Home / Technology / AI Video's Expensive Ghost: Sora's Exit Stuns China
AI Video's Expensive Ghost: Sora's Exit Stuns China
8 Apr
Summary
- OpenAI's Sora video tool shutdown signals high costs, not market opening.
- Chinese AI video firms must court marketers, not casual users.
- Sora's costly operation highlights computing power and regulatory risks.

OpenAI's discontinuation of its Sora video generation platform serves as a critical warning to Chinese AI companies rather than an open market invitation. The platform's operational costs, reported to be around $1 million per day, highlighted significant economic hurdles. This has led to a surge in user growth for Chinese competitors like Kuaishou's Kling.
Chinese AI video firms are advised to pivot their strategies. Instead of pursuing viral, low-quality content, they should focus on attracting marketers and professional content creators. Companies must also proactively address potential regulatory issues, including copyright infringement and the misuse of deepfake technology.
The immense demand for computing power was a major factor in Sora's shutdown. While OpenAI aims for foundational AI models, Chinese specialists focused on application layers could still find a viable business. However, the absence of Sora intensifies competition among these firms, especially with ByteDance's planned integration of its Seedance 2.0 into CapCut.
Sora's prior deal with Disney to license characters for AI video could have set a precedent for copyright holder collaboration. Without this, Chinese competitors face increased scrutiny from artists and regulators, particularly when expanding internationally. Clearer watermarking and robust internal policies are essential to mitigate abuses.
Ultimately, Chinese AI video companies should prioritize empowering genuine creativity over mass generation. The focus needs to be on sustainable business models that provide value to professional users and businesses, rather than solely chasing ephemeral viral trends. This shift is crucial for long-term viability.