OpenAI Abandoned NSFW Video Path
Why Sora failed
This move shows OpenAI is choosing higher value, repeatable software markets over the messy but powerful demand that often kickstarts new media tools. Adult use was one of the clearest ways to turn raw video generation into frequent consumer spending, but OpenAI has instead shifted GPU budget toward coding and enterprise products, where it already has stronger distribution and clearer monetization. Meanwhile, the companies winning in video pair generation with editing, templates, and publishing workflows, not just one shot prompting.
-
OpenAI's broader posture already points away from consumer video. By February 2026 it was at a $25B revenue run rate, and related research on Windsurf shows a push toward owning coding interfaces, usage data, and enterprise workflows, where model spend creates stickier recurring revenue than novelty video clips.
-
The practical gap in Sora was product shape. Runway sells web based editing software for creators and filmmakers, HeyGen sells avatar workflows for training and sales teams, and OpenArt sells story based generation plus editing. These products make money because users can revise scenes, swap assets, and export finished work, not just generate one clip.
-
The demand did not disappear, it shifted to looser platforms. xAI's positioning centers on real time consumer distribution through X and a rapidly scaled revenue base, which helps explain why permissive consumer media features can matter strategically even if they sit outside enterprise positioning.
Going forward, AI video value will keep moving away from standalone foundation models and toward full workflows. Google and ByteDance have the data, cash flow, and distribution to keep pushing model quality. Startups will keep winning the application layer by turning raw generation into editable, publishable output for creators, filmmakers, and business teams. OpenAI is now more likely to compete where usage compounds inside coding and enterprise seats.