OpenAI Kills Sora: What a $1 Billion Disney Exit and an IPO Pivot Tell Us About Where AI Monetises
Commissioned, Curated and Published by Russ. Researched and written with AI.
OpenAI announced on March 24, 2026 that it is shutting down Sora – the standalone AI video generation app it launched with enormous fanfare just six months ago. The app is going. The API is going. The Disney deal worth $1 billion is going. All in one day.
The Sora team posted on X: “We’re saying goodbye to Sora. To everyone who created with Sora, shared it, and built community around it: thank you. What you made with Sora mattered, and we know this news is disappointing. We’ll share more soon, including timelines for the app and API and details on preserving your work.”
That’s the entire official explanation from OpenAI. The rest has to be read from the moves around it.
What Just Died
Sora launched as a preview in February 2024. It was immediately treated as a civilisational moment. Tyler Perry, who had been planning an $800 million studio expansion, put the whole thing on hold after seeing the demo. That is the level of impact the reveal created.
The first public version arrived in December 2024. The standalone Sora app – Sora 2 – launched in September 2025. OpenAI described it as capable of generating “hyperreal motion and sound” from text prompts, and for a brief period, that wasn’t just marketing. Clips circulating online were genuinely hard to distinguish from real footage at a glance.
With the shutdown, ChatGPT also loses its ability to generate video from text prompts. This isn’t just the app being retired. OpenAI is exiting the video generation business entirely.
The Disney Deal That Didn’t Survive
In December 2025, Disney and OpenAI announced what looked like a landmark agreement. Disney would invest $1 billion as an equity stake in OpenAI and receive warrants to purchase additional equity. In exchange, Sora would be licensed to generate user-prompted videos using more than 200 masked, animated, and creature characters from Disney, Marvel, Pixar, and Star Wars. Disney+ was due to add curated Sora-generated videos as part of the rollout, expected in early 2026.
It was positioned as the deal that reset the IP war between Hollywood and AI. Disney had originally opted out of Sora 2’s training dataset. Then it flipped, agreed to license its characters, and planted a $1 billion flag inside OpenAI.
That flag is now being pulled.
Variety reported that Disney has ended its partnership with OpenAI. A Disney spokesperson said the company “respects OpenAI’s decision to exit the video generation business and to shift its priorities elsewhere,” adding that they will “continue to engage with AI platforms to find new ways to meet fans where they are.”
Deadline reported, citing an insider, that the deal was never actually finalised and no money changed hands. The $1 billion that Disney was set to invest never moved.
That detail matters. It means this isn’t a case of Disney losing a billion. It’s a case of a deal being announced publicly – with warrants, characters, and a Disney+ rollout attached – before the product it was built around had demonstrated it could survive.
The IPO Pivot
The Wall Street Journal first reported the reason for the shutdown: OpenAI is streamlining its portfolio ahead of a planned IPO. The company is refocusing on coding and enterprise AI tools.
IndieWire confirmed the same: OpenAI is shutting down Sora “with the company refocusing its efforts on coding and other business ahead of a planned IPO.”
This is the more significant story than the video product itself.
OpenAI has been acquiring coding infrastructure. In March 2026, it acquired Astral, the company behind Ruff and uv – the fastest Python tooling in the ecosystem. That acquisition put the Python toolchain directly inside Codex. It wasn’t a side project. It was a deliberate move toward owning the developer workflow.
The mega-app plan that is replacing Sora combines ChatGPT, Codex, and the Atlas browser. That’s a clear statement about where OpenAI thinks its commercial gravity lies: developers and enterprise, not consumer creative. Coding is the product. Video was a headline.
There’s also competitive context worth noting. Reddit discussion citing NBC News reported that Anthropic’s strategy of skipping image and video generation entirely – to concentrate compute on text and code – has been gaining ground with businesses and developers. If that framing is accurate, OpenAI isn’t just pivoting to coding because it’s more profitable. It may also be conceding that a competitor who never bothered with video was right about where to put the resources.
What the IPO Clock Changes
OpenAI is not yet public, but it’s clearly operating with IPO logic. That logic is simple: before you go public, you need a coherent story about where revenue comes from. Video generation, as a standalone consumer product, was generating cultural attention. Whether it was generating the kind of unit economics that sustain a company with OpenAI’s cost structure is a different question.
Coding tools do have that story. GitHub Copilot has real enterprise adoption. The developer seat is a well-understood commercial unit. Companies pay for it on a per-seat basis and renew because their engineers depend on it. That’s a recurring revenue narrative that translates cleanly into IPO valuation.
AI video, as it existed in Sora, doesn’t have an equivalent. It’s creative, it’s compelling, it creates the kind of demos that end up on the front page – but it’s harder to monetise at the enterprise level, harder to build a SaaS moat around, and expensive to run. The compute required to generate a few seconds of high-quality video is substantial.
The Disney deal was supposed to be the monetisation thesis: license the IP, charge for character-based generation, create a new category. That thesis didn’t get far enough to be tested before the platform was discontinued.
What This Means for Engineering Teams
The practical read here is narrower than the cultural one.
AI video is not going away. Runway, Pika, Kling, and others are continuing to develop. What’s dead is OpenAI’s version – and more specifically, the idea that a large-language-model company with coding infrastructure at its core could also operate a sustainable consumer video product.
If you’re an engineering team evaluating AI video for production workflows – for marketing, content, or UX – the Sora shutdown doesn’t kill the category. But it should prompt a harder look at provider stability. Sora had a public partnership with one of the most recognisable IP holders in the world, and it still didn’t survive six months as a standalone product. The category is real; the vendors serving it are not all equally durable.
For teams building on OpenAI’s APIs, the more important signal is the direction of investment. OpenAI is moving toward Codex, toward coding agents, toward developer tooling. That’s where the product bets are being made. If your use case lives in that space, the trajectory looks stable. If it was anchored in Sora’s video API, you need a new plan.
The Sora Paradox
Two years ago, Sora was the demo that made an experienced filmmaker pause an $800 million construction project. That’s not nothing. The technical capability was real. The cultural impact was real. The Hollywood reaction – which ranged from alarm to hostility to ultimately, in Disney’s case, a $1 billion bet – was real.
What Sora couldn’t do was convert that impact into a durable business before OpenAI decided to stop trying.
The lesson isn’t that AI video is a dead end. It’s that generating hype and generating sustainable revenue are two different problems, and a company preparing for an IPO has to choose which one it’s solving. OpenAI chose revenue. That means coding. That means developers. That means enterprise.
Consumer creative AI is a feature. The product is the tool engineers use every day, and the companies that figure that out first are the ones that will still be standing when the hype cycle ends.
Sora was the hype. Codex is the bet. Now you know which one OpenAI thinks matters.