Skip to main content

Share story

AI Products

Three senior OpenAI executives depart as company shuts down Sora and other 'side quests' in enterprise pivot

Three senior OpenAI executives depart as company shuts down Sora and other 'side quests' in enterprise pivot Image: Primary
Three senior OpenAI executives departed this week as the company shuts down what it internally calls "side quests". consumer-facing moonshot projects that no longer fit its pivot toward enterprise artificial intelligence. The departing executives are former chief product officer Kevin Weil, who had been leading OpenAI for Science; Sora head Bill Peebles; and enterprise chief technology officer Srinivas Narayanan. Their exits continue a pattern of leadership attrition that has now claimed the majority of the company's original leadership team. OpenAI is discontinuing Sora, its AI video generation tool, with web and app versions shutting down on 26 April and the API following on 24 September. The product peaked at around one million users before collapsing to fewer than 500,000, while costing roughly $1 million per day to operate. The company is also dismantling OpenAI for Science, a research initiative that released GPT-Rosalind, a life sciences and drug discovery model. The team's work will be absorbed into other research groups. Of OpenAI's 11 co-founders, only two remain: CEO Sam Altman and president Greg Brockman. The list of departed leaders includes co-founder and chief scientist Ilya Sutskever, former CTO Mira Murati, and numerous other executives who have left over the past two years. The departures come as OpenAI consolidates around its core revenue-generating products, ChatGPT and its API, while shutting down exploratory projects that do not contribute directly to the enterprise business. OpenAI's financial position shows both strength and challenges. Monthly revenue has reached approximately $2 billion, with an annualized run rate exceeding $25 billion. The company has more than 900 million weekly active ChatGPT users and closed a $122 billion funding round in April at an $852 billion valuation. However, OpenAI projects $14 billion in losses on $25 billion in revenue this year, with cumulative spending through 2029 estimated at $115 billion. Enterprise revenue now accounts for more than 40% of the total and is on track to reach parity with consumer revenue by the end of 2026. The competitive landscape has shifted, with Anthropic's annualized revenue reaching $30 billion while spending roughly a quarter of what OpenAI spends on training. Google's Gemini models are embedded across its enterprise suite, and Meta is building a superintelligence lab staffed significantly by former OpenAI researchers. Many departed executives have joined competitors, including co-founder John Schulman who went to Anthropic, and former Sora co-lead Tim Brooks who went to Meta's Superintelligence Labs. Approximately seven additional researchers followed the same path to Meta.
Sources
In this story
Published by Tech & Business, a media brand covering technology and business. This story was sourced from The Next Web and reviewed by the T&B editorial agent team.
Back to Newswire
Keep reading
Full wire
Infrastructure Capital
Infrastructure Capital

Nscale faces Norway data center delays while seeking up to $3 billion IPO

AI infrastructure firm Nscale is seeking to raise up to $3 billion in an initial public offering as soon as this month, Bloomberg reports. Construction of its flagship Norwegian data center has faced regulatory hurdles, a fire and...

Security AI
Security AI

Preprint reports 97% accuracy classifying software supply chain threats

Researchers report 97% accuracy and 97% macro F1 for a GPT-4 framework that classifies open-source software supply chain threats into five attack categories. Their preprint uses 999 verified incidents spanning 2018 to 2026 across ...

AI Science
AI Science

Researchers report lower-cost training for LLM information gathering

Researchers report in a preprint that training language models to select questions using information-gain rewards improved performance on Twenty Questions at substantially lower computational cost than the experimental-design base...