OpenAI Enters a New Phase by Shutting Down Sora

On Tuesday, OpenAI announced the discontinuation of Sora, its AI video application, approximately six months post-launch. The organization also indicated that it would close the Sora API, which enabled developers and Hollywood studios to utilize the text-to-video model.

This decision reflects how the ChatGPT creator is seeking to streamline its focus in anticipation of a forthcoming IPO. OpenAI’s CFO, Sarah Friar, mentioned in a CNBC interview on Tuesday that the company must be “prepared to operate as a public entity.”

Since the debut of ChatGPT, CEO Sam Altman has managed the organization similarly to Y Combinator, the Silicon Valley incubator he once led, making investments in a diverse range of products. This includes Sora, a web browser, a series of hardware devices, robotics, and Codex, the AI-driven coding assistant.

These initiatives have yielded mixed results, with Sora’s growth notably stagnating in recent months. Following a peak of 3.3 million global downloads across iOS and Android in November 2025, Sora’s downloads plummeted to just 1.1 million by February 2026, according to the analytics firm Appfigures.

OpenAI researchers have characterized the company’s culture in recent years as “bottom-up,” indicating that resources are allocated to emerging promising ideas rather than strictly adhering to executive roadmaps. While this approach has fostered an environment conducive to AI research, it has also stretched the company’s GPUs and personnel thin, according to various sources.

As a result, OpenAI’s leadership has instituted a firm directive to concentrate its efforts on a few core areas.

One highlighted focus is a “super app” intended to integrate ChatGPT, Codex, and Atlas. OpenAI’s leadership envisions that merging these products into a single consumer interface can help position ChatGPT as a genuine super assistant. (The Wall Street Journal previously covered the super app and OpenAI’s initiatives to streamline its product offerings.)

Prior to launching ChatGPT in 2022, OpenAI was working on creating an AI agent capable of executing a variety of digital tasks for users. This product, initially called the “super assistant,” aimed to realize the vision of AGI, but sources indicate that it has proven more challenging to develop than anticipated. Instead, OpenAI has attempted to introduce agentic features within ChatGPT, such as Operator and ChatGPT Agent, although adoption has been limited. The firm is optimistic that a consumer agent developed around Codex will resonate more with ChatGPT users.

OpenAI is also enhancing its enterprise sector as it prepares for the public market. While Anthropic had previously been a frontrunner in the AI coding sphere, OpenAI’s Codex team has made significant progress over the past year. Codex has become a highlight for OpenAI, achieving over $1 billion in annualized revenue by January and continuing to expand.

Despite Sora’s initial launch being heralded with enthusiasm, the product didn’t align with OpenAI’s new strategic direction, leading the company to determine that its GPUs and researchers would be better utilized in other areas. In a statement to WIRED, an OpenAI representative noted, “As we streamline our focus and computing demand increases,” the Sora research team will pivot to “world simulation research aimed at advancing robotics to assist in solving tangible, real-world tasks.”

This decision appears to have derailed the company’s partnership with Disney, which had indicated it would invest $1 billion in OpenAI. Disney reportedly felt taken aback by the decision and announced it would no longer pursue the investment.

There remains an unanswered question regarding the implications of OpenAI’s newfound focus for its research teams. OpenAI is in competition with Anthropic, Google DeepMind, and Meta for a limited pool of top-tier talent. In January, OpenAI’s VP of Research, Jerry Tworek, departed the company after facing challenges in securing resources for his next major project. While many employees seem invigorated by the company’s decision, others may consider transitioning to rival labs if their projects lose priority.