OpenAI shut down Sora on March 30, 2026. The company’s AI video generation platform lasted roughly three to four months in public availability before operational costs and user retention made continuation untenable, according to TechCrunch and The New York Times.
The numbers tell a clear story. Sora’s user count reportedly peaked at approximately one million and fell to fewer than 500,000, according to TechCrunch citing Wall Street Journal data. The platform reportedly cost approximately $1 million per day to operate, according to Wall Street Journal reporting cited by multiple outlets. That combination, a shrinking user base against a fixed, substantial cost structure, left no viable path forward.
The Disney dimension sharpens the picture. The New York Times reports the shutdown came approximately three months after OpenAI signed a multiyear deal to bring Disney characters to the Sora platform. That deal is now cancelled. This wasn’t just a product that failed to find an audience. It was a product that secured a marquee enterprise partnership and still couldn’t hold its users.
The BBC confirms that the shutdown covered both the consumer app and the professional platform used by creators to generate video, Sora’s entire operational footprint is gone.
Why it matters. Sora was not a peripheral experiment. It was OpenAI’s flagship entry into generative video, backed by the company’s core brand and technical reputation. Its failure signals something the AI industry doesn’t discuss often enough: capability and commercial viability are different problems. A model can produce impressive outputs and still fail to generate sustainable engagement. The ~50% user drop from peak to shutdown is the metric that deserves attention. Half the people who tried Sora decided it wasn’t worth coming back to.
For practitioners evaluating AI product investments, this is the unit economics question in concrete form. What does retention look like after the novelty wears off? At $1 million per day in operational costs, OpenAI needed sustained engagement, not peak curiosity. It didn’t get it.
Context. Reports attribute the shutdown to high operational costs and declining user engagement, not to data collection concerns that had circulated as earlier speculation, per Dataconomy’s reporting. OpenAI made an announcement confirming the shutdown, per NYT, though no official blog post appeared in the available source record. The shutdown was confirmed independently by the BBC and NBC News, among others.
What to watch. Three questions now matter. First, where do Sora’s former users go? Runway, Kling, and PixVerse, which launched its V6 update the same day Sora shut down, are the obvious candidates for migration. Second, does OpenAI pivot to enterprise AI video rather than consumer, or exit the category entirely? Third, does this affect how other AI labs think about the consumer-facing video generation market? One major exit changes the competitive calculus for everyone still in it.
TJS synthesis. Sora’s shutdown is the clearest data point yet that AI product success requires more than technical capability and brand recognition. OpenAI had both. What it didn’t have was a product that users found indispensable enough to return to. For anyone building or funding AI applications, the lesson isn’t that AI video is impossible, it’s that novelty-driven adoption without workflow integration produces exactly this outcome.