OpenAI shuts down Sora just months after signing Disney deal, pivoting video compute to robotics
The surprise closure of OpenAI's consumer-facing video generator underscores the unsustainable cost of running generative video at scale while the company prepares for a potential IPO.
OpenAI announced on March 24 that it is discontinuing Sora, its AI video generator, just three months after entering a multiyear licensing agreement with Disney that would have allowed Sora users to create videos featuring characters like Mickey Mouse, Cinderella, and Yoda, the New York Times reported. The closure covers both the consumer Sora application and the online service used by filmmakers and production companies to generate video for films and television.
OpenAI did not provide specific reasons for the decision in its public statement, saying only "We're bidding farewell to Sora" and thanking users for their engagement. The company has indicated it will continue to use video-generation technologies internally for training robots, treating video as a simulation of the physical world that can teach robotic systems to understand object dynamics and spatial reasoning.
The timing suggests that the economics of running a consumer video service proved difficult to sustain. Operating a video-generation platform requires substantially more computing power and energy than text-based AI services, and OpenAI anticipates spending approximately $100 billion on infrastructure over the next four years as it prepares for a potential IPO.
Shutting down Sora redirects significant compute resources toward the company's core language and reasoning model development.
Disney, which had framed the partnership as a milestone for both the entertainment and technology industries, said it respects OpenAI's decision and will continue seeking partnerships with AI companies willing to license intellectual property responsibly. The closure is a setback for the broader argument that generative video can be a commercially viable consumer product in the near term, and is likely to sharpen investor focus on the cost structures of AI companies ahead of any public listings.
Read the original reporting at The New York Times.
