Short videos dubbed “Chinese soda-dramas” have been spreading across YouTube Shorts and Instagram Reels. Structured like soap operas, these micro-dramas are built from dozens of one- to two-minute installments. Most episodes finish with a jarring twist designed to keep viewers coming back. Lately, creators have begun using AI to seamlessly alter actors’ faces and backgrounds, and those clips are proliferating quickly.
In China, the micro-drama format is expanding fast. Born in the COVID-19 era, it has become one of the entertainment sector’s fastest-growing niches.
About 660 million people watched these shows in 2024. Industry projections put this year’s revenue north of 120 billion yuan (approximately $19.2 billion) — roughly 25.58 trillion KRW (about $19.2 billion) — nearly double China’s entire box office haul of 68 billion yuan (roughly 14.5 trillion KRW (about $10.9 billion)).
The format is spreading internationally. Chinese short-form apps such as RillShot, DramaBox and TopShot are beefing up their services abroad, and domestic firms like Spoon Labs have started producing AI-assisted short dramas on platforms such as BeGlue.
South China Morning Post reports that much of the micro-drama boom boils down to economics: low production costs and rapid turnaround make profitability far easier than in traditional drama production.
Chinese studios are racing to shave costs and compress schedules with AI tools. Some claim they can produce AI-driven episodes for roughly $30 per minute (about 43,600 KRW per minute).
Content companies say they now spend about 30% of production budgets on AI, shortening production cycles from three months to one and cutting costs to roughly one-fifth of conventional shoots. Tools such as ByteDance’s XiDance 2.0, Kuaishou’s Kling 3.0 and Shengshu’s Bidu have improved markedly; for some viewers, AI-generated material already feels nearly indistinguishable from live-action in certain respects.
That has driven a surge in AI-powered micro-dramas. One report said platforms released an average of more than 470 AI micro-dramas per day in January alone.
Their popularity has jumped: micro-dramas now account for 38% of the top-100 drama chart, up from just 7% a year earlier.
In March, Douyin — China’s version of TikTok — reportedly added about 50,000 AI-produced items, a pace exceeding 1,600 pieces a day.
Local governments have played a role, offering tax breaks, subsidies, GPU access, data privileges and traffic support to studios that produce micro-dramas.
Chongqing built the Liangjiang Film, TV and Animation Cultural Creation Park to support vertical-format production; more than 300 production teams reportedly visit each year. Linping in Zhejiang Province has invested over 100 million yuan (roughly 21.32 billion KRW, about $16.0 million) to back creators and launched China’s first micro-drama production hub.
In January, state broadcaster CCTV even produced an AI comedy drama, a signal of official-level support for the technology.
Analysts say China’s micro-drama sector has achieved international competitiveness. Government backing, a growing ecosystem and rapid AI advances have combined to form an ecosystem that is difficult to replicate elsewhere.
But the fast embrace of AI has also sparked disputes over intellectual property and performers’ rights. Some AI-generated stars have drawn comparisons to well-known Chinese actors, and experts warn that even without using actual photos, AI characters that viewers identify as real performers can trigger portrait-rights claims.
Actors report a sudden drop in opportunities. Li Zhao, a background-actor specialist, told The New York Times that jobs have evaporated in recent months and that the group chat where actors once shared gigs has gone quiet — “as if the rain just stopped all at once,” he said.
In response, China’s National Radio and Television Administration (NRTA) tightened registration and review rules for AI-generated content. Existing works must be reviewed by the end of March 2026 or face removal.
Those rules may slow the pace of releases, but regulators and industry figures say they could push creators toward higher-quality, compliant productions.
Chinese micro-dramas are now finding international audiences. From January through August 2025, overseas micro-drama revenue reached $1.525 billion — a 195% year-over-year jump. Major producers have built substantial user bases in the U.S. and Southeast Asia; short formats alone generated $400 million in 2024.
The model contrasts with South Korea’s drama industry, which typically concentrates resources on a handful of mega-hits.
Yuki B. Lee, CEO of brand consultancy Helios Worldwide, said China may offer an alternative to what she called “K-drama inflation” — showing that low-budget, high-volume content can collectively earn more than a single blockbuster.
When a ByteDance-made XiDance clip went viral in February, a U.S. filmmaker reportedly reacted, “We’re doomed,” reflecting industry unease.
Observers are now watching how China’s system — beyond just AI tools — could reshape global content production. Some argue AI-driven micro-drama factories are turning entertainment into a mass-production business.
Whether a strategy focused on speed and volume can sustain long-term quality is unclear. The question is whether China’s AI-powered content factories can close quality gaps and set a new global standard.
Here are the top stories as of the 6th.
■ [AI & Big Data Show] South Korea’s Largest AI Exhibition Opens at COEX in June
The main exhibition of STK 2026, the AI & Big Data Show, runs June 10–12 at COEX in Seoul. Major domestic and international firms will showcase industry-specific AX solutions and physical AI systems. The event is known for its business focus and is expected to draw many industry professionals.
■ ChatGPT Base Model Replaced with ‘GPT-5.5 Instant’ — Smarter, Fewer Hallucinations
Users who missed GPT-4o’s capabilities will be watching to see whether the new GPT-5.5 Instant can deliver. The challenge is to retain an engaging, near-human tone while boosting accuracy and reducing hallucinations and bias. Consumer reaction will be decisive.
■ Is the RAG Era Over? New Model Offers a 12-Million-Token Context Window
Researchers have released a model that sidesteps Transformers’ structural limits and minimizes computation even for long inputs. Its context window exceeds 12 million tokens — more than ten times that of many rivals — potentially reducing reliance on retrieval-augmented generation (RAG).
AI Times news@aitimes.com
Most Commented