Summary: Three senior OpenAI executives, former CPO Kevin Weil, Sora head Bill Peebles, and enterprise CTO Srinivas Narayanan — departed on the same day as the company shuts down “side quests” including Sora (discontinuing 26 April) and dismantles OpenAI for Science. The exits continue a two-year pattern that has seen only 2 of 11 co-founders remain, with departing talent flowing to Anthropic, Meta’s Superintelligence Labs, and startups, as OpenAI pivots to enterprise AI on $25 billion in annualised revenue against projected $14 billion losses.

Three senior OpenAI executives departed on the same day this week, continuing a pattern of leadership attrition that has now claimed the majority of the company’s original leadership team. Kevin Weil, the former chief product officer who had been leading OpenAI for Science, Bill Peebles, the head of Sora, and Srinivas Narayanan, the chief technology officer of enterprise applications, all announced their exits on Friday. The departures come as OpenAI shuts down what leadership internally calls “side quests,” consumer-facing moonshot projects that no longer fit a company pivoting hard toward enterprise AI.

Weil, who joined OpenAI roughly two years ago from Instagram where he had been head of product, said it had been “a mind-expanding two years.” He had moved from the CPO role to lead OpenAI for Science, a research initiative that released GPT-Rosalind, a life sciences and drug discovery model, one day before his departure was announced. The team is being absorbed into other research groups. Peebles, who built Sora from scratch, described the experience as “the honour and adventure of a lifetime” and credited the project with sparking “a huge amount of investment in video across the industry.” Narayanan, who spent three years at OpenAI helping ship ChatGPT and the API while growing the applied engineering team from roughly 40 people to a major operation, said he was leaving to spend time with family.

The end of side quests

The departures are inseparable from the strategic decisions that preceded them. Sora, OpenAI’s AI video generation tool, is being discontinued. The web and app versions will shut down on 26 April, with the API following on 24 September. The product peaked at around one million users before collapsing to fewer than 500,000, while costing roughly $1 million per day to operate. The Motion Picture Association had reported intellectual property infringement on the platform. Despite its commercial failure, Peebles was right that Sora catalysed industry-wide investment in AI video; its shutdown is a concession that OpenAI could not make the economics work, not that the technology was unimportant.

OpenAI for Science is being “decentralised,” a corporate euphemism for dismantlement. The team’s work will continue within other research groups, but the dedicated initiative that Weil was hired to lead no longer exists as an independent unit. The pattern is consistent: OpenAI is consolidating around its core revenue-generating products, ChatGPT and the API, and shutting down exploratory projects that do not contribute directly to the enterprise business.

The broader exodus

Friday’s departures add to a roster of senior exits that has reshaped OpenAI’s leadership over the past two years. Of the company’s 11 co-founders, only two remain: Sam Altman and Greg Brockman. The list of departed leaders includes co-founder and chief scientist Ilya Sutskever, CTO Mira Murati, chief research officer Bob McGrew, VP of research Barret Zoph, co-founder John Schulman, chief communications officer Hannah Wong, chief people officer Julia Villagra, and board member Lawrence Summers. At least 12 senior executives left in 2025 alone.

The destinations tell a story. Schulman went to Anthropic. Tim Brooks, who co-led Sora before Peebles, went to Google DeepMind and then to Meta’s Superintelligence Labs. Shengjia Zhao, a key architect of ChatGPT and GPT-4, became chief scientist at Meta Superintelligence Labs. Approximately seven additional researchers followed the same path to Meta. Liam Fedus, VP of research, left to co-found Periodic Labs. The talent is not retiring; it is being redistributed to competitors who are, in several cases, building the same products with the same people.

The reasons cited vary. Some executives left over ethical concerns about a Defense Department AI contract. Others described a cultural shift from ambitious research bets toward the operational grind of improving ChatGPT’s speed and reliability for Microsoft and enterprise customers. Multiple sources have noted that Anthropic’s Claude, and particularly Claude Code, has been gaining ground in developer adoption, creating competitive pressure that has intensified OpenAI’s focus on near-term product execution at the expense of longer-term research initiatives.

The leadership vacuum

The timing compounds the problem. Fidji Simo, the chief of product and business who was brought in as one of the most senior hires in OpenAI’s history, took medical leave in early April due to a worsening neuroimmune condition. Brockman is temporarily overseeing product in her absence. Brad Lightcap, the chief operating officer, has been shifted to leading “special projects.” Kate Rouch, the chief marketing officer, is departing to focus on cancer recovery.

OpenAI has added Denise Dresser, the former CEO of Slack, as chief revenue officer, signalling where the company’s priorities lie. But the net effect of the past 18 months is a leadership team that has been almost entirely reconstituted around enterprise revenue and operational execution, with the research-oriented and product-visionary executives who defined the company’s early identity largely gone.

The business context

OpenAI’s financial position is simultaneously impressive and precarious. Monthly revenue has reached approximately $2 billion, with an annualised run rate exceeding $25 billion. The company closed a $122 billion funding round in April at an $852 billion valuation and has more than 900 million weekly active ChatGPT users. Enterprise revenue now accounts for more than 40% of the total and is on track to reach parity with consumer revenue by the end of 2026.

The cost side is less comfortable. OpenAI projects $14 billion in losses on $25 billion in revenue this year, with cumulative spending through 2029 estimated at $115 billion. The company expects to reach cash-flow positive by 2029 and targets $200 billion in revenue by 2030. Those projections require everything to go right: enterprise adoption must accelerate, compute costs must decline, and the competitive threat from Anthropic, Google, and Meta must be contained. Losing the executives who built the products driving that revenue makes the path harder.

The competitive landscape has shifted materially. Anthropic’s annualised revenue has reached $30 billion while spending roughly a quarter of what OpenAI spends on training. Google’s Gemini models are embedded across its enterprise suite. Meta is building a superintelligence lab staffed significantly by former OpenAI researchers. The moat that OpenAI built through first-mover advantage in consumer AI is narrowing, and the company is responding by doing what large technology companies always do when growth becomes the imperative: consolidating around revenue, cutting exploratory projects, and losing the people who were drawn to the mission rather than the business.

Three executives leaving on the same Friday is not, in isolation, a crisis. But it is the latest data point in a pattern that has persisted for two years, and the pattern says something that OpenAI’s valuation does not yet reflect: the company that defined the generative AI era is becoming a different organisation, and the people who built it increasingly do not want to work there.