Sora launched to worldwide attention in early 2024. Two years later, it's being shut down.
The move is part of a wider reset at OpenAI: Kevin Weil and Bill Peebles are both departing, the company's internal science team is being folded into other units, and OpenAI is actively shedding what a TechCrunch report describes as "side quests." These aren't isolated personnel moves - they represent a deliberate narrowing of scope.
Sora found a limited paying audience despite genuine technical achievement. The science team focused on longer-horizon research that doesn't translate easily into near-term revenue. For ChatGPT and the broader OpenAI business to justify their current valuation - which requires substantial recurring revenue - the company needs teams and products that convert directly into paying customers, not well-regarded demos.
This kind of reset is standard for any company that crosses from ambitious startup to revenue-obligated business. The harder question is whether OpenAI can actually win in enterprise, where Microsoft, Google, and dozens of specialized vendors already have years of established procurement relationships, compliance certifications, and dedicated sales teams who know how those deals get done.
Safety researchers, policy leads, and now product and science leadership - the pace of departures has been fast. A more commercially focused OpenAI will likely be more financially stable. It will also be less likely to take the kinds of long-shot bets that produced ChatGPT in the first place.