Sora Shut Down After 4 Months: The Collapse of OpenAI's Video Generator and the $1 Billion Disney Deal

Sora, OpenAI's video generator, was announced in partnership with Disney in December 2025, in a deal valued at $1 billion. It was officially shut down on March 24, 2026, a little more than four months after the announcement. Habr reported this, citing the official Sora account. According to the outlet, billions of dollars were spent on the project, while the revenue it brought in was close to zero.

Timeline of Sora's rise and fall

Sora's story played out in a matter of months, from the loud announcement of a partnership with one of the world's largest media companies to the quiet closure of the service. Habr describes it not as an ordinary failure of yet another AI startup, but as an event that could have serious consequences for OpenAI CEO Sam Altman, for the company itself and for how the AI industry is perceived as a whole, amid already growing talk that the market for AI investment is overheated.

Key facts about the project:

• OpenAI announced Sora in partnership with Disney in December 2025.
• The Disney deal was valued at $1 billion.
• The service was officially shut down on March 24, 2026.
• The project lasted about four months from its announcement.
• Billions of dollars were spent on development and launch, with almost no revenue.

Why the project closed so quickly

AI video generation is among the most compute-hungry tasks in modern machine learning. Unlike text models, each generated video requires far more computation at output, which makes such services expensive to run at any significant scale of use. When infrastructure costs grow faster than revenue from subscriptions or content licensing, a company has few options: sharply raise prices and narrow the audience, or wind the product down before losses spiral out of control. Judging by Habr's account, this is exactly the scenario that played out with Sora. Billions in spending did not convert into comparable revenue, and the Disney partnership, despite the prestige of the deal, did not change that economics.

What this means for OpenAI and the industry

Shutting down a project of this scale, with a media partner like Disney, is a notable reputational blow. Deals at this level are usually announced as strategic, long-term partnerships, not short-lived experiments. Habr states directly that the consequences may reach not only OpenAI but also Sam Altman personally as the company's public face, as well as perceptions of the AI bubble in general.

The fast, costly failure of one of the most high-profile product announcements of recent years gives skeptics a convenient argument. They have long pointed to the gap between colossal investment in generative AI and the actual ability of such products to turn a profit. For the other players in video generation, Sora's story is a warning that even a prestigious partnership with a major media brand does not guarantee a project's economic viability if the cost of compute does not match users' willingness to pay.

In recent years, AI video generation has become one of the most competitive areas of generative AI. Besides OpenAI's Sora, other major industry players are developing their own models that create video from text or visual prompts. Against this backdrop, the quick shutdown of OpenAI's project, backed by a partnership with a major content rights holder, raises a question for the whole industry: does the same scenario apply to competitors? If even the combination of a recognizable brand, access to premium content and multibillion-dollar funding did not give Sora a sustainable business model, the problem lies deeper, in the fundamental economics of AI video generation, rather than in the management decisions of a single company.

It is also telling how fast the whole story unfolded: from the December 2025 announcement of the Disney partnership, presented as a landmark event for the entertainment industry, to the quiet wind-down of the service less than half a year later. Such a rapid change of course is unusual even by the standards of the fast-moving generative AI market. It suggests that actual usage and monetization diverged from the expectations built into the deal much faster than usually happens with large corporate partnerships of this kind.