educating the next generation of AI researchers through YouTube, Substack, and other platforms, with an influence on the field's human capital development that no academic institution currently matches. Mira Murati has not yet disclosed her next chapter.

The competitive landscape these departures have created is, from OpenAI's perspective, genuinely concerning: the most important safety and alignment researchers who built OpenAI's foundational techniques are now distributed across the field's competing institutions, making safety research simultaneously more diverse and more diffuse. From the perspective of anyone who cares about the long-term trajectory of AI development, however, the distribution of that talent across multiple independent institutions — each with a different approach to the problem of building safe AI — may be the most important structural fact about the current AI landscape.