Last September, OpenAI launched Sora 2. The app, which could generate realistic videos from simple text descriptions, immediately shot to number one in the Apple App Store. People created absurd clips, such as Princess Diana doing parkour or dogs driving cars. The hype was enormous. And now, six months later, OpenAI has announced the shutdown of Sora.
OpenAI bid farewell to its video app with the words, “To everyone who created with Sora: thank you.” Nice words, but hard business realities lie behind them.
The numbers speak for themselves
Just look at the data. In November 2025, Sora peaked at 3.3 million downloads worldwide. By this February, that figure had fallen to just 1.1 million, a two-thirds decline in only three months. And revenue? Over its entire existence, Sora generated approximately $1.4 million in in-app revenue. During the same period, ChatGPT made more than $1.9 billion. That is a staggering difference that can only be described as a resource black hole with minimal prospects of turning a profit.
At the same time, Sora’s operating costs were enormous. Generating videos consumes computing power on a completely different scale than producing text responses. OpenAI was effectively paying for every generated clip without users covering those costs through subscriptions, for example.
Legal issues with content
But Sora was not just a financial disaster. It also became a moderation nightmare. Soon after launch, violent videos, racist content, celebrity deepfakes, and copyright infringement began proliferating on the platform.
OpenAI fought back. It tightened its terms of use, restricted the generation of content featuring famous faces, and addressed complaints from intellectual property owners. The day before announcing the shutdown, it even published a blog post titled “Creating with Sora Safely” about how it was safely improving the app. A day later, it shut it down. That must have made for uncomfortable reading for those who wrote the blog post.
One of the most high-profile moments of the Sora era came in December 2025. OpenAI and Disney signed a three-year, billion-dollar deal that allowed Sora users to create videos featuring more than 200 licensed Disney characters, including characters from Marvel, Pixar, and Star Wars. Hollywood and Silicon Valley shook hands. The media hailed it as a breakthrough.
It lasted three months. The media later reported that no money had actually changed hands between Disney and OpenAI when the deal was canceled. A Disney spokesperson responded diplomatically: “We respect OpenAI’s decision to exit the video generation space.” But the company was reportedly caught off guard. Hardly surprising.
OpenAI has a new direction
So what now? Since its inception, OpenAI has operated in a way reminiscent of a startup incubator. CEO Sam Altman bet on many different projects at once: the Atlas web browser, hardware devices, robots, Sora, and Codex. Now it is time to sharpen the focus. The company is preparing to go public. CFO Sarah Friar explicitly said that OpenAI must be “ready to be a public company.” And that means presenting investors with a clear story, not a fragmented product portfolio.
The result is a plan for a so-called super app. ChatGPT, the Codex coding agent, and the Atlas browser are to be merged into a single interface. Codex is also having an excellent year, with more than $1 billion in annualized revenue since January 2026, and is becoming the natural centerpiece of a strategy focused on enterprise customers.
The computing resources consumed by Sora will now be redirected to robotics research. OpenAI wants to use the technology that taught AI to create realistic videos to train physical robots. It sounds like a major pivot. But with this company, we have grown accustomed to surprises coming quickly.
After all, who would have thought in September 2025 that the number-one app in the App Store would be history six months later?
Sources: theguardian.com and wired.com
