DooDooLamb News
Why OpenAI killed Sora
Brief published March 29, 2026 · Original source published March 27, 2026
Original reporting by Hayden Field at theverge.com.
Automated brief. Verify important details at the original source.
OpenAI Discovers Video Generation Costs More Than Disney's Annual Catering Budget
OpenAI announced Tuesday that it's officially pulling the plug on Sora, its much-hyped video generation tool, after apparently discovering that running AI models capable of creating Hollywood-quality clips requires roughly the same amount of electricity as a small European nation. The company also quietly shelved its plans to integrate video generation into ChatGPT, which is corporate speak for "we tried to make the math work and our calculators started crying."
The news broke just months after OpenAI had been parading Sora around like a prized show pony, generating breathless coverage about AI's inevitable conquest of Hollywood. Remember those demo videos of photorealistic puppies frolicking through fields that definitely weren't generated by burning through enough compute cycles to power Delaware? Those same videos that had film executives simultaneously salivating and updating their résumés? Well, turns out making more of them would have required OpenAI to either charge customers approximately $47 per video or accept venture capital funding from several small countries.
Sources close to the matter suggest that OpenAI's leadership finally did the math on what it would actually cost to let millions of ChatGPT users generate even short video clips. The revelation reportedly came during a budget meeting where someone innocently asked, "So what happens if our user base actually uses this feature?" The resulting silence was broken only by the sound of spreadsheet cells bursting into flames. Video generation, it turns out, is computationally expensive in the same way that launching rockets is "mildly costly." Each frame requires the AI to essentially dream an entire image into existence, pixel by pixel, while maintaining temporal consistency across sequences. It's like asking someone to paint a thousand individual masterpieces that happen to tell a coherent story when flipped through quickly, except the painter is a massive neural network consuming electricity at rates that would make Bitcoin miners blush.
The competitive landscape didn't help OpenAI's case either. While Sora was busy being "the future of video generation" in carefully curated demo reels, companies like Runway, Pika, and a dozen other startups were actually shipping products that real people could use without requiring a personal loan. These competitors had the audacity to build sustainable business models instead of relying on the "we'll figure out the economics later" approach that has become Silicon Valley's unofficial motto. Even more embarrassing, some of these companies were producing comparable results while consuming significantly less computational resources, which is the AI equivalent of watching your neighbor achieve the same workout results while eating pizza instead of your $200-per-month supplements.
Meanwhile, the rumored $1 billion partnership with Disney apparently fell through faster than a Marvel movie's third act. Disney, a company that knows a thing or two about expensive creative endeavors, reportedly took one look at the projected operational costs and decided they'd rather stick with paying actual humans to create their content. When Disney, the company that once spent $200 million on a movie about John Carter of Mars, thinks your business model is financially questionable, it might be time for some serious soul-searching. The partnership would have theoretically allowed Disney to generate custom content using Sora's technology, but apparently the entertainment giant preferred the predictable expenses of unionized animators over the chaotic cost structure of AI inference at scale.
The broader implications here extend beyond just one company's failed product launch. OpenAI's Sora shutdown represents a rare moment of economic reality breaking through the AI hype cycle like a fire alarm during a particularly enthusiastic pitch deck presentation. The company essentially admitted that some AI applications, no matter how technically impressive, simply don't make financial sense at current infrastructure costs and usage patterns. This is the kind of honest assessment that venture capitalists usually hear only in private board meetings, delivered in hushed tones alongside phrases like "runway optimization" and "strategic pivoting."
For developers and founders watching from the sidelines, OpenAI's retreat offers a valuable lesson in the difference between technical capability and business viability. Sora could generate stunning videos, but stunning videos that cost more to produce than traditional methods solve a problem that mostly exists in conference room whiteboards. The real winners here might be the companies building more modest video tools with sustainable unit economics, rather than trying to boil the ocean with every inference call. Sometimes the future arrives not with a bang but with a spreadsheet showing that the bang costs too much money.
OpenAI's decision to kill Sora feels less like a strategic pivot and more like that moment when you realize your ambitious home renovation project is going to cost three times more than your house is worth. The technology was real, the demos were impressive, and the vision was compelling, but somewhere between the boardroom dreams and the AWS bills, reality came knocking with a very large invoice.