OpenAI Enters Its Focus Era by Killing Sora

OpenAI said Tuesday it would abandon Sora, its AI video app, about six months after its launch. The company also announced that it would shut down the Sora API that allowed developers and Hollywood studios to access the text-to-video model.
The move shows how the ChatGPT creator is trying to focus its efforts ahead of a planned IPO. OpenAI CFO Sarah Friar said in a CNBC interview on Tuesday that OpenAI needs to be “ready to become a public company.”
Since ChatGPT’s launch, CEO Sam Altman has run the company like Y Combinator, the Silicon Valley incubator he led, betting on a wide range of products. This includes Sora, as well as a browser, a family of hardware devices, robots, and Codex, its AI-powered coding agent.
These efforts have seen varying levels of success, and Sora’s growth in particular has stagnated in recent months. After peaking at 3.3 million downloads worldwide on iOS and Android in November 2025, downloads of the Sora app fell to just 1.1 million in February 2026, according to third-party analytics firm Appfigures.
OpenAI researchers have described the company’s culture in recent years as “bottom-up,” meaning the company allocates resources to promising ideas as they emerge, rather than following executives’ roadmaps. While this created fertile ground for AI research, it also reduced the company’s GPUs and employees, according to multiple sources.
Now, OpenAI executives have given a strict mandate to refocus the company around a few key areas.
One focus area is a “super app” that will combine ChatGPT, Codex and Atlas. OpenAI executives hope that combining these products into a unified consumer interface will help the company transform ChatGPT into a true super assistant. (The Wall Street Journal previously reported on the super app and OpenAI’s efforts to simplify its offerings.)
Before OpenAI launched ChatGPT in 2022, it planned to create an AI agent that could perform all kinds of digital tasks for users. This product, then dubbed “super assistant,” was supposed to bring the promise of AGI to life, but it was more difficult to build than OpenAI expected, according to sources. OpenAI has instead tried to launch agentic features in ChatGPT, such as the ChatGPT Operator and Agent, although adoption has been limited. The company hopes that a consumer agent built around Codex will resonate more with ChatGPT users.
OpenAI is also strengthening its enterprise operations as it prepares for the public market. While Anthropic was previously a leader in the AI coding race, OpenAI’s Codex team caught up last year. Codex is now a bright spot for OpenAI, surpassing $1 billion in annualized revenue in January and continuing to grow.
Although Sora launched with great fanfare, the product didn’t quite fit the new era of OpenAI, and the company decided its GPUs and researchers would be better used elsewhere. In a statement to WIRED, an OpenAI spokesperson said, “As we focus and demand for computing increases,” Sora’s research team will work on “global simulation research to advance robotics that help people solve real-world physical tasks.”
The move appears to have blown up the company’s partnership with Disney, which previously announced it would invest $1 billion in OpenAI. Disney was reportedly blindsided by the decision, and the company said it no longer planned to invest.
It remains an open question as to what OpenAI’s era of focus means for its research teams. OpenAI competes with Anthropic, Google DeepMind and Meta for a small pool of high-level talent. In January, Jerry Tworek, OpenAI’s vice president of research, left the company after struggling to secure resources for its next big bet. While many employees appear energized by the company’s decision, others may decide to join competing labs if their projects lose priority.




