OpenAI’s Sora video generation technology could be making a comeback, but not necessarily for everyday users. The artificial intelligence (AI) company is reportedly in talks with a startup seeking access to Sora to modify its capabilities and develop software and services for the film industry.
According to a report by The Information, the startup was co-founded by Hollywood producer Jeffrey Katzenberg, former Sora research lead Bill Peebles, and former Dropbox chief financial officer Sujay Jaswa. The company’s name has not yet been disclosed.
If the discussions result in a deal, the partnership could mark a shift in how OpenAI approaches AI-generated video, with Sora potentially being used as a specialised tool for professional filmmakers rather than a publicly available consumer product.
Credits: Digit
OpenAI in Talks to Adapt Sora for Hollywood
The proposed partnership reportedly centres on giving the startup access to Sora so it can build and commercialise tools tailored to the entertainment industry. These could help filmmakers incorporate AI-generated video into their production workflows, although the specific applications have not been revealed.
However, it remains unclear how much access OpenAI is willing to provide. In particular, the discussions have not established whether the startup would receive Sora’s model weights, which contain the parameters learned during training and are essential to how the model generates video.
Access to the model weights could allow the startup to make deeper modifications to the technology. Even without confirmation on this point, the reported discussions suggest that OpenAI may be considering a more flexible approach to deploying Sora beyond its original consumer-facing applications.
Peebles’ involvement could also prove valuable. Having previously led research on Sora, he brings experience with the technology’s development and underlying capabilities. His expertise could help the startup adapt the model to the requirements of professional video production.
Katzenberg and Jaswa have also been associated with WndrCo, a technology investment firm. Their involvement brings together experience in Hollywood production, AI research and technology business development, potentially positioning the venture to explore commercial opportunities in AI-assisted filmmaking.
Sora’s Rise and Discontinuation
OpenAI introduced Sora in 2024, attracting considerable attention for its ability to generate videos from text prompts. The technology demonstrated how AI models could turn written descriptions into moving visuals, opening up possibilities for content creation, advertising and filmmaking.
The company subsequently expanded Sora’s availability through a standalone application and application programming interfaces (APIs), allowing users and developers to experiment with AI-generated video. These offerings reportedly experienced strong initial interest, with the standalone application briefly reaching the top of download charts.
However, the technology also faced significant commercial challenges. Generating videos requires substantial computing resources, making large-scale operation expensive. Despite the initial excitement surrounding Sora, its ability to generate sufficient revenue to justify these costs reportedly remained limited.
In March 2026, OpenAI announced that it would discontinue Sora. The decision appeared to bring an end to the technology’s public-facing journey, raising questions about the future of the company’s video generation ambitions.
The reported startup discussions could offer a different path forward by shifting the focus from mass-market adoption to specialised commercial applications.
Credits: CNBC
Could Professional Filmmaking Give Sora a Second Life?
The film industry could offer a distinct opportunity for AI video generation. Professional production involves complex workflows, and filmmakers may be interested in tools that help them develop visual concepts, experiment with scenes and create supplementary footage.
A modified version of Sora could potentially be tailored to these requirements, although the startup’s intended products and capabilities remain undisclosed. Its commercial model could also differ from a consumer application, with software licences or industry-specific services providing alternative ways to generate revenue.
Nevertheless, significant questions remain about the potential partnership, including the extent of OpenAI’s technology-sharing arrangements and whether the resulting tools will meet the industry’s technical and creative requirements.
For OpenAI, the discussions could represent an opportunity to extract further commercial value from technology that has already attracted widespread attention. Rather than returning Sora to the general public, the company may be exploring whether a more focused, professional offering can deliver a sustainable business.
For now, no final agreement has been confirmed. If the reported talks lead to a deal, Sora could return in a considerably different form, serving filmmakers and entertainment companies instead of everyday users.
The development would highlight a broader question facing AI companies: whether advanced generative tools can find more sustainable business models by targeting specialised industries rather than relying solely on mass-market adoption.
