China Turns AI-Generated Video Into a Fast-Growing Commercial Industry

China’s push into AI-generated filmmaking is moving rapidly beyond technology demonstrations. Filmmaker Zhu Zhili told Reuters that two years ago Shenzhen was the obvious location for an AI film studio because of its technology ecosystem, but that he is now regularly approached by officials from cities and industrial parks across China offering support to establish AI production businesses. Local governments are using subsidized offices, rent reductions, living allowances and computing support to attract studios and startups, effectively creating regional AI-content clusters. Shenzhen’s recently released 2026–2028 AI action plan illustrates the direction of policy: the city wants AI to reshape the content-production and consumption value chain and explicitly supports large-scale production of AI comics, AI-generated dramas and immersive entertainment. This turns filmmaking into another testing ground for China’s broader “AI+” strategy, which aims to embed artificial intelligence across industries rather than treating it solely as a standalone technology sector.

The economics explain why adoption has been so rapid. Reuters reported that producing one minute of an AI short drama cost around 5,000 yuan at the beginning of 2026 but had fallen to only a few hundred yuan by the first half of the year. One filmmaker told Reuters that a wedding sequence that might have cost about 60,000 yuan using conventional production could be created with AI for roughly 1,400 yuan. That cost compression fundamentally changes the production model because small teams can now attempt projects that previously required actors, sets, cameras, visual-effects teams and much larger budgets. At the technology layer, Chinese developers are also competing aggressively. Kuaishou’s Kling and ByteDance’s Seedance have emerged as internationally competitive video-generation models, while MiniMax launched its H3 model this year with video, audio and multimodal generation capabilities aimed at commercial applications including advertising, gaming, ecommerce and product design. Together, cheaper computing, increasingly capable models and government-backed infrastructure are lowering the entry barrier to professional-quality video production.

The result, however, is an enormous increase in supply. DataEye figures cited by Reuters show that 221,900 new AI shows were launched on Douyin during the first half of 2026, yet only 1,055 surpassed 100 million views, meaning less than half of one percent reached that level. Separate figures released by China’s broadcasting regulator show just how large the broader short-drama economy has become: China had more than 800 million micro-drama viewers, the market exceeded 100 billion yuan in 2025, and about 430,000 micro-dramas were released during the first eight months of 2026. More than 90% reportedly used AI technologies in their production. These figures illustrate both the opportunity and the problem. Generative AI is making content dramatically cheaper to produce, but it is not creating more consumer attention. As production capacity becomes abundant, distribution, intellectual property, recognizable characters, storytelling quality and audience retention become more valuable than the ability to generate footage itself.

That dynamic could eventually make the AI-video market resemble other Chinese industries in which technological progress and policy support produced explosive capacity growth before profitability caught up. Content prices are falling, creators complain about plagiarism and repetitive material, and actors and voice performers are increasingly concerned about digital replicas and employment displacement. Regulators are responding at the same time that the government continues to promote the sector. China’s rules requiring explicit and embedded identification of AI-generated content took effect in September 2025, and enforcement has intensified in 2026. Major short-video platforms have also been required to make creators classify uploads that contain AI-generated material. For investors and companies, this suggests that the biggest long-term value may not necessarily sit with the thousands of studios entering the market. Model providers, cloud and computing infrastructure, large distribution platforms and production companies able to build defensible intellectual property may capture more durable economics. China has already demonstrated that it can industrialize AI video at extraordinary speed; the harder question now is whether the industry can convert that scale into sustainable returns.