China leverages short-video ecosystem, lower costs to carve out lead in AI video

Chinese artificial intelligence companies are gaining a significant lead over US rivals in video generation, according to analysts. While US tech giants like OpenAI and Anthropic lead in frontier large language models, Chinese firms are emerging as formidable competitors in AI video generation. Companies such as Alibaba Group Holding, MiniMax, and ByteDance dominate benchmark rankings on industry evaluation platform Artificial Analysis.

Alibaba's Wan 3.0 leads the text-to-video leaderboard with audio, followed by MiniMax H3 and ByteDance's Seedance 2.0 in third place. Chinese models also lead in image-to-video and video editing categories based on blind user comparisons. The success of Chinese video AI models can be attributed to proprietary training data, rapid iteration cycles, and strong domestic demand for short-form digital content.

ByteDance, in particular, has created a positive feedback loop, turning AI-generated short-form video into a profitable industry. Chinese companies have an advantage due to comparatively flexible copyright enforcement, which allows them to test and commercialize their models more freely than US developers. This is evident in ByteDance's ability to generate a 15-second video of a Ratatouille character, which Google's Gemini rejected due to copyright concerns.

Chinese models offer significant cost advantages as well. MiniMax's H3, which can generate 15-second 2K videos with stereo sound, is one-third the cost of mainstream competing products. Effective commercial prices for Chinese models are often heavily discounted through enterprise agreements, contributing to their growing commercial success.

Kuaishou Technology's Kling AI video-generation platform has generated over 850 million yuan in revenue in the second quarter, up over 200% year on year, with a user base of over 100 million globally.