In a studio in Jiaxing, one person clicks a few buttons and a finished short drama comes out — no camera, no set, no actors. A year ago the same result took a crew, a location and the better part of a month. The shift is not a tweak to the pipeline. It is a different kind of factory, and it is reshaping what Chinese entertainment looks like abroad.
Key takeaways
- More than 90% of micro-dramas (短剧) newly released in China from January to August 2026 were made with AI, according to China Daily citing industry data.
- A live-action season once cost about US$300,000; an AI-made season now runs US$10,000–20,000.
- Overseas micro-drama apps reached 258 million monthly active users by mid-2026, with the overseas market forecast above US$6 billion for the year.
- The bottleneck has shifted from production to distribution and localisation, not the other way around.
- China introduced its first sector-specific microdrama regulation in September 2026, requiring AI content to be labelled per episode.
What changed on the production floor
The numbers come from companies and industry bodies that have lived the change. Jiuzhou Culture Media, a short-drama company in Jiaxing, says its AI production centre now delivers a finished micro-drama (短剧) in seven to ten days, down from one to two months with traditional shooting, at roughly 10,000 yuan (≈US$1,400) versus hundreds of thousands before.
In Nanjing, Xuanjia (炫佳) runs its Kino-AIGC audiovisual large model (大模型) across the whole chain — script, storyboard, translation, editing. The company says production fell from about one and a half months to under ten days, efficiency rose around 95%, and cost dropped to roughly one-tenth of the old figure. Overseas business is close to 40% of its output, and the model now covers all ten ASEAN countries.
At the industry level, the China Netcasting Services Association reported about 128,000 micro-dramas released in the first quarter of 2026, of which titles made with AI accounted for roughly 122,000 — over 95%.
Why the format travels
AI collapses the two costs that kept short drama local: production and translation. A 60–80 episode micro-drama that once needed a film crew can now be finished in under two weeks. Yuewen Group, a major online-fiction platform, says AI now handles more than 80% of the translations on its overseas bestseller lists and has cut translation costs by about 90%; by the end of June 2026 its WebNovel platform had launched over 30,000 AI-translated works, making up 40% of the platform's novel revenue.
In Qiantang district in Hangzhou, after ByteDance's video-generation model Seedance 2.0 launched in early 2026, daily AI drama output reached roughly 50 times that of live-action equivalents, and district enterprises grew revenue at an average annual rate of about 74% over two years.
The overseas audience is real
DataEye, a research firm, found monthly active users of overseas micro-drama apps reached 258 million by mid-2026, and it forecasts the overseas market will exceed US$6 billion for the year, with AI and animated dramas making up more than US$4 billion — nearly 70%. The United States, Japan and Canada were the top three markets by in-app purchase revenue from January to July, together close to half the total. India, Indonesia and Brazil led downloads, at 21.5%, 19.4% and 9.4%.
Crazy Maple Studio's ReelShort, one of the highest-grossing vertical-drama platforms in the United States, now keeps local production teams in Los Angeles. American viewers prefer stories rooted in their own culture, so Chinese teams handle the format while local writers adapt the scripts — a reminder that AI lowered production cost without removing the need for localisation.
According to Li An, chief scientist at BrainNet (脑机网), the durable advantage in AI content will not be generation quality, which is converging quickly, but the depth of localisation — matching story, voice and cultural reference to each market.
A rule arrives with the boom
The speed has drawn a regulator. China's first sector-specific microdrama rule, the Measures for the Administration of Microdrama Development (《微短剧发展管理办法》), took effect on 1 September 2026. It classifies dramas by investment and subject tier and requires AI content to be labelled on a per-episode basis. For studios that already produce at machine speed, the practical effect is a labelling step baked into the export pipeline rather than a brake on output.
Honest limitations
- The "more than 90% AI-made" share counts titles released, not viewing time or revenue; most AI titles break even and only a minority of creators profit, DataEye's vice-president noted.
- Figures come from industry associations and one analytics firm (DataEye), not an audited government census, and the China Netcasting Services Association and DataEye use different methodologies.
- Quality, not throughput, is the open question: a large share of AI titles see little audience, and the cultural fit that drives retention still needs human judgement.
- The September 2026 labelling rule is new; enforcement depth and its effect on overseas platforms are not yet measurable.
What readers can do now
- If you produce or commission short video, pilot an AI pipeline for translation and dubbing first — that is where cost drops are largest and risk lowest.
- If you invest or partner in content, judge studios by localisation capability and audience retention, not raw output volume.
- If you operate in a regulated market, build per-episode AI-content labelling into your workflow now; the compliance baseline is moving.