During China’s 2026 Spring Festival a new category of entertainment elbowed its way onto the home screen: AI-generated short dramas, from illustrated “AI manhua” to photorealistic AI‑actors, became mainstream hits on short‑video platforms. Titles in apocalyptic, suspense and historical niches clocked hundreds of millions of views, and what began as a technical curiosity a year earlier has evolved into a commercial production model that looks as much like a factory as a creative atelier.
The industry’s rapid expansion is being driven by a confluence of technical progress and economic incentives. Video‑generation tools such as Seedance 2.0—released by a ByteDance‑linked team in February 2026—have pushed usable output rates above 90 percent in independent tests, solving prior pains in shot continuity, camera movement and audio‑visual alignment. That technical leap makes minute‑long, plot‑driven episodes plausible at scale instead of being limited to short, disjointed clips.
Supply has surged. Independent trackers and platforms report dozens of AI short dramas with playbacks in the tens of millions, several crossing the 100‑million mark on Douyin. Market research firms estimate the 2025 AI short‑drama market at over RMB 12 billion, a year‑on‑year jump of roughly 300 percent and supply increases measured in multiples compared with 2024. Producers are now able to deliver content in days rather than months and at a fraction of traditional costs.
The economics are stark. Conventional short dramas typically cost RMB 400,000–700,000 per title, with premium series running into the low millions. By contrast, current AI workflows can push production costs to dozens of thousands of RMB per minute—translating into a full short series for a few hundred thousand yuan or less. Platforms also sweeten the math: incentive schemes and revenue guarantees for top AI titles can exceed RMB 3.6 million, making low‑cost productions a potentially high‑margin business.
That combination of lower capital needs and faster iteration is already reshaping the competitive landscape. Several platforms and major short‑drama channels are pausing or pruning traditional, actor‑led projects in favour of AI pipelines; a recent wave of project suspensions on one leading short‑drama aggregator underlines how quickly commissioning priorities can shift. For middle‑tier production houses and cast crews this represents an existential squeeze: fewer orders, tighter margins and a reallocation of media‑buyer budgets toward algorithmically generated content.
Yet the flood of inexpensive content carries well‑documented artistic and commercial limits. Early viral experiments revealed uneasy facial animation, stilted character interaction and jump‑cut editing; newer models have improved continuity and camera logic, but many AI scenes still converge on similar colour palettes, effects and staging. That standardisation risks audience fatigue: like pre‑made food, AI dramas can efficiently fill time but struggle to match the sensory depth and emotional texture viewers expect from higher‑end productions.
The medium’s future will therefore hinge on two forces working in tandem: continued advances in generative models and a durable human‑in‑the‑loop production model. User reactions and editorial curation will both train algorithms and map where human craft is still essential—narrative nuance, complex emotional beats, culturally specific performance. If that hybrid model takes hold, AI pipelines could free resources to finance riskier original IP and deeper storytelling; if not, the market risks sliding toward volume‑driven sameness and a bifurcated ecosystem of cheap filler and expensive prestige.
On the international stage, the industrialised production chain—web fiction to short video to AI generation—could give Chinese short‑form content new export potential, provided platforms and rights frameworks adapt. But anywhere this model scales, it will raise familiar questions about labour displacement in production crews, copyright over training material and the cultural consequences of algorithmically amplified tastes. For now the immediate story is economic: an efficiency shock that is already reordering commissioning desks, production houses and the small but influential bubble where hits are born.