The set is a desk. On it: a laptop, a ring light, a printed beat sheet, and a small fan keeping the laptop from throttling. There are no actors waiting, no lighting rig, no caterer.
A producer watches a vertical frame — a woman turning in a doorway, then the cut — and marks it for regeneration. The episode is not shot. It is selected.
Behind her, a season takes shape in the time it used to take to cast one. That compression, more than any model release, is what China is now exporting at scale.
Key takeaways
- Overseas microdrama (短剧) MAU reached 258 million by mid-2026, in a market DataEye puts above US$6 billion for the year.
- Synthetic and animated drama accounts for more than US$4 billion of that, roughly 70% of the total.
- More than 90% of newly released microdramas from January to August 2026 were synthetic, per DataEye data cited by China Daily.
- A live-action season cost about US$300,000; a synthetic one now costs US$10,000–20,000, and a 60–80 episode series can finish in under two weeks.
- Revenue concentrates in the US, Japan and Canada, while downloads are led by India, Indonesia and Brazil.
The cost collapse
The headline comparison is blunt. A live-action season of a microdrama cost roughly US$300,000; a synthetic season now costs US$10,000–20,000, and a series of 60 to 80 episodes can be finished in under two weeks (China Daily, 3 October 2026). The saving is not a discount on the same product. It removes whole categories of cost: cast, location, crew, catering, travel, and the schedule that binds them together.
That is why the format, not the technology, is the export. A vertical drama is short, cheap and disposable, which makes it the ideal customer for a pipeline that produces acceptable frames rather than perfect ones. Nobody is grading the lighting. They are grading whether the viewer stays for the next episode.
The numbers show what happens when the constraint lifts. More than 90% of newly released microdramas between January and August 2026 were synthetic, according to DataEye data cited by China Daily. Where the constraint used to be crews, it is now how many stories can be localised.
What the money actually buys
The market's shape is stranger than its size. Overseas microdrama MAU reached 258 million by mid-2026, and the 2026 market is put above US$6 billion, of which synthetic and animated drama is more than US$4 billion — roughly 70% (DataEye). Most of the category is now machine-made.
But the same dataset shows where the value sits, and it is not where the volume is. The United States, Japan and Canada take nearly half of in-app revenue. India, Indonesia and Brazil lead downloads, at 21.5%, 19.4% and 9.4% of installs between January and July (DataEye).
Read those two facts together and the business model appears. Attention is global and cheap; payment is concentrated and expensive. A producer who optimises for views in Jakarta is running a different business from one who optimises for subscriptions in a US app store, and the second one is where the margin is.
Localisation is the product
This is why the model is the least interesting part of the stack. A synthetic scene can be regenerated on request; a story that lands in another market cannot. The work moves to dubbing, cultural substitution, episode pacing, title testing, platform operations and the unglamorous business of buying traffic.
Jiangsu's Xuanjia is a useful illustration of the shape. Its Kino-AIGC tools cover all ten ASEAN countries, cut production time from about a month and a half to under ten days, cut cost to roughly one-tenth and raise efficiency by 95%, with overseas business around 40% of output (Xinhua Daily, 23 June 2026). Note what the company sells: coverage and speed in specific markets, not a model.
The pipeline, in other words, is a service business wearing a technology costume:
- Script selection and cliffhanger pacing, tuned per market.
- Generation and regeneration of scenes to a fixed visual grammar.
- Dubbing and subtitle localisation, including idiom replacement.
- Platform operations: upload cadence, ad creative, retention testing.
The rules come home
On 1 September 2026, the Measures for the Administration of Microdrama Development took effect — the first sector-specific rule, with a three-tier classification system and per-episode labelling of synthetic content (Sina Finance, 15 October 2026). The rule applies to the domestic market. Most of the growth is outside it.
That mismatch is the interesting part. A producer can comply at home, label every episode, and still ship an unlabelled season to an overseas app store. Whether destination markets will demand the same labels is now the open question, and it is a commercial question before it is a legal one: platforms that carry the content inherit the reputational risk.
There is also a subtler cost. A format that was already dismissed as lowbrow now has a second reason to be dismissed, and the labelling requirement makes the synthetic share visible in a way that the export market has so far avoided. Regulation at home may end up shaping the product's reputation abroad.
Honest limitations
DataEye is an app-analytics vendor with an interest in a growing market, and its MAU figures may count the same viewer across several apps. The US$6 billion market figure and the US$4 billion synthetic share are not clearly defined on the same basis, so the 70% should be read as a company estimate rather than an audited split. "Synthetic" may cover anything from fully generated video to partly assisted post, which makes the 90% figure softer than it looks. No public data shows how much of the revenue reaches Chinese producers rather than platforms and advertisers.
What readers can do now
- Operators: budget for localisation and user acquisition, not generation; the model is now the cheap input.
- Investors: ask for revenue by market rather than downloads, and for platform concentration, which decides bargaining power.
- Researchers: track labelling rules in destination markets; they will determine whether the pipeline can keep shipping unmarked.