Sora burned roughly $1 million a day in compute costs. Across its operational lifespan, it grossed $2.14 million in total revenue. OpenAI shut it down anyway, and in doing so killed a Disney deal that had been valued at $1 billion.
The numbers underneath that decision deserve more than one read. Sora, OpenAI's text-to-video generator, peaked at around 1 million monthly users in November 2025 before collapsing to fewer than 500,000 by the time it was discontinued on March 29, according to TechCrunch. Monthly downloads fell from 6.1 million that November to 1.1 million in March, Sensor Tower data shows. The product had no plateau. It had a cliff.
The unit economics confirm what the user numbers suggest. Across roughly six months and 11.7 million downloads, Sora grossed $2.14 million in revenue, per Appfigures Intelligence cited by Ars Technica. The app was burning through approximately $1 million every day at its peak, TechCrunch reported. At that rate, OpenAI spent more on compute in a single month than Sora generated in its entire lifetime. OpenAI has not disclosed its internal financials for Sora. The figures come from public reporting and analyst estimates. But the gap between burn and revenue is not close enough to dispute.
Disney was supposed to be the exception. The entertainment giant had negotiated a three-year licensing agreement covering more than 200 Marvel, Pixar, and Star Wars characters, plus a separate equity investment reported at $1 billion, The Guardian reported. The deal lasted three months. Disney learned of the shutdown less than an hour before the public announcement, TechCrunch reported. No money had changed hands, Reuters noted. The partnership existed on paper and collapsed before it fully existed in practice.
The memo that best explains what happened inside OpenAI surfaced from Fidji Simo, the company's VP of Applications. "We cannot miss this moment because we are distracted by side quests," she wrote, per The Verge. "We really have to nail productivity in general and particularly productivity on the business front." The memo did not name Sora. The inference costs named it anyway.
Sam Gregory, a program director at Witness, a nonprofit that works on media authenticity, frames the shutdown differently. "What angers me actually is what they normalized in the period of six months, a set of things that I think have really long-lasting implications," he told The Verge. Sora shipped into a world where hyper-realistic video could be generated and distributed without watermarking infrastructure, industry-wide authenticity standards, or meaningful public understanding of what had changed. The normalization happened in six months. The consequences, Gregory suggested, will outlast the app.
Trevor Harries-Jones, a researcher at the Render Network, told The Verge the competitive position had narrowed fast. "The state of innovation and the plethora of choice means there is little to no moat and it is very simple to switch between," he said. OpenAI showed a striking demo. The actual product arrived late, at high cost, and into a market that had moved.
That market is being absorbed by others. Kling AI, the video generation tool from Chinese short-video company Kuaishou, saw global weekly active users jump 4 percent to 2.6 million in the week following Sora's shutdown, Sensor Tower found. By January 2026, Kling's annualized revenue run rate exceeded $300 million. Runway, a San Francisco-based AI video startup, raised $315 million in a Series E in February 2026, nearly doubling its valuation to $5.3 billion, TechCrunch reported. General Atlantic led the round, with Nvidia, Fidelity, and Adobe Ventures among participants. Runway's Gen 4.5 model had outperformed both Google and OpenAI on several video-generation benchmarks, with native audio, long-form generation, multi-shot consistency, and advanced editing.
OpenAI said it will continue world simulation research. The consumer app and API are gone. The question underneath the decision is one the company has not answered: did it kill Sora because the race was lost, or because the race was never winnable on those economics? The memo suggests the direction was already set. Productivity, not video. The side quest is over.
The side quest may have been the point.
