The $7.8B Short Drama Market Forecast Is Real — and It’s the Wrong Number to Plan On
By the Modellix Editorial Team. This is an independent editorial analysis of publicly reported market data, not a product pitch. Disclosure: Modellix builds AI media generation tools, so we have an industry stake in how this market develops — we flag that openly and separate our views from any commercial interest.
Deloitte’s headline landed and went everywhere: global in-app revenue from micro-series will more than double in 2026, from $3.8 billion to $7.8 billion. The Deloitte 2026 TMT Predictions chapter on short-form serials, published in November 2025, also predicts the United States will hold half of global revenue in 2025 before sliding to roughly 40% as other markets learn to convert views into cash.
That number is defensible. It is also, for anyone deciding whether to staff a production pipeline, the least useful figure in the category.
The reason is definitional. Three respected forecasters are measuring three different things, and much of the coverage this week quietly stacks them as if they describe the same market.
Three forecasts, three definitions, one misleading comparison
Deloitte counts in-app revenue only — coins, unlocks, and subscriptions purchased inside micro-drama apps. That narrow aperture is why its number looks modest next to the others. Omdia sizes a broader microdrama market at $11 billion in 2025 rising to $14 billion by the end of 2026. That wider scope pulls in advertising and licensing alongside consumer spend.
Media Partners Asia, meanwhile, runs a global series that starts at $5 billion in 2023, reaches $12 billion in 2024, and is forecast at $26 billion by 2030 — while separately projecting China alone at $16.2 billion by 2030. So does the often-quoted “$14 billion in 2026” belong to MPA? No. That is Omdia’s figure. And “$16 billion in 2026” is not MPA’s global number at all; the $16.2 billion in MPA’s work refers to China by 2030.
| Source | What it measures | 2025 | 2026 | Long horizon |
|---|---|---|---|---|
| Deloitte | In-app revenue only | $3.8B (forecast) | $7.8B (prediction) | — |
| Omdia | Broader microdrama market | $11B | $14B | — |
| Media Partners Asia | Global short-drama market | — | — | $26B by 2030 |
| Media Partners Asia | China only | ~$9.4B | — | $16.2B by 2030 |
Source: Deloitte 2026 TMT Predictions, Omdia, and Media Partners Asia, retrieved September 14, 2026.
None of these is wrong. They are simply not interchangeable, and treating them as a single trajectory produces a market that appears to be growing in three directions at once.
Worth noting how conservative the Deloitte measure is. It captures only in-app purchases, and Sensor Tower puts that global category at $2.98 billion for all of 2025, up 115% year over year. Deloitte’s $3.8 billion for the same year sits above the app-store-only tally because it takes a slightly broader view of micro-series consumer spend. Either way, the number that is doubling is money users hand over inside an app — a demand signal, not a statement about whether the content being made earns its money back.
Key takeaway: Before quoting any microdrama market size, check whether the number is in-app revenue, total consumer spend, or a broader market definition that includes advertising and licensing. Mixing the three inflates the category.
The engagement gap is closing faster than the revenue gap
The revenue forecasts dominate coverage, but the more consequential number is time spent, because attention is what the format is actually winning.
Global users now spend an average of 25 minutes per day in short-drama apps as of April 2026, up 85% from January 2025, according to Sensor Tower’s State of Short Drama Apps 2026 report. Established streaming services held broadly flat at around 35 minutes. Southeast Asia is already approaching 40 minutes.
That ten-minute gap is narrower than it sounds, and the comparison is not apples to apples. Sensor Tower’s 35 minutes is a global app-category average across all streaming services. Omdia’s analysis of the same dataset reaches a blunter conclusion in the US: ReelShort users averaged 35.7 minutes per day in Q4 2025, ahead of Netflix mobile at 24.8 minutes, Prime Video at 26.9, and Disney+ at 23.0. Netflix still leads on monthly active mobile users in the US by roughly twelve million to ReelShort’s 1.1 million — as Omdia’s Maria Rua Aguete put it, microdramas are winning the battle for attention, rather than scale.
Read those two findings together and a clearer picture emerges: the category has already won individual engagement intensity, and it is still far from winning breadth of audience. The revenue forecast measures breadth. The engagement data measures intensity. Both are true, and they point at different strategic conclusions.
In the short drama market, volume is not the bottleneck. Value is
China is running the largest production experiment in the medium’s short history, and the results so far are sobering.
About 128,000 microdramas were released in China in the first quarter of 2026, and more than 95% of them — roughly 122,000 titles — were produced with AI-generated footage rather than cameras and live actors, according to China Netcasting Services Association figures reported by CNBC. A year earlier that share was effectively zero.
Two things about that 95% deserve care. It counts titles published, not viewing or revenue — a series shot with real actors that uses AI for subtitles is still filed under live-action. And the growth comparison depends entirely on which denominator you use. China’s National Radio and Television Administration counted roughly 33,000 microdramas released in all of 2025, while the association’s own white paper measured about 40,000 titles produced independently and in full. Against the NRTA count, one quarter of 2026 equals almost four years of the old pace. Against a like-for-like definition of finished, independently produced series, the multiple is smaller. The direction is not in doubt; the exact multiplier is.
Output grew. Hit rates did not follow.
Among the 221,900 AI-made short dramas new to Douyin in the first half of 2026, only 1,055 exceeded 100 million plays — a Douyin hit rate of about 0.47%. Fewer than 1.3% cleared a 50 million-play break-even line, and more than 90% of the companies producing AI short dramas have reportedly already shut down, based on industry estimates compiled from Chinese production data. The association’s own release notes that no AI series has yet produced a mainstream hit — over Lunar New Year 2026, live-action titles drew roughly 25 times the views of their AI counterparts despite being outnumbered fifty to one.
The revenue side is equally concentrated. Sensor Tower’s dataset tracks roughly 717 short-drama apps, and in the twelve months to April 2026 the top five captured 68.8% of tracked revenue, with the top twenty taking 95.0%, according to analysis of that dataset. Two apps alone, ReelShort and DramaBox, each approached $140 million in in-app purchase revenue in Q1 2026 against a category total of roughly $750 million. Five apps that took nearly seven-tenths of revenue drove only 28.8% of downloads — a gap that says everything about where pricing power actually sits.
Retention compounds the problem: once users exhaust their free episodes and hit a paywall, drop-off is steep, which is why so much of the category’s revenue concentrates into so few apps. Cheap production solves the supply problem and leaves the demand problem untouched. The labour side of that shift has been documented separately in what AI microdramas did to Hollywood production work.
The unit economics that actually govern staffing decisions
Where the forecast becomes actionable is cost per finished minute. That is the number a production lead can put against a client budget.
The range in published 2026 estimates is wide, and the spread is mostly workflow maturity rather than model choice:
- China’s most efficient AI pipelines run near $30 per finished minute, with some teams turning out a 100-minute AI drama every three to seven days, per Caixin’s investigation into how AI took over the industry, which also reports AI-enabled titles produced for under 20,000 yuan.
- A disciplined pipeline outside China lands around $50–80 per finished minute. A first production team building the workflow from scratch typically spends $100–200 per finished minute.
- India’s cost curve is falling fastest in relative terms. Redseer expects generative AI to cut the cost of a roughly 100-episode micro-drama series from ₹15–20 lakh today to ₹5–6 lakh within three to four years — a drop of about 60–70%, as reported in the Economic Times coverage of AI-assisted microdrama production.
The part most budgets get wrong is regeneration. At API list prices, a ten-episode run of 90-second vertical episodes costs roughly $70–260 in model calls if every clip succeeds first time at 720p, and $120–450 at 1080p or 2K. Real production multiplies that by two to three, because first attempts rarely survive review. Labor, not inference, is what dominates a season’s cost once the pipeline stops being fully automated. The API-side mechanics behind those numbers are covered in this walkthrough of AI media generation infrastructure.
There is a second cost that the production math hides entirely. CNBC’s reporting on the Chinese market quotes analyst Ashley Dudarenok estimating that the cost of buying 1,000 promotional impressions for a short drama rose from 50–80 yuan in 2023 to around 150–200 yuan in 2025, sometimes exceeding 300 yuan in competitive periods. Production got cheap. Distribution did not.
That arithmetic has a strategic implication the growth figures hide. AI collapsed the cost of attempting a series. It did nothing to the cost of acquiring an audience. Notes on how those attempts are sequenced in practice have been collected in a comparison of AI short drama production patterns in India and Indonesia, where vertical-native workflows matured earlier than in most Western markets.
Pro tip: Put cost per finished minute, day-seven retention, and customer acquisition cost in the same spreadsheet as revenue. The forecast tells you the category is big; those three numbers tell you whether your pipeline survives inside it.
India is buying the manufacturing playbook, not just the format
If China proved the production model, India is now underwriting it at company-building scale.
Kuku Technologies — parent of the audio platform Kuku FM and the short-drama app Kuku TV — is scaling an AI-native production unit with 1,000 hires by the end of the current financial year, staffing writers, filmmakers, and engineers. Reporting puts FY2026 revenue near ₹14 billion (about $146 million), roughly a fivefold increase, alongside an IPO targeting a valuation around ₹15,000 crore, or roughly $1.57 billion.
What makes the India case interesting is that the market it is attacking is still small. Lumikai values India’s microdrama sector at about $300 million with 100 million monthly active users, forecast to reach $4.5 billion by 2030 — a trajectory through a market that barely existed two years ago.
So the India bet is not that the market is large today. It is that AI-native production collapses the cost of the format to a level where an ad-supported model works, and that whoever builds that capacity first owns the supply.
The predictions being made publicly point the same way. Vinod Kumar Meena, co-founder and COO of Kuku Tech, told the Economic Times he expects AI-assisted content to reach roughly 80% of Indian microdramas within a year, leaving about 20% to live-action. Anshita Kulshrestha of TukTuki Entertainment gave a more conservative 50–60% on the same timeline. Those are named forecasts rather than measurements, and the gap between them is the honest range.
Two other moves are worth watching because they show where the category’s money now flows. In Korea, Spoon Labs opened the 2026 Vigloo Short Drama Festival in September — a $20,000 prize pool for 3-minute AI pilots from creators in Korea, the US, and Japan, submitted through Vigloo Studio, with twenty finalists posted to the platform in late October and ranked on viewing data before five winners are chosen. Spoon Labs itself took a ₩120 billion equity investment from Krafton in 2024, at the time the game publisher’s largest non-gaming bet.
In India, ZEE Entertainment has taken a stake in the micro-drama startup Bullet and is integrating a vertical micro-drama app into ZEE5, pairing AI-driven pricing with gamified retention for younger viewers. ZEE did not disclose the investment amount; a follow-on ₹100 crore injection into its ZBULLET subsidiary was approved in May 2026.
The clearest sign that the category is converging on the audio playbook, though, comes from Pocket Entertainment. The parent of Pocket FM and the microdrama app Pocket Saga crossed $500 million in annual revenue run rate, up 70% year over year, with 96 titles having passed $1 million each and 13 past $10 million, according to the company’s September 2026 announcement. ARR is a run rate, not audited revenue — the distinction matters when reading the headline. The company also credits AI with cutting time-to-market in a new country from about twelve months to two.
That last figure is the one to take seriously. Twelve months to two is not a cost story. It is a distribution story, and it is the first time this cycle that AI has been credibly linked to market entry rather than market output.
What to watch next
Four indicators will separate the forecast from the fantasy, and all are checkable over the next two quarters.
1. Whether the US share actually falls to 40%. Deloitte made a falsifiable prediction, not a directional one. If US revenue share holds near half through 2026, the thesis that other markets are monetizing better is wrong.
2. Whether India’s AI-native share approaches 80%. Meena’s prediction and Kulshrestha’s 50–60% bracket it. The spread is wide enough that either outcome is informative.
3. Whether the hit rate moves at all. A 0.47% hit rate on Douyin is a function of supply flooding every distribution surface simultaneously. If output keeps climbing and the hit rate does not, the constraint is discovery and retention, not production — and no amount of model efficiency fixes it.
4. Whether the top-five revenue concentration breaks. If 68.8% of tracked revenue keeps consolidating into a handful of apps, the category is becoming a studio business, not a platform business, and the returns accrue to whoever controls the funnel.
Deloitte, notably, flags its own uncertainty. Its chapter states that micro-dramas “may not signal a great change in the new mass media” and observes that algorithmic feeds and interest graphs “are not geared to support serialized narratives” — the very behavior the format depends on. That is the most useful sentence in the whole report, and almost nobody quoting the $7.8 billion is repeating it.
Pro tip: When a forecast doubles, find out what exactly doubled. In this case it is in-app consumer spending. That makes it a demand signal well worth tracking, and a poor proxy for whether a production business will be profitable.
Market data compiled from publicly available third-party sources — Deloitte, Omdia, Media Partners Asia, Sensor Tower, and industry press — each cited inline. All figures are third-party estimates or forecasts as of September 14, 2026; market-sizing definitions differ between providers. Treat ranges as directional rather than precise.
