Sora is sliding: OpenAI hits reality after the hype

Remember the collective shock when OpenAI first showed what Sora could do? It seemed that camera operators and editors might as well update their résumés en masse, and that Hollywood should switch to making potato chips. Videos of mammoths drifting across clouds and neon Tokyo convinced everyone that the future had arrived, and arrived for good, with no hope left for traditional video production. But then January came, and the dry statistics say otherwise: downloads of the Sora mobile app fell 45%. Subscription revenue went down with them. It looks like the great video revolution has begun a prolonged and rather painful hangover.

The problem is not that the technology suddenly got worse or that the neural network forgot how to draw. The problem is that OpenAI seems to have slightly overestimated how willing the average user is to pay for generating video on the go. Asking ChatGPT to write a polite letter to a grumpy neighbor or put together a workout plan is one thing. Trying to make a good clip on a small smartphone screen is quite another. In practice, the result still often resembles a strange neural-network dream, with extra limbs and floating object physics. Hype is high-octane fuel: it burns very bright, but it burns out almost instantly unless it is fed with real everyday usefulness.

Look back at the industry's recent history and we have seen such rises and falls before. Think of Lensa AI, which got practically every second Instagram user to generate cyberpunk or fantasy portraits of themselves. Download charts shot into the stratosphere then, too, and then dropped like a stone once the fashion for AI avatars passed.

The difference is that people expected something far bigger from Sora than a fun toy for stories. Sam Altman and his team promised a professional tool that would turn media production upside down. For now, though, Sora in its mobile form looks like an unjustifiably expensive toy that people played with for one evening, were sincerely amazed by, and then put on the shelf when the notification about the second month's subscription charge arrived.

A drop in interest of nearly half in a single month is not just a market correction; it is a serious alarm signal for the whole development team. Users quickly realized that Sora still lacks precise control tools for real work, and that it costs too much for simple entertainment. On top of that, competition in this segment has grown fierce.

While OpenAI rested on its laurels after the stellar launch, projects such as Runway, Pika and Luma did not sit idle. They steadily rolled out features professionals actually need: motion brushes, camera control and frame extension. As a result, OpenAI has fallen into the trap of its own aggressive marketing: it created inflated expectations that the current version of the mobile app simply cannot meet.

What does all this mean for the AI market as a whole? The period of blind enthusiasm and unlimited credit for any product with the letters AI in its name is coming to an end. Users and investors now want more than a pretty picture or a funny video: they want predictable results, integration into workflows and clear value for their money. If Sora does not soon evolve from a generator of random visual wonders into a full working tool with deep settings and a clear pricing policy, it risks becoming the most expensive and technologically advanced monument to unfulfilled expectations in OpenAI's history.

Bottom line: The wow effect is officially over, and harsh economics have begun. Can OpenAI prove that Sora is a tool for creation, and not just a flashy trick to show off at tech conferences?

*Meta has been designated an extremist organization and is banned in the Russian Federation.