Sora was breathtaking. It was also burning $1 million a day. Last week, OpenAI pulled the plug.

Impressive isn’t a business model

Last week, OpenAI shut down Sora, its AI video generation app, just six months after launch. And it wasn't a quiet exit. It also killed a $1 billion Disney partnership that would have let users generate videos featuring over 200 characters from Marvel, Star Wars, and Pixar.

The technology wasn't the issue.

Each 10-second clip cost OpenAI roughly $1.30 to generate —  and even with a shrinking user base, that added up to roughly $1 million a day in operational costs. Total lifetime revenue? $2.1 million. Every new user made the economics harder. At some point, the only rational move was to stop.

The curiosity trap

How many AI tools are your teams actually using versus just exploring?

​

Generating an AI video of a cat surfing is fun. It's also a one-time thing. Sora's downloads dropped 65% from their November peak by February 2026. The novelty wore off fast, because it always does. What Sora never answered: what problem does this solve that someone needs solved again tomorrow?

​

There's real value in exploring new AI capabilities — that's how teams figure out what's worth keeping. But exploration has to graduate into a harder question. At GTC 2026, Jensen Huang made the same point from the infrastructure side: the unit of competition in AI is no longer the most impressive model. It's how efficiently a system converts compute into economically useful output. OpenAI's own applications CEO put it more bluntly internally: the company could no longer afford "side quests."

Curiosity got you in. Now what?

Sora wasn't shut down because nobody tried it. It was shut down because trying it never became needing it. That gap is where most AI adoption quietly stalls.

​

The question worth sitting with isn't "are we using AI?" It's: which part of your sales motion would break if AI disappeared tomorrow? If the answer is nothing, you're still exploring. Which is fine. But exploration without a conversion point is just a very expensive hobby.

​

The teams seeing results stopped rotating through tools and went deep on two or three workflows. Account research. Follow-ups within the hour. Proposals pulled from past wins. Unglamorous — but it runs every week.

​

Sora's users showed up for the wow. They didn't stick around for the work. The opportunity for enterprise teams is to be the ones who do.

‍