Srinivas Narayanan, Kevin Weil, Bill Peebles: OpenAI lost 3 executives in one day as science division shuts down
In a single Friday, OpenAI ousted three senior leaders, shuttered a chapter it had publicly championed just months earlier, and quietly signaled that its scientific AI ambitions were being incorporated into a coding exercise; All this while the company is accelerating towards a reported initial public offering.
Three OpenAI Executives Leaving the Same Day
Approved on April 17, 2026, the separations were implemented as a restructuring memo. Kevin Weil, who first served as OpenAI’s chief product officer before leading OpenAI’s science initiative, has publicly announced his departure. Srinivas Narayanan, the company’s chief technology officer for enterprise applications, told colleagues within the company that he was leaving to spend time with his family.
Bill Peebles, who was head of the Sora video production product, announced at X that he is also leaving.
Three top-level departures in a single day at any major tech company is remarkable. At OpenAI, where executive turnover is becoming closer to a recurring event, it appears to be the latest chapter in a leadership story that has been unraveling for some time.
Weil addressed his departure directly on social media: “Today is my last day at OpenAI, as OpenAI for Science is being rolled out to other research teams. The two years from Chief Product Officer to joining the research team to launching OpenAI for Science have been a mind-expanding process.”
What Was OpenAI for Science and Why Was It Dissolved?
Until recently, OpenAI for Science was one of the company’s most visible bets on the transformative potential of advanced AI. Weil joined OpenAI as chief product officer in June 2024, bringing a product background from Instagram and Twitter. By September 2025, he had moved internally to lead the initiative designed to attract world-class academics and apply GPT-5’s capabilities to challenging problems in physics, biology, and chemistry.
In January 2026, the team launched Prism, a custom web application that aims to provide researchers with a dedicated AI workspace. That same month, MIT Technology Review profiled the startup as active and ambitious, noting GPT-5.2’s score of 92 percent on the GPQA graduate-level science benchmark, a striking jump from GPT-4’s 39 percent. In this profile, Weil spoke at length about the importance of “epistemological humility” and self-verification in AI models.
Three months later the episode no longer exists. Prism ended and the team of about ten people who built and ran it were folded into Codex (OpenAI’s AI coding application) under the direction of Codex president Thibault Sottiaux.
An OpenAI spokesperson explained the changes WIRED As part of a broader effort to unify the company’s business and product strategy. At the same time, OpenAI announced GPT-Rosalind, a new model series designed for life sciences researchers.
OpenAI’s Codex is Positioned as the ‘Everything App’
The inclusion of the Prism team in Codex is not an isolated decision. This is consistent with the strategic pivot that has been evolving since at least March 2026, when Fidji Simo, OpenAI’s general manager of AGI deployment, told staff that the company needed to simplify its product offerings. Sora video creation app has already been discontinued. Prism is no more. The pattern becomes clear.
Accordingly WIREDOpenAI “has broader ambitions to transform AI coding app Codex into an ‘everything app’.” Rather than maintaining a portfolio of specialized tools for different domains like science, video, enterprise, and more, the company appears to be consolidating its brokerage targets into a single product surface. For developers and researchers building using OpenAI’s APIs, this means the integration points they rely on are actively changing.
The fates of Sora and Prism show that custom implementations in OpenAI have a short shelf life when their commercial viability is not immediately obvious.
OpenAI’s Loss of Leadership: A Pattern That Preceded This Week
Friday’s triple-header does not exist in a vacuum. The scale of OpenAI’s executive turnover over the last eighteen months is, by any measure, unusual for a company of this caliber.
OpenAI’s chief technology officer, Mira Murati, leaves in 2024. Greg Brockman, the company’s co-founder and president, stepped back to an active role overseeing products after Simo took medical leave.
Marketing chief Kate Rouch also resigned for health reasons. Operations chief Brad Lightcap moved into a loosely defined “special projects” role. Now, in a single day: gone is a former CPO, a company CTO, and a product manager.
IPO Pressure Behind OpenAI’s Restructuring
OpenAI raised $6.6 billion at a $157 billion valuation in October 2024 and reported 300 million weekly active ChatGPT users at the time. But the competitive environment it now faces is significantly different from the one it dominated just two years ago.
Anthropic, OpenAI’s closest rival in the lead model race, reportedly grew ninefold in revenue year over year. Google DeepMind continues to have significant credibility in scientific AI applications, powered by the influence of AlphaFold. With OpenAI reportedly aiming for an IPO in late 2026, the pressure to deliver a coherent, simplified product narrative to prospective public market investors is considerable.
What’s Next for OpenAI? and Products Created on It
For those building products and integrations on top of the OpenAI infrastructure, this is a consolidation event with real consequences. The Codex ecosystem is about to become larger and more complex, absorbing the capabilities of Prism and other potentially retired products. The API surface is changing. And the leadership instability that now touches product, enterprise, video, science, operations and marketing raises legitimate questions about the continuity of roadmaps.
