From hit product to discarded project: Why did OpenAI axe the video-generation app Sora?

A sudden about-face

OpenAI quietly abandoned Sora, its short-lived video-generation experiment, after the app failed to meet internal expectations and ran headlong into hard technical and policy realities. Why kill a product that briefly captured headlines and imagination? The short answer: cost, quality and safety — three triangles that are hard to square for any company trying to scale generative video.

Technical limits and rising costs

Video generation is computationally thirsty. It requires far more GPU time than image or text models, which blows up operating costs and complicates a sustainable business model. It has been reported that Sora’s outputs still lagged human expectations on coherence, temporal consistency and fine-grained detail — problems that are less visible in single-frame image generation but fatal in moving pictures. Combine that with the staggering infrastructure bills and the need for heavy-handed moderation, and the economics look risky.

Safety, legal exposure and strategic refocus

There are also non-technical reasons. Reportedly, OpenAI’s leadership worried about deepfake risks, copyright exposure and the regulatory heat that comes with easy-to-produce synthetic video. For a company already navigating partnerships, API customers and global scrutiny, adding a high-risk consumer-facing video product may have seemed unwise. Instead, OpenAI appears to be prioritizing core models and chat-focused products that are easier to monetize and govern.

What this means for the market

Does Sora’s demise slow down progress? Not necessarily. Chinese players such as ByteDance (字节跳动) and Baidu (百度) are racing to commercialize multimodal video tools, and many startups worldwide will continue to push boundaries where others pause. But the episode is a reminder: technical novelty is not the same as product readiness. Policymakers and platforms will watch these developments closely — and companies will have to balance ambition with cost, safety and regulation.