An AI microdrama is a serialized scripted series built for vertical 9:16 mobile viewing, with episodes running 60 to 120 seconds each, produced partly or entirely with generative AI video, voice, and audio models instead of traditional cameras and crews. A typical season runs 60 to 100 episodes, ends nearly every episode on a cliffhanger, and monetizes through coin-based episode unlocks, ad views, or subscriptions. As of mid 2026, AI microdramas have moved from experiment to industrial supply chain, with more than 10,000 AI-generated titles releasing every month in China alone and Western platforms commissioning AI-native series for global distribution.
The microdrama format, defined
Microdramas, known as duanju in China where the format originated around 2018, compress the emotional arc of a full television drama into one to three minute fragments. The defining traits are consistent across every major platform. Episodes are vertical, shot or generated natively in 9:16 for full-screen phone viewing. Episode length sits between 60 and 120 seconds, occasionally stretching to three minutes. Seasons are long, usually 60 to 100 episodes, sometimes 150 or more, so a complete series totals roughly the runtime of one or two feature films consumed in bites. Pacing is relentless: an emotional spike lands within the first 15 to 30 seconds of episode one, and every episode closes on an unresolved beat designed to pull the viewer into the next.
The dominant genres are romance, revenge, billionaire and CEO fantasies, werewolf and supernatural arcs, and increasingly true crime and thriller as the format expands. The core paying audience skews female: Omdia data puts the primary viewer base at women aged 25 to 45, while Media Partners Asia describes the most lucrative US segment as affluent urban women between 30 and 60.
What makes a microdrama an AI microdrama
The AI in AI microdrama refers to the production method, not the story. In an AI-native production, the live shoot is replaced by a generative pipeline: scripts become structured scene prompts, image models establish characters and locations, video models such as Kling 3.0, Seedance 2.0, and Veo 3.1 generate the shots, lipsync and voice models handle dialogue, and music and SFX models complete the soundtrack. The entire chain can be orchestrated as a single repeatable workflow. On node-based platforms like MinionArts Vertex, a full episode pipeline lives on one canvas, which is why a small team can output serialized volume that previously required a studio.
The economics explain the shift. Traditional live-action microdrama seasons already run cheap, often shot in 7 to 12 days for under $300,000, which works out to roughly $2,000 to $3,000 per finished minute. AI production compresses that further. Chinese industry reporting puts AI microdrama costs at roughly one tenth of an equivalent live-action shoot, with the usable rate of generated footage now above 90 percent. South Korean platform Vigloo completed its AI-generated vertical drama in six weeks at a reported 90 percent cost reduction. When the marginal cost of an episode collapses, the format's natural appetite for volume takes over.
How big is the market in 2026
The numbers are no longer niche. Omdia estimates global microdrama revenue reached $11 billion in 2025 and will hit $14 billion by the end of 2026, with $3 billion generated outside China and the United States accounting for roughly half of that international total at about $1.5 billion. Deloitte forecasts in-app microdrama revenue alone will more than double in 2026 to $7.8 billion. In China, the format grew from about $500 million in 2021 to $7 billion in 2024 and surpassed the country's entire theatrical box office in 2025.
Engagement data is even more striking than revenue. Sensor Tower measured $2.98 billion in in-app purchases across short drama apps in 2025, up 115 percent year over year. In Q4 2025, microdrama apps drove 733 million global downloads, more than long-form streaming apps combined. US users of ReelShort average over 35 minutes per day in the app, ahead of mobile Netflix, Prime Video, and Disney+. As Omdia put it at MIP London in February 2026, microdramas are winning the battle for attention before they win the battle for scale.
The AI inflection: what changed in 2026
2026 is the year AI-native production became a supply chain decision rather than a demo. In January 2026, a new AI-generated microdrama went live on a Chinese platform roughly every 90 seconds. In March, around 50,000 AI-native titles were added to Douyin in a single month, and the share of AI-generated content in China's top charts jumped from 7 percent to nearly 40 percent within a year. China's state broadcaster CCTV released its own AI-generated drama in January 2026, and government subsidies now support the production base.
The platform layer industrialized in parallel. TikTok launched PineDrama, a dedicated serialized short drama app, in early 2026. ByteDance entered international markets with Melolo. Holywater, GammaTime, and Vigloo are commissioning AI content for the US and Europe, and the SAG-AFTRA 2026 contract added specific provisions for vertical microdrama casting, a sign that the format is now negotiated at the institutional level.
How AI microdramas make money?
Monetization is built into the story structure. The first five to ten episodes are free. From there, viewers unlock episodes with coins typically priced between $0.50 and $1.00 per episode, watch rewarded ads for free unlocks, or take weekly subscriptions in the $10 to $20 range. Many viewers spend $20 to $40 to finish a single series. Because each cliffhanger is a purchase prompt, writers place the biggest reveals immediately after payment thresholds. The model resembles free-to-play gaming more than television, and that gaming DNA is exactly what makes the format compatible with AI production: it rewards rapid iteration, data-driven rewrites, and volume testing.
What are the saturation risks?
Saturation and quality are the format's twin risks. With tens of thousands of AI titles releasing monthly, content homogeneity at the low end is severe, and researchers note that AI-generated video still shows gaps in portraying believable human behavior. Audience preference data also cuts both ways: one 2026 survey found 93 percent of viewers still say they prefer human actors, even as producers swap in synthetic performers for cost reasons. The winners in this market will not be the studios that generate the most footage, but the ones that pair AI throughput with disciplined story engineering.
Where does microdrama go from here?
This guide anchors our full AI microdrama series. From here, read how AI microdramas are actually made inside a node-based pipeline, how episode length and structure rules work, and how the coin economy converts cliffhangers into revenue. If you want to skip straight to building, the MinionArts Vertex canvas lets you chain image, video, voice, lipsync, and music models into one reusable episode workflow and start producing your first vertical series today.
