China's Short Drama Industry Found Its Perfect Match: AI

A frightened young woman is thrown onto a bed by a tall, muscular man. Flame-like vines crawl across her skin. She levitates, then drops. A dragon tattoo materialises across her chest. 'Two months,' the man says. 'Give me an heir, or I will eat you.'

This is Carrying the Dragon King's Baby, and something about it looks slightly off. Glossy lighting, sure. But the visual texture sits in a strange uncanny valley between prestige television and a mid-budget video game cutscene. That's because there were no actors on set. No camera operators, no cinematographers, no CGI team. The whole thing was generated by AI.

Welcome to the new frontier of Chinese short drama.

For context: China's short drama industry has been running since 2018, producing ultrashort, melodramatic, frequently smutty serialised content built for smartphones. Episodes run one to two minutes. You can burn through an entire series in under an hour. The format is engineered for doom-scrolling, stuffed with betrayals, physical confrontations, and emotional detonations every thirty seconds. In 2024, the market turned over roughly $6.9 billion, overtaking China's entire theatrical box office for the first time.

Since 2022, Chinese companies have been pushing hard into international markets, translating hits and producing localised versions with regional actors. Apps like ReelShort and DramaWave have racked up close to a billion cumulative downloads globally. The US alone accounts for around half of all revenue outside China.

Now the industry is going fully algorithmic. An average of 470 AI-generated short dramas were released every single day in January, according to research firm DataEye. Companies like Kunlun Tech, which owns DramaWave and FreeReels, now offer more than 1,000 AI-generated titles. FlexTV has scrapped traditionally shot productions entirely. StoReels is targeting 100 AI dramas per month.

The economics are stark. A short drama series produced in North America with real cast and crew used to cost around $200,000. AI cuts that by 80 to 90 percent. Production timelines that once stretched three to four months now take under thirty days.

'No one comes to short dramas expecting high art,' says investor Shangguan Hong, former partner at Legend Capital. 'The short drama industry was already data-driven and real-time. AI just accelerates that logic.'

The content itself has always been optimised for formula. Platforms categorise projects with granular keyword taxonomies: 'campus romance,' 'enemies to lovers,' 'rags to riches,' 'reborn revenge.' Writers are expected to maintain near-constant emotional intensity throughout, cycling the same plot devices repeatedly because anything less will cause viewers to scroll away. 'You sacrifice narrative logic for shock value,' says Phoenix Zhu, a freelance short drama writer based in Suzhou. 'That's just how it works.'

Those simple, repeatable structures happen to be exactly what AI video models are good at reproducing. The format and the technology are almost suspiciously well suited to each other.

The production pipeline looks quite different now. Where there were once camera crews, lighting rigs, makeup departments, and VFX teams, there are now small groups of producers, writers, and something called AI asset curators. This last role involves translating scripts into prompts and generating reference images of characters, settings, and costumes for AI video models to work from. Chinese job listings for the role reportedly number in the hundreds, with most requiring little more than familiarity with the relevant tools.

Studios tend to mix and match platforms: Google's image model, ByteDance's Seedance, Kuaishou's Kling. One Beijing-based AI drama producer, Hanzhong Bai, notes that combining celebrity faces via prompt is fairly standard practice when designing characters. He also points out that AI makes previously prohibitive genres financially viable. Elaborate fantasy sequences, dragons, underwater kingdoms, all the visual excess that once made short drama budgets unworkable. 'We'll see many more dragon and mermaid shows,' he says, without apparent irony.

For the humans still in the pipeline, the shift has not been painless. Phoenix Zhu graduated in 2024 with a philosophy degree, spent months bouncing between rejection letters from traditional media companies, and eventually broke into short drama writing. She sold her first script in April 2025 for around 2,900 US dollars. More followed. Then two contracted projects were cancelled outright, rates started dropping across the industry, and the pay increases she expected as she built experience never materialised.

Writers have arguably fared better than most. Entire on-set roles have simply evaporated from AI productions. But even the writing process has been reshaped. Scripts are now expected in weeks rather than months. Descriptions need to be exhaustively specific because visual decisions once left to a cinematographer now have to be embedded in the text itself.

'Before AI, writing "he gave her a cold stare" might have been enough,' Zhu says. 'Now I might need to write, "cold beams of light shot out from his eyes."'

Kunlun Tech's CEO Han Fang frames the quality question as a volume problem. More people making shows means more chances for strong ideas to surface. The company plans to grow AI content to 20 percent of its platform output, while keeping some traditionally produced drama in the mix.

Research firm Omdia puts the global microdrama market at $11 billion in 2025, rising to $14 billion by end of 2026, with the US expected to generate $1.5 billion of that this year.

The formula has always been the point. AI just makes it cheaper to run it a thousand times a day.