ByteDance's Seedance 2.0 launched in early February 2026, and within days, the conversation in China's production industry shifted from "can AI make usable video" to "how fast can we ship it." The model, which generates multi-shot sequences with synchronized audio from text prompts, landed in a market already primed to exploit it: China's booming short-form drama and manhua (comic drama) sector, which Zheshang Securities projects will exceed 20 billion yuan this year alone.
Feng Ji, CEO of Game Science (the studio behind Black Myth: Wukong), didn't mince words. "The content field is bound to experience unprecedented inflation," he wrote on Weibo, predicting that traditional production costs would collapse toward the marginal cost of computing power. Coming from someone who shipped one of 2024's biggest games, that's not idle speculation.
The numbers behind the hype
China's micro-drama market hit roughly 50 billion yuan by mid-2025, surpassing theatrical box office revenue for the first time. The manhua subset is where AI tools are making the deepest cut. According to 36Kr reporting, Kuaishou's data showed daily revenue in the AI comic drama segment grew 900% between Q4 2024 and Q3 2025, with monthly episode production up 567%.
Those figures deserve some skepticism. Platform-reported growth percentages on a small base can look enormous, and "daily revenue" for a young category means the absolute numbers may still be modest. But the production cost data tells a more convincing story. Individual manhua drama production costs have dropped to 100,000-150,000 yuan per series, roughly a quarter of what traditional short dramas cost. Teams of 10-12 people are turning around 50-episode runs in 30-45 days.
The show that keeps getting cited is "Tomorrow is Monday" (明日周一), a manhua drama produced by a 10-person core team with about 40 freelancers. Fifty episodes in 45 days. It racked up millions of views quickly and reportedly generated substantial merchandise revenue, though the specific figures circulating online are hard to independently verify. The production used AI tools from Shengshu Technology's Vidu, not Seedance 2.0, which is a distinction the breathless coverage tends to skip.
What Seedance actually changes
The official model page describes a unified multimodal architecture handling text, image, audio, and video inputs simultaneously. The practical breakthrough, according to China Daily's coverage, is character consistency across shots, the thing that made previous AI video tools useless for anything narrative. Pan Helin, a member of an expert committee under China's Ministry of Industry and Information Technology, attributed Seedance's performance partly to ByteDance's data advantage: running the world's largest short-video platforms gives you unmatched training data on what makes video compelling.
That's a polite way of saying ByteDance has something none of its Western competitors can replicate. OpenAI doesn't run TikTok. Google doesn't run Douyin.
The 36Kr analysis is more specific about where the cost savings land. E-commerce video production, where clips don't need artistic nuance, just clear product demonstration, is getting flattened first. Concept verification for games and film is next: the work of showing "what this could look like" before committing real production budget.
So who's actually making money?
ByteDance's own Jimeng platform runs an AIGC Short Drama program that provides up to 50-70% investment for qualifying productions, with maximum backing reportedly reaching $2 million per project. The first paid AI short drama produced through the program, "Mysteries in the Greater Khingan Mountains," hit 30 million views in three days on Douyin.
Kuaishou isn't sitting idle. Kling AI hit $240 million in annual recurring revenue by December 2025, just 19 months after launch, with nearly 70% of revenue coming from professional content creators. Kuaishou released Kling 3.0 on February 4, 2026, days before Seedance 2.0 dropped, in what looks like a deliberate counter-programming move.
The competitive dynamic here matters more than any single model's specs. Two of China's largest content platforms are pouring resources into AI video tools and then funneling the output directly into their own distribution networks. That closed loop, from generation tool to content platform to monetization, doesn't exist anywhere else in the world right now.
The copyright problem nobody wants to talk about
Almost immediately after Seedance 2.0 launched, users started generating convincing recreations of Stephen Chow's classic films. His agent, Chen Zhenyu, publicly challenged whether the platform was turning a blind eye to mass infringement. The 36Kr article frames this as proof that top-tier IP becomes more valuable in an AI-saturated market, which is a convenient argument for the company selling the tool that enables the infringement.
The deeper issue: China's manhua drama boom runs heavily on adapted IP from platforms like Zhihu and web novel sites. As AI collapses production costs, the bottleneck shifts entirely to who owns the stories worth telling. Several platforms launched manhua creator incentive programs in late 2025, and Zhihu opened its IP library for adaptation. The scramble for licensable content is real.
Seedance 2.0 remains primarily available through ByteDance's Jimeng platform in China, with subscriptions starting around 69 RMB ($9.60) per month. International API access has been delayed. Kling 3.0 is globally accessible through Kuaishou's app, with a free tier and paid plans starting at $6.99 monthly.
