OpenAI’s decision to discontinue Sora represents a drastic change in their product development strategy. Although the AI video generation tool made a spectacular first impression, OpenAI chose to exit the project after only six months. This represents a clear shift towards products that have the potential to scale and retain users to ensure consistent revenues.
Sora was first released to the public on September 30, 2025. It made an immediate impact. Millions of users attempted to use the tool in the first few weeks.
It was featured prominently on app store charts and was one of the most discussed AI products of the year. Short texts could be converted to cinematic experiences, and this was a novel and impressive feature to have. However, this was not enough to sustain the product.
Why OpenAI is Pivoting from Video to Agentic AI
As of March 24, 2026, OpenAI declared that it will shut down Sora completely across all platforms. This includes the application, developer tools, and video capabilities in ChatGPT. Altman explained that the reason is to concentrate more on “agentic AI systems” that can be used in business settings.
The main reason is the cost.
Creating videos requires significant computing power. Large models need to process each frame of the video. At the time of maximum usage, Sora was consuming around $1 million per day. It is difficult to sustain such expenses when the number of users is declining.
Even though usage dropped off, the cost of infrastructure remained high. Unlike text or image models, video models cannot be scaled back easily without compromising quality.
The way users behaved contributed to the problem.
Sora gained 1.5 million active users per week shortly after launch. However, within a few months, user numbers dropped below 500,000. Users were trying out the tool, producing a few pieces of content, and then stopping. Retention remained low. The product worked as a demo but had difficulty becoming a daily habit.
This is a common pattern for new creative tools. Users will often try out a new tool, play around with it, and then go back to their old way of doing things. This makes it hard to justify high ongoing costs.
Another high-profile partnership ended during the shutdown as well.
The End of Sora: Why OpenAI Swapped Disney Magic for Serious Work
Disney had signed a deal in December 2025 for a three-year partnership. This deal included a $1 billion equity investment based on Sora and access to franchises such as Marvel, Pixar, and Star Wars. This deal had significant potential for providing premium content for users.
However, OpenAI informed Disney just less than one hour before announcing the shutdown publicly. This deal had not been funded at this time. This deal ended before it ever had a chance to come together. This may mean that the internal decision to shut down Sora came quickly or at least sooner than partners were aware of.
This shutdown represents a significant change in overall strategy.
OpenAI is instead focusing on enterprise products that help solve ongoing problems. This includes coding assistants, data analysis tools, and AI agents that can perform various tasks. This is more likely to encourage consumers to pay for the tool. It is also more likely to encourage consumers to continue using it.
Consumer products, however, require something more.
Consumer products require something more than just being popular. They require something more than just being interesting. They require something more than just being useful. They require something more than just being entertaining. They require something more than just being popular.
The company is consolidating its efforts.
OpenAI’s Pivot Toward Utility and World Simulation
OpenAI is instead planning to incorporate its key features into a broader “super app” based on ChatGPT and Codex.
At the same time, the Sora team would be focusing on robotics and world simulation research. These areas are more in line with the long-term goals of AI systems. They involve the interaction of the model with the world instead of producing one-off outputs.
Sora shutdown does not mean the end of AI video. Instead, it means the current model of high cost, low retention, and focus on consumers is not quite right. Perhaps the next model would be more suited to the video game industry, advertisers, or business users who can afford the cost.
Essentially, OpenAI has traded stability for flash. They created the Sora model to show what was possible. Now they’re focusing on what they can actually deliver.
