What Sora’s Rise And Sudden Fall Means For OpenAI, Disney And AI Video - Forbes
OpenAI killed Sora on March 24, 2026 — twenty-four hours after publishing a blog post titled Creating with Sora safely, detailing watermarking standards, C2PA metadata requirements, teen access controls, and consent-based likeness protections. That sequencing is the lede, not the footnote. Disney and OpenAI had been in an active project meeting that same evening, March 23. Thirty minutes after it ended, Disney got a call: Sora was gone. Not sunset. Not pivoted. Discontinued entirely. One person familiar with the matter called it a big rug-pull. Some OpenAI staffers on the Sora team were reportedly surprised when informed the following morning, a day after the safety post had gone live.
The obvious question the timing raises — were those guardrails a launch requirement, or were they a regulatory pre-statement made after the decision to kill the product had already been made? — is one OpenAI has not answered.
Sam Altman told staff the Sora team would pivot toward robotics and longer-horizon research. The commercial framing has been made explicit: OpenAI is now openly positioning to catch Anthropic in the enterprise and coding market, where Anthropic's Claude Code product holds a meaningful lead. Fidji Simo, previously CEO of Applications, is now CEO of AGI Deployment — a title shift that says everything about where product priority lives. Separately, Altman said OpenAI security and safety teams would no longer report directly to him. The company is simultaneously narrowing its commercial focus and restructuring internal accountability while courting private equity ahead of a possible IPO.
OpenAI is in advanced talks with TPG, Advent International, Bain Capital, and Brookfield Asset Management about a joint venture that would distribute enterprise products across the firms portfolio companies, according to Reuters. The PE pitch includes pre-release model access and financial terms described as well above market, with a 17.5 percent guaranteed return to lock in distribution relationships before going public — a structure that absorbs upfront deployment costs and provides cleaner segment reporting for the IPO narrative. That is a company building its institutional anchor tenants before it has to answer to public markets.
Disney is now looking for a new AI video partner. Google, whose Veo model runs on Vertex AI and already powers a range of content creation tools, is the structural beneficiary — and not only because it was still standing when OpenAI stepped away. Google is the only AI video player at scale without major IP licensing exposure. Every other competitor in this space — Runway, Pika, Lumiere — has had to negotiate content rights, talent agreements, or studio liability frameworks. Google sidestepped that. For a company like Disney that operates at the intersection of intellectual property and content distribution, that distinction is not minor.
The Disney deal never closed. Two people familiar with the matter told Reuters no money changed hands despite the December 2025 announcement. The billion-dollar figure is real as a stated intent; it is not real as a contracted obligation. That gap — between headline investment number and actual committed capital — is something I flagged in a story about AI funding announcements last week, and the Sora episode is a textbook example of why that decomposition matters. The number was always a signal, not a balance sheet entry.