Three Top OpenAI Executives Exit As Sora And Science Unit Shut Down In Major Company Restructuring

Three senior executives have left OpenAI on the same day as the company dismantles key experimental units and reshapes how it develops and deploys artificial intelligence, signalling a clearer shift towards commercially focused products and enterprise services.

Three Senior Leaders Exit As OpenAI Reshapes Its Structure

Kevin Weil, Bill Peebles and Srinivas Narayanan have all departed OpenAI on the same day as the company dismantles key experimental units and restructures how it develops and deploys artificial intelligence, signalling a clearer shift towards commercially focused products and enterprise services.

Kevin Weil, Bill Peebles and Srinivas Narayanan have all departed OpenAI in a single day of leadership changes that coincided with a wider internal restructuring.

Weil had been leading “OpenAI for Science”, Peebles worked on the company’s short-form video tool Sora, and Narayanan oversaw enterprise technology and business applications.

The exits come as OpenAI reorganises internal teams, merging several standalone projects into core engineering and product groups.

A spokesperson said the company is “decentralizing OpenAI for Science” to bring its work closer to model, product and infrastructure teams.

Why Was Sora Shut Down And What Happened To Openai For Science

Two high-profile initiatives have now been wound down or absorbed into other divisions.

Sora, OpenAI’s text-to-video generation tool, was discontinued after struggling to maintain sustained user growth and facing high operating costs reportedly reaching around $1 million per day at peak usage.

The product, which had briefly surged in popularity after launch, has now been closed, with its API expected to wind down in September.

Bill Peebles, who led the project, said:

“I’m proud of all the sleepless nights before and after the launch this team endured in order to deploy the technology in a responsible way and help steer societal norms.”

Meanwhile, OpenAI for Science, aimed at building AI tools for scientific research in fields such as biology and physics, has been dismantled as a standalone unit.

Its work, including tools like Prism and GPT-Rosalind, will now be integrated into existing research and product teams.

Is Openai Moving Away From Experimental Projects

The restructuring reflects a broader shift away from standalone experimental programmes towards tighter integration with core product development.

Kevin Weil described his time at the company as “a mind-expanding two years”, adding:

“It’s been a mind-expanding two years, from Chief Product Officer to joining the research team and starting OpenAI for Science.”

Internally, several experimental programmes have been merged or discontinued over recent months as OpenAI focuses on aligning research output more directly with product delivery and commercial use cases.

Enterprise Push And Revenue Pressure Drive Strategy Shift

The company is increasingly prioritising enterprise AI and scalable products as it seeks more predictable revenue streams.

Enterprise services now account for more than 40 percent of OpenAI’s income, with expectations that this will continue to rise as adoption grows among businesses.

OpenAI is also balancing rapid revenue expansion with heavy infrastructure costs.

The company is estimated to generate around $2 billion per month, roughly $25 billion annually, but projections suggest losses could reach about $14 billion in 2026 due to compute and development expenses.

Despite this, it continues to scale rapidly, with around 900 million weekly ChatGPT users reported.

Rising Costs And Competition Intensify Pressure On AI Race

The restructuring comes amid intensifying competition from rivals including Anthropic, Google and Meta, all accelerating development in both consumer and enterprise AI.

Industry rivals are also improving efficiency, with some achieving stronger revenue performance at lower operational costs, increasing pressure on OpenAI to streamline its own structure.

At the same time, Meta has been actively recruiting AI researchers, adding further competition for talent across the sector.

Departing Executives Reflect On Internal Changes

Srinivas Narayanan, who helped scale ChatGPT and OpenAI’s API infrastructure, said he was stepping back for personal reasons after contributing to the company’s enterprise expansion.

Bill Peebles noted that Sora helped “spark a huge amount of investment in video across the industry”, while Kevin Weil pointed to the scientific tools developed under his leadership before the unit was folded into broader teams.

The exits also follow recent leadership changes within the company, including temporary departures and role shifts among senior executives, highlighting a period of continued internal transition as OpenAI reworks both its structure and long-term direction.