🦝 The Most Expensive Demo Reel Ever Made
Here's the claim: OpenAI told Disney that Sora was a production-ready creative platform worth a billion-dollar partnership.
Here's reality.
Sora launched in September 2025. By April 1, 2026 — six months later — OpenAI killed it. Not pivoted. Not sunset gracefully. Killed. And they did it quietly, burying the announcement under the same-day Microsoft MAI news cycle, because apparently even OpenAI knows you don't advertise a funeral. The Disney deal, worth a reported $1 billion, collapsed with it.
Let me do the math because the math is devastating 💰
Sora's operating cost, per leaked financial documents: $15 million per day. Total revenue generated across its entire lifespan, according to internal figures reported by The Information: $2.1 million. That's not a rounding error — that's the entire commercial output of what OpenAI pitched as a revolutionary creative tool. For every dollar Sora earned, it burned roughly $1,300. You could have literally set cash on fire on YouTube and generated more views. Usage never exceeded 2% of ChatGPT traffic — despite an estimated $200 million in marketing spend.
But here's the part that should terrify anyone considering an OpenAI partnership 🔍
ChatGPT Plugins — launched 2023, killed 2024. GPT Store — launched January 2024, abandoned by summer. Sora — launched September 2025, dead by April 2026. Three major platforms in three years, all with partner ecosystems built on top, all terminated. The pattern goes deeper than bad product management.
Disney didn't lose a billion-dollar partnership because Sora's video quality wasn't good enough. Disney lost it because they trusted OpenAI to maintain a product for longer than it takes to shoot a Pixar short. The $1 billion deal was structured as milestone-based payments. Disney hit zero milestones. Not one. Their own Pixar team tested Sora internally and called the character consistency problem — the inability to maintain the same character across frames — "fundamental, not fixable with fine-tuning." OpenAI's flagship customer publicly declared the core technology broken before a single milestone check cleared.
The verdict: OpenAI doesn't have a technology problem. They have a product discipline problem. Every launch is a proof of concept disguised as a platform. They invite partners to build on foundations they haven't committed to maintaining. And when the unit economics don't work — and they never work, because nobody stress-tests them before launch — the product dies and the partners eat the loss.
Sora wasn't a product. It was a $15-million-a-day screensaver with a press kit.
