Three senior OpenAI executives announced their departures on the same day last week, the latest jolt in a prolonged leadership shakeup that has been reshaping one of the world’s most closely watched artificial intelligence companies. Kevin Weil, Bill Peebles, and Srinivas Narayanan each shared the news of their exits in separate posts on social media, adding fresh momentum to what observers have described as a brain drain at OpenAI.
The triple departure lands as OpenAI pulls the plug on Sora, its once-celebrated AI video generation tool, and dismantles OpenAI for Science, a research initiative focused on using AI to accelerate discoveries in fields like medicine and drug development.
Three Leaders, Three Different Exits
Kevin Weil had joined OpenAI about two years ago, coming from Instagram, where he served as head of product. At OpenAI, he rose to chief product officer before shifting to lead OpenAI for Science. In his farewell message, Weil described the experience as “a mind-expanding two years.” One day before his departure was announced, his team released GPT-Rosalind, a model built for life sciences and drug discovery research. OpenAI said the Science division is being “decentralized” — folded into other research teams, including the Codex team, which focuses on AI coding tools.
Bill Peebles built Sora from the ground up and led it through its high-profile launch, which briefly topped Apple’s App Store charts. In his departure post, he called the work “the honour and adventure of a lifetime” and credited Sora with sparking widespread investment in AI video across the industry. Srinivas Narayanan, who served as OpenAI’s chief technology officer for B2B applications, spent three years at the company helping ship ChatGPT and its developer API while growing the applied engineering team from roughly 40 people into a much larger operation. According to a person familiar with the matter, Narayanan’s exit is unrelated to the Sora and Science shake-ups; he said he is leaving to spend more time with family.
Sora’s Rise and Rapid Fall
Sora launched with enormous fanfare and quickly became one of the most-discussed AI products in the world. But the numbers told a harder story. The app peaked at around one million users before sliding to fewer than 500,000, and it was costing roughly $1 million per day to operate. The Motion Picture Association had flagged intellectual property infringement concerns tied to the platform. By the end of March, OpenAI had made it official: Sora is being discontinued. The web and mobile app versions are set to go dark on April 26, with the API following on September 24.
OpenAI has described the shutdown as a move to redirect its energy toward other areas, including robotics designed to assist people with real-world tasks. Despite the product’s commercial struggles, Peebles’ claim that Sora changed the AI video landscape is hard to dispute — competitors raced to build similar tools after its debut.
A Pattern of Leadership Attrition
Friday’s three departures are part of a much larger trend. Of OpenAI’s 11 co-founders, only two remain: CEO Sam Altman and president Greg Brockman. Over the past two years, the company has lost co-founder and chief scientist Ilya Sutskever, former CTO Mira Murati, chief research officer Bob McGrew, VP of research Barret Zoph, co-founder John Schulman, and at least 12 additional senior executives in 2025 alone.
Many of those who left have ended up at direct competitors. Schulman joined Anthropic. Tim Brooks, who co-led Sora before Peebles, eventually landed at Meta’s Superintelligence Labs. Shengjia Zhao, a key architect behind ChatGPT and GPT-4, became chief scientist at Meta Superintelligence Labs as well. Roughly seven additional researchers followed a similar path to Meta.
The timing of the latest exits is especially difficult for OpenAI because other senior leaders are also unavailable. Fidji Simo, the company’s chief of product and business, took medical leave in early April due to a worsening neuroimmune condition, leaving Brockman to temporarily oversee product. Kate Rouch, the chief marketing officer, stepped down to focus on her cancer recovery. Brad Lightcap, the chief operating officer, transitioned into a newly created role focused on “special projects.”
Enterprise Focus Comes at a Cost
OpenAI’s pivot toward enterprise AI is unmistakable. The company now generates roughly $2 billion in monthly revenue, with an annualized run rate exceeding $25 billion. Enterprise business already makes up more than 40% of total revenue. To signal where things are headed, OpenAI brought in Denise Dresser, the former CEO of Slack, as chief revenue officer.
But the pivot has a price. OpenAI projects $14 billion in losses this year against the $25 billion in revenue. Multiple former employees have described a cultural drift — from bold, long-horizon research bets toward the faster, narrower work of making ChatGPT faster and more reliable for business customers. That shift has driven away some of the people who were drawn to OpenAI’s original mission in the first place.
Meanwhile, competition is intensifying. Anthropic’s annualized revenue has reportedly reached $30 billion, achieved at a fraction of OpenAI’s spending. Google’s Gemini is embedded across its enterprise products. Meta is building a superintelligence lab largely staffed by former OpenAI researchers. Three executives leaving on one Friday is not a company in freefall — but as a pattern, it points to an OpenAI that is becoming something quite different from the one that launched ChatGPT.
