OpenAI has Shut Down Sora and a Billion-Dollar Disney Deal

OpenAI is shutting down Sora. The app, the API, and any plans for video inside ChatGPT are all going away. Disney’s $1 billion investment and character licensing deal, announced just three months ago, is done too.

The stated reason is focus. Fidji Simo, OpenAI’s applications CEO, told employees the company “cannot miss this moment because we are distracted by side quests.” Sora was classified as one of those side quests.

It’s a surprising label for a product that hit #1 on the App Store and reached a million downloads in five days. But the usage data makes the decision more understandable. Sora had 9.6 million total downloads and generated about $2.1 million in consumer spending across its entire life. Downloads dropped 32% in December and another 45% in January, and the app fell out of the top 150 within months of launch. People tried it, made a few clips, shared them, and stopped coming back.

Compare that to Codex, OpenAI’s coding product. Over 2 million weekly active users, 3x user growth, 5x usage growth since January. Last week OpenAI acquired Astral, the team behind uv, Ruff, and ty (Python’s most popular developer tools), and folded them into Codex. They’re also building a desktop app that merges ChatGPT, Codex, and their Atlas browser into one product.

When your projected negative cash flow through 2029 is $143 billion (per Deutsche Bank estimates) and you’re preparing for an IPO, you keep the product with better retention and unit economics. Codex is that product.

Developer Tools as the Core Business

I think this was the right call. Consumer video generation is expensive to run, hard to monetize at scale, and the usage patterns aren’t sticky. Developer tools have better economics across the board: predictable usage, high retention, and real willingness to pay.

OpenAI isn’t the only company making this calculation. Anthropic is investing heavily in coding agents. Google is expanding Gemini’s developer tools. The whole industry is converging on developer productivity as its primary product, not a side feature.

If you build software with AI tools, that convergence matters for you directly. The capabilities you use every day, coding assistance and API inference, are the ones these companies have the strongest business reason to keep running and improving. They’ll keep getting compute allocated to them because they’re the ones generating revenue.

Platform Risk and the Disney Situation

The other angle worth considering is platform risk.

Disney committed $1 billion and licensed over 200 characters from Disney, Marvel, Pixar, and Star Wars for use in Sora. Three months later, that partnership is gone because OpenAI decided the compute was better allocated to other products.

If you’re building on AI platforms, the practical lesson is simple: understand what the platform’s core business is, and build your critical dependencies on the parts that generate revenue for them. For OpenAI, that’s coding tools and enterprise. Those are the capabilities most likely to survive priority shifts, because they’re the revenue engine.

Features that sit outside the core business, consumer experiments, entertainment integrations, viral novelties, carry real risk. The fact that Sora had a billion-dollar partner and millions of users didn’t save it once compute allocation became the deciding factor.

For most developers, the alignment already works in your favor. Your daily workflow runs on the same tools these companies are betting their IPOs on. This isn’t a coincidence, and it’s the most reliable kind of platform stability there is.