OpenAI officially shuts down Sora: the video generation app didn't pay for itself

In a single Tuesday, OpenAI made a series of decisions that completely reshaped its video product strategy: it shut down the Sora app, canceled plans for video generation inside ChatGPT, scrapped its $1 billion deal with Disney and announced an additional $10 billion from investors.

Sora debuted in December 2024 to applause from the press and users. The model could create realistic videos from text descriptions and was positioned as the flagship of a new line of business. But within a few months it became a symbol of another OpenAI problem: a huge gap between infrastructure costs and actual revenue.

Generating video takes incomparably more computing resources than working with text or even images. Each Sora request cost the company several times more than a ChatGPT request, and the company failed to monetize that volume of traffic at profitable rates. According to The Verge's sources, Sora consumed vast amounts of compute without a corresponding financial return, and that was the main reason for the shutdown.

Dropping the Disney deal sends a separate signal. The $1 billion partnership envisioned using OpenAI's AI tools in content production at the world's largest media corporation. Pulling out of such an agreement means either failing on the technical side or admitting that the terms were unprofitable given the current cost structure.

The same day, the company announced a new funding round: another $10 billion from investors brings the total of the latest round to $120 billion. It is the largest venture raise in the history of the technology industry, and also an indirect admission that OpenAI still has no clear path to profitability.

Competitors in video, including Runway, Kling, Vidu and Pika, are carrying on and actively attracting users. OpenAI's exit from this segment frees up a niche and will likely speed up consolidation around more specialized players. For OpenAI itself, this is a turn toward focus: fewer experimental directions, more concentration on profitable products, above all ChatGPT and the API for enterprise customers.

Shutting down Sora is not a failure of the technology. It is an acknowledgment that even an impressive product does not survive if the unit economics don't add up. Under pressure from investors and the need to prove the business is viable, OpenAI made a pragmatic choice: better to shut down a money-losing product now than to keep subsidizing it at the expense of everything else.