Sora Shipped. Nobody Needed It.

Sora launched in December with a wave of impressive demos and a multiyear deal with Disney. By March, Disney had exited the deal and OpenAI was winding the product down. That’s three months from “AI’s TikTok moment” to the graveyard.

Vine had the same arc. Initial burst, huge hype, then — nothing. Except Vine’s problem was constraints and corporate mismanagement. Sora’s problem is different: it was a demo that became a product before anyone figured out who actually needed it.

The Disney deal is instructive. They signed on for character creation. Which sounds promising until you realize Disney’s actual problem isn’t “we can’t make enough character variations” — it’s “we can’t let a model hallucinate Mickey Mouse with six fingers in something we publish.” The use case looked obvious from the outside and fell apart in contact with the actual workflow.

This is what happens when you build for impressiveness instead of usefulness. A technically stunning thing demonstrates a capability, and then the product team has to work backwards to find who actually has the problem and whether they care enough to pay for it. Sometimes that works. Often it doesn’t.

In my experience, the tools I use every day are not impressive — they’re useful. They fit into what I was already doing, they save me a specific kind of friction, and I’d notice immediately if they disappeared. I couldn’t say that about Sora.

The AI graveyard has been filling up for two years with technically impressive products nobody needed. Sora just joined.

Source: The New York Times