TL;DR

OpenAI killed Sora on March 24, 2026 — app, API, and website, all gone. The move blindsided Disney, whose $1 billion partnership deal collapsed the same day. Downloads had plummeted 67% from their peak, in-app revenue barely hit $2.1 million, and OpenAI is burning $4 billion a month with an IPO on the horizon. This is what happens when an AI company tries to be everything at once and reality catches up.

Six Months from #1 to Dead

Sora launched in September 2025 with a second-generation model that produced eerily realistic videos with audio and accurate physics. It hit #1 in the App Store’s Photo & Video category within 24 hours. Downloads blew past ChatGPT’s early pace. The hype was real.

By February 2026, monthly downloads had cratered from 3.3 million to 1.1 million. Total in-app purchase revenue across the app’s entire life: $2.1 million. For a company burning through $4 billion every month, that’s a rounding error.

OpenAI posted a goodbye on X on Tuesday: “We’re saying goodbye to Sora. To everyone who created with Sora, shared it, and built community around it: thank you.”

No real explanation. No warning. Just done.

Disney Got Blindsided

This is the part that stings the most.

In December 2025, Disney struck a deal with OpenAI worth $1 billion. The plan: license over 200 iconic characters — Marvel, Star Wars, Pixar — for AI-generated short videos. It was the biggest entertainment-AI partnership anyone had seen.

On Monday evening, March 24, Disney and OpenAI teams were in a working session on a Sora-linked project. Thirty minutes after that meeting ended, the Disney team learned OpenAI was killing the product entirely.

“It was a big rug-pull,” a person familiar with the situation told The Hollywood Reporter.

No money ever changed hands. Disney’s official response was diplomatic: “As the nascent AI field advances rapidly, we respect OpenAI’s decision to exit the video generation business and to shift its priorities elsewhere.”

Translation: we got burned, but we’re not making a scene.

Why OpenAI Pulled the Plug

Four factors converged:

1. The math didn’t work. Video generation eats GPU cycles like nothing else. Every Sora video consumed compute resources that could power far more profitable ChatGPT requests. With 900 million weekly ChatGPT users and 50 million paying subscribers, the opportunity cost was massive.

2. Usage fell off a cliff. The novelty wore off fast. After the initial spike, users discovered that AI video generation is fun to demo but hard to build a workflow around. Monthly downloads dropped 67% in three months.

3. Deepfakes became a liability. Despite guardrails, users quickly found ways to generate videos of real public figures. Deepfakes of Martin Luther King Jr. and Robin Williams surfaced, prompting their daughters to publicly ask people to stop. OpenAI’s content moderation team was playing whack-a-mole — every new filter got bypassed within days. For a company heading toward an IPO, that kind of press is toxic.

4. The IPO clock is ticking. OpenAI is targeting a Q4 2026 public listing at a valuation north of $840 billion. Wall Street analysts already flag OpenAI’s $14 billion projected 2026 loss and the fact that profitability isn’t expected until 2030. Killing a money-losing product with mounting PR problems is IPO hygiene 101.

The Anthropic Factor

There’s a competitive angle here that’s hard to ignore.

Anthropic — OpenAI’s most direct rival — has deliberately avoided image and video generation. While OpenAI spread resources across ChatGPT, DALL-E, Sora, and a shopping feature (also recently killed), Anthropic concentrated everything on text and code.

The result: Anthropic’s annualized revenue hit $19 billion by March 2026, and Claude has become the default AI assistant for a growing number of businesses and developers. OpenAI still leads with $25 billion in annualized revenue, but the gap is narrowing fast.

OpenAI CEO Sam Altman reportedly stepped back from overseeing safety and security teams to focus on fundraising, supply chains, and data center construction. The company’s new model, internally codenamed “Spud,” is in development. The signal is clear: OpenAI is trimming everything that doesn’t serve its core business.

Sora was the most visible casualty.

What Happens to Existing Users

If you’ve built anything on Sora, here’s what we know:

  • The iOS app, Sora.com, and the API will all shut down. No exact date yet.
  • OpenAI says it’s “exploring ways to support export and preservation” of user content.
  • Users will get a timeline for saving their generated videos.
  • The Sora 2 model will remain available behind the ChatGPT paywall briefly, but the full creative app experience is ending.

Creators who invested months building workflows around the platform are left scrambling. Some had built entire content pipelines around the Sora API — automated social media clips, product demos, storyboarding tools. All of that infrastructure is now worthless.

The AI video generation space remains volatile. Runway, Pika, Kling, and Google’s Veo all exist, but none have matched Sora’s brand recognition or distribution. And after watching OpenAI pull the plug this fast, creators will think twice before going all-in on any single platform.

What This Means for AI Video

Sora’s death doesn’t mean AI video generation is dead. It means it’s not profitable yet.

The technology works. The demand exists for specific use cases — marketing clips, prototyping, storyboarding. But the gap between “cool demo” and “sustainable business” remains enormous. GPU costs for video generation dwarf those for text, and users aren’t willing to pay enough to close that gap.

For developers building on AI video APIs: treat every platform as potentially temporary. Export your data regularly. Don’t build critical workflows on top of experimental products from companies chasing quarterly metrics. And keep abstraction layers between your application logic and any single provider’s API — the switching cost from Sora to Runway or Pika should be hours, not weeks.

The real question is whether any standalone AI video company can make the economics work, or whether video generation will end up bundled inside larger platforms — a feature of ChatGPT or Gemini rather than a product in its own right.

The Bigger Picture

OpenAI in March 2026 looks very different from OpenAI in March 2025. A year ago, the strategy was expansion — more modalities, more products, more partnerships. Now it’s contraction. Sora gone. Shopping feature gone. Sam Altman focused on infrastructure instead of research.

The company raised $110 billion in February but only $25 billion of that is confirmed immediate cash. Amazon’s $35 billion is milestone-dependent. SoftBank’s $20 billion arrives in quarterly installments. Nvidia’s $30 billion is mostly GPU compute credits, not cash.

At $4 billion per month burn rate, even OpenAI can’t afford to fund experiments that don’t pay for themselves. The era of “launch everything and see what sticks” appears to be ending at the company that popularized it. Meanwhile, competitors are doubling down on securing the compute supply chain — Musk is betting $25 billion on building his own chip fab, and SpaceX just filed for the largest IPO in history partly to fund that infrastructure.

FAQ

Is Sora completely shutting down?

Yes. The iOS app, the website (Sora.com), and the API are all being discontinued. OpenAI’s Sora research team will continue working on “world simulation research” for robotics applications, but the consumer-facing video generation product is finished.

What happened to the Disney deal?

It’s dead. Disney had agreed to invest $1 billion in OpenAI and license over 200 characters for Sora videos. No money changed hands before the shutdown, and Disney has confirmed the partnership is over.

Can I still use AI video generation?

Yes. Runway, Pika, Kling, and others offer AI video generation tools. Google’s Veo is also available through select channels. But none have matched Sora’s peak quality, and all face similar compute cost challenges.

When is OpenAI’s IPO?

OpenAI is reportedly targeting Q4 2026 or early 2027 for its public listing. The company is valued at approximately $840 billion based on its most recent funding round and is aiming for a $1 trillion IPO valuation.

Why did OpenAI kill Sora instead of raising prices?

The core problem wasn’t pricing — it was engagement. Downloads dropped 67% in three months, and total in-app revenue was $2.1 million. Even at higher prices, the user base was shrinking too fast to justify the compute costs, especially with deepfake controversies adding legal and reputational risk ahead of an IPO.

Bottom Line

OpenAI just admitted something the AI industry doesn’t like to say out loud: not every AI product is a viable business. Sora was technically impressive and culturally significant — and it still died because the economics didn’t add up. The Disney debacle will make future partners think twice before betting big on OpenAI’s experimental products. And for everyone building on AI platforms: if OpenAI can kill its flagship creative tool with 30 minutes’ notice to a billion-dollar partner, your integration is never safe either.