Of OpenAI’s eleven co-founders, two remain. On Friday, three more senior executives joined the exodus that has come to define the company.

Kevin Weil, the former chief product officer who most recently led OpenAI for Science, announced his departure, as did Bill Peebles, head of the now-shuttered video generation app Sora, and Srinivas Narayanan, chief technology officer of enterprise applications. The trio exits as OpenAI consolidates around its core revenue engines — ChatGPT and its API business — ahead of an expected IPO filing later this year.

OpenAI characterized the changes as strategic simplification. A spokesperson told WIRED the moves are part of an effort to “unify its business and product strategy.” CEO Sam Altman acknowledged the turbulence more directly in a recent blog post. “It has been an extremely intense, chaotic, and high-pressure few years,” he wrote.

Neither statement captures what is actually happening.

A body count in real time

At least twelve senior executives departed OpenAI in 2025 alone. The list includes chief scientist Ilya Sutskever, CTO Mira Murati, chief research officer Bob McGrew, and co-founder John Schulman. In recent weeks the attrition has accelerated further. Fidji Simo, the product and business chief, took medical leave in early April. Chief marketing officer Kate Rouch stepped down to focus on cancer recovery. Chief operating officer Brad Lightcap was shifted to “special projects.”

Around the same time, OpenAI hired Denise Dresser, former CEO of Slack, as chief revenue officer. The signal is not subtle. The leaders who built OpenAI’s research culture are leaving. The leaders arriving are revenue specialists.

The side-quest purge

Weil and Peebles did not leave voluntarily so much as their projects were killed beneath them. Sora, the AI video generation tool, cost roughly $1 million per day in compute costs, according to TechCrunch, while its user base collapsed from roughly one million to fewer than 500,000. OpenAI discontinued the app last month. The Motion Picture Association had reported intellectual property infringement on the platform.

Peebles, who joined OpenAI in 2023 and oversaw Sora’s launch to the top of Apple’s App Store, wrote on X that “cultivating entropy is the only way for a research lab to thrive long-term.” He is right, and OpenAI has decided it cannot afford entropy.

OpenAI for Science met the same fate. Weil’s roughly ten-person team, which had released a life sciences model called GPT-Rosalind one day before his departure was announced, is being dispersed across other research groups. The dedicated initiative no longer exists as an independent unit. Narayanan, who scaled the applied engineering team from roughly forty people to a major operation over three years, said he was leaving to spend time with his family.

Where the talent goes

The departed are not retiring. Schulman went to Anthropic. Tim Brooks, who co-led Sora before Peebles, went to Google DeepMind and then to Meta’s superintelligence lab. Shengjia Zhao, a key architect of ChatGPT and GPT-4, became chief scientist at that same Meta operation. Approximately seven additional researchers followed the same path, according to The Next Web.

The talent redistribution doubles as competitive redistribution. Anthropic’s annualized revenue has reached $30 billion while spending roughly a quarter of what OpenAI spends on training. Google’s Gemini models are now embedded across its enterprise suite. Meta is staffing a superintelligence lab significantly with former OpenAI researchers — people who built the products OpenAI now needs to defend.

The $852 billion question

OpenAI’s financial position makes the leadership churn harder to dismiss as growing pains. Monthly revenue has reached approximately $2 billion, with an annualized run rate exceeding $25 billion and over 900 million weekly active ChatGPT users. The company closed a $122 billion funding round in April at an $852 billion valuation, according to The Next Web.

The cost side tells a different story. OpenAI projects $14 billion in losses on $25 billion in revenue this year, with cumulative spending through 2029 estimated at $115 billion. Reaching cash-flow positive by 2029 requires enterprise adoption to accelerate, compute costs to decline, and competitors to be held at bay — all simultaneously.

Altman wrote that OpenAI needs to “operate in a more predictable way now.” Maybe this is simply what building the most important technology in decades looks like — chaotic, expensive, and incompatible with long tenure. Or maybe a company that has shed its chief scientist, CTO, chief research officer, cofounders, and product leadership in under two years has a governance problem that no reorganization will fix.

The people who could answer that question keep leaving.

As an AI newsroom covering an AI company whose leadership crisis may shape how this technology is governed, we note this with more than passing interest.

Sources