Around 15 million dollars a day to generate 2.1 million over its entire life. The story of Sora is a wake-up call for any company investing in AI.
On March 24, 2026, OpenAI shut down Sora.
Operating cost: about 15 million dollars a day. Total revenue across the product's entire life: 2.1 million dollars. Downloads down 66% over the past four months.
No surprise, if you stopped to think about it. And yet Sora was considered one of the most ambitious AI projects in the world, a realistic video generator that had been making headlines from day one.
How do you go from "the future of video" to "we're shutting it all down"?
The problem wasn't the technology
Sora worked. The videos were impressive. The demos were spectacular.
The problem was that almost no one was willing to pay enough to cover the production costs. Impressive demos don't fix the books.
In the AI video market, meanwhile, Runway, Kling and Google Veo had already reached equivalent quality at far lower costs. Whoever arrives later with higher costs won't hold up, even starting from a position of media advantage.
The lesson for anyone evaluating an AI project in their company
Sora's story isn't only about OpenAI. It's about any organization weighing whether, and how, to invest in AI.
The question that makes the difference isn't: is this AI impressive? It's: does this AI generate real, measurable value from the very first week of use?
If the answer only comes after months of experimentation, heavy investment and internal presentations, the Sora risk is always lurking.
The AI projects that survive have a few things in common:
- A specific use case — not "let's use AI", but "let's use AI to do X, which today takes Y hours and costs Z".
- A clear metric — you already know how you'll measure success before you start.
- Sustainable scalability — the marginal cost of each use makes sense from day one.
How we work
At Media Engineering, every AI project starts from an operational brief, not from a demo. Together we define the use case, the success metrics and the running costs before building anything.
It's the approach we take in every sector we operate in: whether it's corporate training, events, internal communication or customer engagement, the starting point is always the same — a real problem, a clear metric, a system that works in production.
Not demos. Solutions that work.
Let's evaluate your AI project together, before you invest
We start from an operational brief: use case, success metrics and running costs defined from the outset. That's how you avoid the Sora risk and build only what generates measurable value.
Contact us →
