Sam Altman arrived in Los Angeles for Vanity Fair’s Oscar afterparty earlier this month with OpenAI on the verge of licensing its Sora video generation tools to Hollywood studios, only to pull the plug on the project weeks later. According to the Wall Street Journal, Sora proved to be an expensive mistake.
A recent report from the Wall Street Journal provides further insight into OpenAI’s sudden decision to shut down its highly anticipated video-generation product that was positioned as the natural successor to ChatGPT’s breakthrough success. The decision caught major partners off guard, including Disney, whose chief executive Bob Iger had signed on to a vision that would have allowed users to create videos featuring iconic characters like Mickey Mouse and Darth Vader.
Disney executives learned of the shutdown less than an hour before it was publicly announced, leaving them shocked by the sudden reversal. What they did not know was that Sora had quietly become a liability for OpenAI in the months following its release, particularly as the startup tightened its operational focus ahead of a planned initial public offering.
The company was weeks away from completing work on a new AI model code-named Spud and needed to reallocate computing resources toward coding and enterprise products that would run on it. AI chips represent the most valuable commodity at leading research labs, and Sora was consuming far too many of them. The product was not profitable, and every user generating content drew down finite resources that OpenAI needed elsewhere.
Sora now stands as an expensive strategic miscalculation led by key employees who were central figures in the competitive AI talent war across Silicon Valley. Altman characterized the move as a necessary sacrifice for larger company goals, writing in an employee note that he was encouraged by staff willingness to make “difficult trade-offs” for the organization’s benefit.
The project had once represented Altman’s ambition to transform OpenAI into a creative pioneer of the AI era and a potentially lucrative revenue source. The company first previewed Sora two years ago, showcasing dreamlike landscapes that evoked the fantastical worlds of Hayao Miyazaki or the surrealism of Salvador Dalí. When OpenAI launched a standalone Sora consumer app last September, Altman compared it to the original ChatGPT release moment.
However, the app failed to achieve the traction its creators envisioned, with users finding it produced more “AI slop than AI magic.” Usage flatlined by year-end, and with OpenAI’s financial constraints tightening before its IPO, executives began scrutinizing Sora more critically and did not like what they discovered.
The research team was preparing to begin training a new model intended to power video generation within ChatGPT. Unlike language models that learn from text, video models must interpret entire moving worlds, making them substantially more expensive to develop. After calculating the costs, OpenAI decided to cancel the initiative.
The company now plans to redirect resources toward a new “superapp” incorporating agentic AI tools that can autonomously execute tasks like writing software, analyzing data, and booking travel. These productivity-focused products are gaining widespread workforce adoption, and OpenAI has trailed rival Anthropic in capturing this market, threatening its lead in the AI race. Altman informed staff that the Sora team would pivot to longer-term initiatives such as robotics.
An OpenAI spokeswoman stated the company is ruthlessly prioritizing computing resources based on long-term economic value. “This disciplined focus on where we apply that compute allows us to grow, innovate faster, and deliver more efficiently to enterprises and developers,” she said.
Researchers could track chip allocation through an internal dashboard, with some surprised by the resources devoted to Sora given its limited revenue and lack of contribution to language model capabilities. The project’s secrecy led former employees to describe it as a startup within a startup.
As 2024 progressed, OpenAI showed signs of falling behind competitors. Google’s Gemini gained consumer popularity, while Anthropic’s Claude Code coding tool won over Silicon Valley software engineers through its ability to write programs with minimal oversight. OpenAI rushed to release an updated version of its own coding product, Codex.
Yet Altman wanted OpenAI to also reshape popular culture and entertainment through technology. In early 2025, he enlisted former Twitter CEO and leftist censorship guru Parag Agrawal to informally consult on a separate social media project similar to X, and worked with then-Disney CEO Iger on a deal enabling fans to bring beloved characters to life through Sora.
OpenAI previewed the new Sora app to employees in late September before public release. Internal reception was mixed, with some staff concerned that a social media app built around engagement would damage the company brand, while others worried about safety implications of user-generated AI videos even with guardrails in place.
Sora reached the top of the App Store shortly after launch despite invitation-only access. Users who gained entry found it remarkable: typing any request would generate a 10-second video within minutes. The face-upload feature enabled users to star in short, wild films themselves, with Altman volunteering his own likeness for absurd and sometimes violent or disturbing content that he appeared unfazed by.
Loose copyright guardrails allowed boundary-pushing content to proliferate, including videos featuring Martin Luther King Jr. instructed to share dreams about topics ranging from new Fortnite seasons to changing Sora’s content policies. King’s estate complained, prompting the platform to remove his likeness.
Global user count peaked at roughly one million shortly after launch but never recovered, declining to under 500,000 in subsequent months according to Similarweb data. Sora was losing approximately one million dollars daily.
OpenAI attempted to salvage the product, announcing a multiyear Disney deal in December to license over 200 characters. Disney agreed to become a major OpenAI customer and invest $1 billion in the startup. Iger said in a CNBC interview that the deal offered Disney opportunity to participate in AI’s rapid growth and new media forms. Altman expressed hope that the partnership would enable new creative possibilities for users.
For Disney, the deal demonstrated a viable business model for licensing intellectual property for AI use. The day before announcing the OpenAI partnership, Disney had sent Google a cease-and-desist letter accusing the tech giant of massive copyright infringement.
In February, Iger announced on an earnings call that Sora-created short-form videos would soon appear on Disney+’s upcoming vertical video feed. Disney was also negotiating company-wide ChatGPT deployment.
OpenAI had recently begun piloting an enterprise Sora version with safeguards for companies like Disney. The tool, expected to launch this spring, would have allowed select Disney executives to use Sora for marketing campaigns and special effects without exposing their work to OpenAI.
However, OpenAI was already considering Sora’s elimination. The company fell further behind Anthropic, whose recent advances sparked fears that AI could replace traditional software and services, briefly triggering a stock market selloff. OpenAI recognized the need to prioritize productivity tools and began identifying areas for deprioritization. After initially planning to continue video generation through ChatGPT, the company opted to eliminate Sora entirely.
Disney’s $1 billion investment never materialized, and the relationship is now effectively dormant. Under new CEO Josh D’Amaro, Disney is actively discussing AI implementation with more than a dozen other partners.
“As the nascent AI field advances rapidly, we respect OpenAI’s decision to exit from the video generation business and to shift its priorities elsewhere,” Disney said in a statement. “We appreciate the constructive collaboration between our teams and what we learned from it.”
The landscape of the AI industry is changing at light speed, making it even more crucial that conservatives catch up on this technology. Code Red: The Left, the Right, China, and the Race to Control AI by Breitbart News social media director Wynton Hall is the blueprint for conservatives to create effective AI policies to take on Silicon Valley leftists and China at the same time.
Read more at the Wall Street Journal here.
Lucas Nolan is a reporter for Breitbart News covering issues of free speech and online censorship.
Discussion about this post