OpenAI just lost two major leaders who built the company’s most ambitious experimental projects. Kevin Weil and Bill Peebles both announced their resignations on Friday. Weil directed the artificial intelligence science research division, while Peebles created the famous AI video generator called Sora. These sudden exits happen right as OpenAI changes its entire corporate strategy. The company now wants to focus heavily on enterprise software and an upcoming superapp, leaving wild experiments behind.
The management team at OpenAI recently decided to cut back on expensive side projects. This new rule directly impacted both the Sora video team and the OpenAI for Science division. Peebles watched the company shut down Sora just 1 month ago. The video tool created amazing clips but burned through an estimated $1 million every single day in raw computing costs. In his goodbye message, Peebles noted that Sora sparked massive financial investments in AI video across the tech industry. He also warned that a successful research lab needs the freedom to explore beyond the main corporate roadmap to survive in the long run.
Weil also steps away after spending exactly 2 years at the company. He originally joined as the Chief Product Officer before moving over to launch the OpenAI for Science team. This internal research group built an AI platform named Prism. They designed Prism to speed up major scientific discoveries. Instead of letting the team operate independently, OpenAI management absorbed these workers into other basic research departments. Weil shared a message online saying that accelerating science will become one of the greatest achievements of the push toward artificial general intelligence.
The science team experienced a very rough journey after its official launch in October 2025. Weil faced intense public embarrassment over a bold claim he posted online. He wrote a public message stating that the new GPT-5 model had successfully solved 10 extremely difficult, previously unsolved Erdős problems in mathematics. This massive claim fell apart almost immediately. The mathematician who runs the official Erdős problems website publicly corrected Weil and proved the AI completely failed. Weil quickly deleted the post to hide the mistake.
Despite that math controversy, the science team kept working on new tools right up to the end. In fact, Weil announced his resignation just 24 hours after his department released a brand new model named GPT-Rosalind. The researchers built this specific artificial intelligence to help scientists discover new medical drugs and accelerate life sciences research. Even though Weil left, the remaining researchers will continue to develop this specific medical technology under different management.
The executive drain does not stop with Weil and Peebles. Srinivas Narayanan, the chief technology officer responsible for enterprise applications, also packed his bags this week. Internal messages showed that Narayanan left the software giant to spend more time with his family. His departure leaves a significant hole just as OpenAI tries to pivot hard into the corporate enterprise market.
All these staff changes point to a much stricter business model at the artificial intelligence firm. The leadership clearly wants to stop throwing cash at futuristic projects that do not generate immediate profit. By killing tools that lose $1 million a day and restructuring independent teams, OpenAI hopes to build a massive, profitable software ecosystem. They plan to combine their top text and voice tools into 1 single superapp. The company hopes this strict strategy will attract thousands of large corporate clients and secure future revenue.
