These departures are a signal, not just a staffing update
Kevin Weil and Bill Peebles leaving on the same day matters because both men were closely tied to projects that sat a little off OpenAI’s new main road. Weil had moved from chief product officer into OpenAI for Science, while Peebles led Sora, the company’s high profile video effort. Their exits came right after OpenAI cut back on what insiders and reporters have described as “side quests,” including Sora and the science workspace Prism. That does not automatically mean crisis. It does mean the company is drawing a harder line between work that feels intellectually exciting and work that fits the commercial shape of the business it is now building.
The side quest era was always going to be temporary
For a while, OpenAI could afford to look like a lab with almost endless optionality. It had the consumer heat of ChatGPT, the aura of a frontier research outfit, and enough momentum to explore tools that reached beyond its most obvious revenue engine. That kind of expansion was always attractive because it made the company look bigger than a chatbot business. It could be a video company, a science company, a coding company, a research company, and eventually a platform for almost everything. The problem is that scale changes the meaning of experimentation. Once a company becomes this large, every extra initiative competes for scarce compute, management attention, and strategic clarity. At that point, curiosity starts to look expensive unless it can be turned into something that clearly strengthens the core.
Sora showed how costly ambition can get
Sora was probably the clearest example of that problem. It was flashy, culturally loud, and easy to understand as a symbol of OpenAI’s reach. But behind the scenes, it was also costly and difficult to justify in a business that now seems more focused on practical AI work that companies will pay for again and again. Reuters reported that OpenAI dropped Sora in March, ending a planned Disney deal and shifting attention toward coding tools, corporate customers, and AGI related priorities. Reuters also said the app required significant computational resources and that some employees were surprised by how abruptly the decision landed. What this really means is that Sora did not just fail as a product. It failed the new test OpenAI appears to be applying to everything: does this strengthen the company’s next phase, or does it drain power from it.
Science is not dead, but it is being reorganised
It would be too simple to say OpenAI is abandoning science just because Weil is leaving and Prism is being shut down. The picture is more interesting than that. Prism only launched in January as a free AI native workspace for researchers, built around GPT-5.2 and designed to handle scientific writing, collaboration, citations, and LaTeX heavy workflows in one place. Now it is being folded into Codex, while Weil says OpenAI for Science is being decentralised into other research teams. At almost the same moment, OpenAI launched GPT-Rosalind for life sciences research, made it available through ChatGPT, Codex, and the API for qualified users, and positioned it for workflows in biology, drug discovery, and translational medicine. So the shift is not from science to no science. It is from science as its own freestanding workspace and identity, toward science as a domain inside a more unified product and infrastructure strategy.