Industry

Kevin Weil, Bill Peebles Leave OpenAI Amid Side Cuts

OpenAI sees departures of Kevin Weil, head of science research, and Bill Peebles, creator of AI video tool Sora. The moves align with the company's pullback from side projects like Sora, which shut down last month after high compute costs, and OpenAI for Science, now merging into other teams. Srinivas Narayanan, CTO for enterprise apps, also plans to leave for family time.

Neura News

Neura News

Neura Market Editorial

April 17, 20263 min read
Kevin Weil, Bill Peebles Leave OpenAI Amid Side Cuts

Kevin Weil, Bill Peebles Leave OpenAI Amid Side Cuts

OpenAI faces the loss of two key figures behind its bold projects. Kevin Weil oversaw the science research effort. Bill Peebles developed the AI video system Sora. Both shared news of their exits on Friday. These changes happen while OpenAI focuses more on enterprise AI and plans for a superapp.

Pullback from Ambitious Side Efforts

The company has trimmed what it calls side quests. These include projects aimed at customers, such as Sora and OpenAI for Science. Sora stopped operations last month. It faced compute expenses of about $1 million each day. OpenAI for Science handled internal work on Prism. That tool uses AI to speed up scientific breakthroughs. Now it moves to other research groups, as Weil noted in his social media update.

OpenAI started as a research lab in 2015 with a mission to build safe artificial general intelligence. Over time, it launched products like ChatGPT, which brought massive attention to generative AI. The firm shifted partly to a for-profit model to fund its goals. Recent steps show a turn toward business applications and core products. Side efforts like video generation and specialized science tools take a backseat.

Challenges for OpenAI for Science

The group faced a brief and rough path. It launched formally in October 2025. Weil once posted about GPT-5 tackling 10 unsolved Erdős math problems. He removed that message quickly. A mathematician from erdosproblems.com pointed out issues with the claim. Just one day before Weil left, his team unveiled GPT-Rosalind. This model aims to boost life sciences research and drug development.

Prism represented hopes for AI in science. The platform sought to aid discoveries across fields. Weil's role evolved from chief product officer to research leadership. OpenAI for Science formed under his guidance. Now, its work integrates elsewhere in the organization.

The #1 Newsletter in AI

Stay ahead of the AI curve

The most important updates, news, and content — delivered weekly.

No spam. Unsubscribe anytime.

Insights from Leaving Researchers

Weil reflected on his time there. "It's been a mind-expanding two years, from Chief Product Officer to joining the research team and starting OpenAI for Science," he wrote. He added, "Accelerating science will be one of the most stunningly positive outcomes of our push to AGI."

Peebles highlighted Sora's effects. He said it sparked major funding for video AI industry-wide. He believes such work needs distance from main priorities. "Cultivating entropy is the only way for a research lab to thrive long-term," he stated.

Sora debuted as a text-to-video generator. It produced realistic clips from descriptions. The tool drew praise but also concerns over costs and use cases. Peebles' contributions helped push boundaries in multimodal AI.

Additional Leadership Change

OpenAI also parts with Srinivas Narayanan. He served as chief technology officer for enterprise applications. Wired reports he shared his plans inside the company. Narayanan aims to spend more time with family.

These exits point to a refocus. OpenAI builds on successes in language models and chat interfaces. Enterprise tools gain priority. The superapp hints at broader consumer reach ahead.

Related on Neura Market

More from Neura News

General

Open-weight AI mirrors Kubernetes ecosystem shift

Tobi Knaup, co-founder of Mesosphere, draws parallels between the rise of Kubernetes and the current trajectory of open-weight AI models. He argues that open-weight models are becoming a neutral substrate for innovation, attracting a global ecosystem of developers, startups, and enterprises. The piece warns against US restrictions on Chinese open-weight models, advocating instead for American leadership through open releases, procurement strategies, and standards.

Jul 25·7 min read
General

Open-weight AI mirrors Kubernetes rise, US warned on bans

The author, a Mesosphere co-founder, draws parallels between the rise of Kubernetes and the current open-weight AI ecosystem. He argues that open-weight models are becoming a neutral platform for innovation, and warns that US restrictions on Chinese open-weight models could isolate American developers from a global ecosystem. The piece urges the US to compete by releasing frontier models, using procurement to create demand, building the stack, and setting standards rather than imposing bans.

Jul 25·7 min read
Industry

Power line failure reveals AI data center grid risks and solutions

A fallen power line near Washington, DC caused over 3 gigawatts of data center load to vanish from the PJM grid in seconds, spiking voltage across the region. The event, which made lights flicker from Northern Virginia to Chicago, highlights a growing problem as AI data centers become larger and more concentrated. Experts warn that without better coordination or technology like ON.Energy's battery-backed uninterruptible power supply, such disruptions will become more frequent and severe.

Jul 25·5 min read