What Happened
On March 24, 2026, OpenAI announced it is shutting down Sora — its AI video generation app, API, and web experience — just six months after the standalone app launched in September 2025. The Sora web and app experiences will be discontinued on April 26, 2026. The Sora API will be discontinued on September 24, 2026. Users have been told to export their content before the app cutoff.
The shutdown came with a Disney deal implosion, an internal memo confirming compute reallocation to the Spud model, and an org rename that signals what OpenAI now considers its core business. For developers who built on the Sora API: you have until September 24 to migrate.
The Numbers That Killed Sora
The narrative around Sora's shutdown has been framed as a strategic pivot. The actual story is a cost structure that was never viable.
Sora was burning through an estimated $15 million per day in inference costs at its peak. Its entire lifetime revenue from in-app purchases came to $2.1 million. Video generation is compute-intensive by nature — every text-to-video render draws heavily on GPU capacity that OpenAI needed elsewhere.
After a splashy launch, Sora's worldwide user count peaked at around one million and then collapsed to fewer than 500,000. The app was burning through roughly $1 million every day — not because people loved it, but because video generation is so costly to run.
By February 2026, downloads had fallen to just over 1.1 million — a decline of roughly 66% in three months. Monthly active users also peaked in December 2025 and fell through early 2026. The app was not just failing to grow. It was actively shrinking.
The gap between cost and revenue was not closable at consumer pricing levels. Video generation requires orders of magnitude more compute per output than text generation, and users were not willing to pay what it actually cost to produce those outputs.
The Disney Deal: $1 Billion, Zero Execution
The most damaging collateral from the shutdown is the Disney partnership — announced with significant fanfare in December 2025 and dissolved before a single deliverable shipped.
Under a three-year licensing agreement announced in December 2025, Sora would have been able to generate user-prompted videos from a set of more than 200 masked, animated, or creature characters from Disney, Marvel, Pixar, and Star Wars. Disney also planned a $1 billion equity investment in OpenAI as part of the agreement.
The entertainment giant had committed $1 billion to the partnership, yet found out Sora was being shut down less than an hour before the public.
According to reporting, no money ever actually changed hands between Disney and OpenAI as part of this agreement. Disney issued a statement saying it "respects OpenAI's decision to exit the video generation business" — diplomatic language for a billion-dollar deal dissolving before execution.
The irony is notable: Disney had previously filed copyright complaints against other AI video companies for using its IP without permission. It then partnered with OpenAI — and watched the partnership collapse before a single Disney+ Sora video went live.
Where the Compute Goes: Spud and the Superapp
OpenAI's official statement framed the shutdown in terms of focus and compute demand. An OpenAI spokesperson said: "As we focus and compute demand grows, the Sora research team continues to focus on world simulation research to advance robotics that will help people solve real-world, physical tasks."
The direct beneficiary of freed compute is Spud — OpenAI's next frontier model, which completed pretraining on approximately March 24–25, 2026 (the same day as the Sora announcement). Every GPU previously running Sora inference is now available for Spud evaluation and post-training.
The Sora team's reassignment to "world simulation research for robotics" is not a consolation prize — it is a genuine strategic move that puts them in direct competition with Google DeepMind and NVIDIA in physical AI. Whether that research surfaces as a product is a separate question.
While a whole team inside OpenAI was focused on making Sora work, Anthropic was quietly winning over the software engineers and enterprises that drive revenue. Claude Code, in particular, was eating OpenAI's lunch. So CEO Sam Altman made the call: kill Sora, free up compute, and refocus.
The Org Restructuring That Explains Everything
The Sora shutdown is one piece of a broader reorganization that clarifies OpenAI's actual product bets heading into 2026.
"Product" renamed to "AGI Deployment" — Fidji Simo's division, which oversaw Sora, has been renamed. The message: OpenAI's product organization now exists to deploy AGI-level capabilities, not to ship consumer media apps.
Superapp consolidation — OpenAI is building a desktop superapp that merges ChatGPT, the Codex coding agent, and the Atlas browser agent into a single interface. Sora did not fit that architecture. Maintaining a separate video app with its own infrastructure, its own moderation stack, and its own content partnerships was pulling engineering resources in the wrong direction.
IPO pressure — OpenAI crossed $25 billion in annualized revenue in February 2026 and is preparing for a public listing targeting a $1 trillion valuation. Investors want a focused, simple business story backed by predictable enterprise revenue. A growing portfolio of experimental consumer apps with unclear monetization does not help that story.
On March 19, OpenAI was still shipping new editing tools inside Sora. On March 24, it moved to deprecate the API and announced it was "saying goodbye" to the app. That is not the pattern of a product drifting into obscurity. The decision was abrupt and deliberate.
What Sora's Failure Means for AI Video
Sora's shutdown is not just an OpenAI story — it is data about the AI video market as a category.
The product launched in September 2025 as the most-downloaded app in the iOS App Store's Photo and Video category within a day. By early 2026, strong alternatives from Google Veo, Runway, Kling, and others had emerged, matching or surpassing Sora in certain use cases. Rather than dominating the category it helped define, Sora quickly became one option among many.
The underlying problem: AI video does not solve a recurring paid workflow for most users. There really needs to be something that people feel like they're getting meaning out of it for it to stick around. People did not treat an AI video app as a daily-use product the way they use a search engine or messaging app.
For developers building in the AI video space, the lesson is structural: video generation has a fundamental cost-per-output problem that consumer pricing cannot bridge unless inference efficiency improves by an order of magnitude. The companies that survive in this space will either serve professional workflows at professional pricing (Runway's model) or find dramatically cheaper inference paths.
Key Dates and Deadlines
| Event | Date |
|---|---|
| Sora app publicly launched | September 2025 |
| Disney deal announced | December 2025 |
| Sora peak downloads (~3.3M/month) | December 2025 |
| Sora new editing tools shipped | March 19, 2026 |
| Shutdown announced | March 24, 2026 |
| Sora web and app discontinued | April 26, 2026 |
| Sora API discontinued | September 24, 2026 |
| Disney $1B deal dissolved | March 24, 2026 |
What Developers Should Do Now
If you're on the Sora API: You have until September 24, 2026 to migrate. ChatGPT/Sora credits can still be used for Codex if you choose. Evaluate Runway ML, Google Veo 2, and Kling as API-accessible alternatives for video generation workflows.
If you have Sora-generated content: Export it before April 26, 2026. After Sora is discontinued and after the period of any final export window passes, OpenAI will permanently delete any data associated with your use of Sora.
If you're building an AI video product: Sora's numbers should inform your unit economics assumptions. At $15M/day in inference costs against $2.1M in lifetime revenue, the gap between compute cost and willingness to pay for AI video was not marginal — it was existential. Build accordingly.
FAQ
Why did OpenAI shut down Sora?
Three compounding reasons: inference costs estimated at $15 million per day against $2.1 million in lifetime revenue, rapid user count decline from 3.3 million monthly downloads to under 1.1 million by February 2026, and the need to free compute for Spud — OpenAI's next frontier model — and the planned ChatGPT superapp.
What happens to the Disney deal?
Dissolved. Disney had planned a $1 billion equity investment in OpenAI and a three-year character licensing agreement for Sora. No money changed hands. Disney found out about the shutdown less than an hour before the public announcement. Disney issued a statement saying it "respects OpenAI's decision to exit the video generation business."
When does the Sora API shut down?
The Sora web and app experience ends April 26, 2026. The Sora API is discontinued September 24, 2026. Developers have until September 24 to migrate workloads to alternative APIs.
What happens to the Sora team?
OpenAI reassigned the Sora team to world simulation research targeting robotics applications — a direct move into territory occupied by Google DeepMind and NVIDIA's physical AI research groups.
What are the best Sora alternatives in 2026?
For API-accessible video generation: Runway ML Gen-3, Google Veo 2, and Kling (Kuaishou). Each has different pricing and capability trade-offs. Runway is the most developer-friendly for professional workflows. Google Veo 2 is available via the Gemini API for teams already on Google infrastructure.
Next step: If you have content in Sora, export it at sora.com before April 26, 2026 — hover over your media, click the ... menu, and select Download. If you're on the API, review the OpenAI API deprecations page and begin migration testing against Runway or Veo 2 before the September 24 cutoff.
Sources: TechCrunch, March 29, 2026 (https://techcrunch.com/2026/03/29/why-openai-really-shut-down-sora/) · OpenAI Help Center discontinuation FAQ (https://help.openai.com/en/articles/20001152-what-to-know-about-the-sora-discontinuation)