Ninety-second episodes, shot vertically, sold a coin at a time. It is the fastest-growing format in video and almost nobody outside it can say how big it actually is.
- A microdrama is a 60–90 second episode, by Deloitte’s description, shot vertical, in a series of dozens — built for a phone held one-handed.
- Most often you buy coins and spend them to unlock the next episode at a cliffhanger — though subscriptions, ads and hybrids are all being tried.
- Deloitte forecasts $7.8 billion of in-app revenue in 2026, up from $3.8 billion in 2025.Deloitte, TMT Predictions 2026, “Tiny episodes, massive appeal”, read at source 23 Sep 2026: “In 2025, in-app revenue for micro-series content is forecast to reach US$3.8 billion.” “In 2026, Deloitte predicts that the revenue growth of in-app micro-series will more than double, reaching US$7.8 billion.”
- But published estimates run from under $4bn to $14bn, because they do not count the same things. Be suspicious of any single confident figure.Media Partners Asia via Variety, 6 Aug 2026, read at source 23 Sep 2026: “MPA forecasts outside-China microdrama revenue overall will grow from $2.7 billion in 2025 to $3.6 billion this year”. Omdia via Variety, 20 Jan 2026, read at source 23 Sep 2026: “Omdia expects the category to grow to $14 billion by the end of 2026, with $3 billion coming from markets outside China”.
- The economics are the story, not the AI. Cheap to make, sold by the episode, marketed on the same platform that hosts it.
- AI is arriving here first precisely because the format tolerates rough edges that cinema does not.
Take a soap opera. Cut each episode to about ninety seconds, turn the camera sideways, and charge for the next one at the moment it gets interesting.
That is the core of it. Everything else — the budgets, the AI, the revenue numbers — follows from those three decisions, even where the payment model varies.
What the format actually is
The craft constraint that follows: a hook in the first three seconds and a turn in the last five. There is no room for a slow open, because the viewer is one thumb-flick from something else and the next episode has to be worth paying for.
How big — honestly
The number depends entirely on what you count
Deloitte forecasts in-app micro-series revenue of $7.8 billion in 2026, up from $3.8 billion in 2025. That is the most cited credible figure and it counts one specific thing: money spent inside the apps. On audience, Deloitte cites approximately 662 million micro-drama users in China as of 2024, and separately notes that only about 30% of US Gen Z and millennials were even familiar with the format — a useful pairing, because it shows how much of this is one market and how early the others are. It expects the US share of global revenue to fall from about half in 2025 to roughly 40% in 2026.
Other published estimates for the same year range from under $4 billion to around $14 billion, depending on whether they include China, advertising, licensing, or the wider "short drama" category. Figures around $20 billion circulate too, in trade write-ups and marketing copy — they do not match Deloitte’s basis, and this page cannot tell you what they do count.Media Partners Asia via Variety, 6 Aug 2026, read at source 23 Sep 2026: “MPA forecasts outside-China microdrama revenue overall will grow from $2.7 billion in 2025 to $3.6 billion this year” (outside China only). Omdia via Variety, 20 Jan 2026, read at source 23 Sep 2026: “Omdia expects the category to grow to $14 billion by the end of 2026, with $3 billion coming from markets outside China” (China included). Deloitte, TMT Predictions 2026, "Tiny episodes, massive appeal", read at source 11 Sep 2026: “in-app revenue for micro-series content is forecast to reach US$3.8 billion” in 2025 and “will more than double, reaching US$7.8 billion” in 2026; “approximately 662 million micro-drama users nationwide” in China as of 2024; and “the United States will account for half of global revenue in 2025, but its share will decline to 40%”.
A four-fold spread is not a rounding difference. It means the category has no agreed boundary yet — so when you see one confident number, the useful question is what it excluded.
Who is actually making the money
The market is concentrated despite having, reportedly, more than two hundred platforms. A small handful take most of the revenue — ReelShort and DramaBox lead outside China, with a long tail of apps sharing what is left.
Two details worth having. Profitability is thin, and marketing is where the money goes: one research firm estimates the largest platform ran at a net loss in 2025 while spending just over half its revenue on user acquisition and marketing, and expects its first meaningful profit only this year. And the US is both the largest single market and a shrinking share of the whole — Deloitte expects it to fall from about half of global revenue to roughly 40% as other markets scale.Media Partners Asia via Variety, 6 Aug 2026, read at source 23 Sep 2026 — a research-firm estimate, not a company filing: “After an estimated $12 million net loss last year”; user acquisition and marketing costs “falling from just over half in 2025 to under 45% by 2028”; the leader will “post its first meaningful profit at scale”. Deloitte, TMT Predictions 2026, “Tiny episodes, massive appeal”, read at source 23 Sep 2026: “the United States will account for half of global revenue in 2025, but its share will decline to 40%”. First-hand: until 23 Sep 2026 this paragraph said one platform had “reportedly run deliberately unprofitable to buy growth” and another “reported a modest net profit on a few hundred million in revenue”, from trade reporting it did not link; neither could be traced, so both were replaced.
The reliable facts are the shape — short, vertical, paid per episode, concentrated among a few apps. The dollar figures are contested, and anyone quoting one without saying what it counts is repeating a press release.
Why the economics work
The cost side is the innovation
Traditional television spends heavily per finished minute and recovers it through licensing, advertising or subscription over years. Microdrama inverts that. Production is fast and cheap — compressed shoots, small crews, reused locations — and revenue arrives within days of release, direct from viewers.
That changes what a failure costs. A series that does not land loses a small amount quickly, so platforms can afford to try a great many and let the audience pick. It is closer to how mobile games are published than to how television is commissioned.
And marketing runs on the same rails
The clips that advertise a series are cut from the series, and they run on the same vertical feeds where people already watch. The product and its trailer are the same shape, which removes an entire adaptation step and most of the cost of it.
This is why the format is a natural home for generated video — not because the storytelling is advanced, but because the unit is short, disposable, and produced in volume, which is precisely where cheap generation has an advantage.
Where AI actually fits
It fits where the tolerance is highest. A ninety-second vertical episode watched on a phone forgives things a cinema screen does not: slightly odd hands, an inconsistent background, a face that shifts between shots. The format's own conventions — tight framing, fast cuts, close-ups — happen to be the conditions generated video handles best.
What it does not fix is the writing. The economics depend on the cliffhanger landing and the viewer paying for episode nine. That is structure and pacing, and no generation tool supplies it.
If you are producing in this format, the constraints are the ordinary ones: identity has to hold across dozens of episodes (character consistency), and every re-roll costs the full clip (what generation costs).
This is a category with enormous headline numbers, contested measurement, heavy paid user acquisition, and a payment model designed to be spent in small increments at moments of maximum emotional pull.
All four of those describe a real business. They also describe how bubbles look from the inside. The growth is genuine and documented; whether the current spending survives the end of subsidised acquisition is not yet answerable, and anyone telling you it is has something to sell.
If you are considering making one
Deloitte, TMT Predictions 2026 — in-app micro-series revenue of $7.8 billion forecast for 2026 against $3.8 billion in 2025, approximately 662 million micro-drama users in China as of 2024, and a US share moving from about half toward 40% · a published estimate range of roughly $3.6–14 billion for 2026 across other sources · platform share and profitability from trade reporting. Checked 6 September 2026.Deloitte, TMT Predictions 2026, “Tiny episodes, massive appeal”, read at source 23 Sep 2026: “reaching US$7.8 billion”; “approximately 662 million micro-drama users nationwide”. Media Partners Asia via Variety, 6 Aug 2026, read at source 23 Sep 2026: “MPA forecasts outside-China microdrama revenue overall will grow from $2.7 billion in 2025 to $3.6 billion this year”. Omdia via Variety, 20 Jan 2026, read at source 23 Sep 2026: “Omdia expects the category to grow to $14 billion by the end of 2026, with $3 billion coming from markets outside China”.
Deliberately, this page attaches no number to any named company. Platform revenue, user counts and profitability come from trade reporting rather than filings, disagree between sources, and move fast enough that a figure here would be wrong before it was useful. The structural description is the durable part; the market sizing is contested and is presented as contested rather than resolved.
The through-line: the innovation is not the video, it is the pricing. Ninety seconds, a cliffhanger, and a coin — repeated episode after episode.