A viewer in São Paulo opens an app, picks a 90-second drama about a billionaire's secret heir, and watches it in fluent Portuguese with dubbed voices and localized subtitles. The show was written, translated, voiced and cut by algorithms in China. This is no longer a pitch; it is a balance-sheet line item.
China's microdrama (短剧) format — vertical, hyperbolic, bingeable — has become one of the country's most quietly successful cultural exports. The engine behind it is increasingly AI, and the clearest case study is DramaWave, run by Kunlun Weiwei (昆仑万维).
How big the overseas wave actually is
Third-party data from Sensor Tower paints the scale. In Q1 2025, global in-app-purchase revenue for short-drama apps hit roughly US$700 million — about four times Q1 2024's US$178 million — with cumulative revenue near US$2.3 billion and quarterly downloads over 370 million (6.2× the prior year). The United States is the single largest revenue market at close to US$350 million, about 49% of the total.
The leaders are Chinese-owned: ReelShort took about US$130 million in Q1 2025, DramaBox about US$120 million, with DramaWave (under Kunlun Weiwei) a fast riser at roughly US$47 million as of late April 2025, with 53 million cumulative downloads.
What DramaWave's numbers show
Kunlun Weiwei's 2025 annual report turns the trend into hard figures for its short-drama and AI short-drama platform business:
- Revenue of ¥1.617 billion (≈ US$228 million), up 864.92% year on year.
- Revenue share jumping from 2.96% to 19.73% of the company.
- Monthly revenue near US$36 million and an annualized run-rate (ARR) above US$400 million as of December 2025.
- Two platforms — paid DramaWave and free FreeReels — whose combined monthly active users rank first among overseas short-drama apps per third-party data.
Why the format travels
The microdrama (短剧) formula — a hook in three seconds, a cliffhanger every episode, a paywall at the emotional peak — was perfected on Chinese smartphones and turns out to be near-universal. Vertical video fits how people actually watch on buses and in bed, and the emotional beats (wealth, revenge, romance, rebirth) need little cultural translation. That is what lets a Chinese studio ship a Portuguese- or Spanish-dubbed drama and land in Brazil or Mexico's top charts. AI then compresses the cost of making dozens of localized variants, turning a single hit template into a portfolio of market-specific bets — the real machinery behind the download numbers above.
Where AI actually enters
The format's economics only work because production is cheap and fast, and AI is doing the repetitive parts: script generation, translation, dubbing and editing. DramaWave has pushed into "AI manga" (animated-style dramas) and AI live-action themes, with users averaging over 30 minutes of watch time. According to DataEye, DramaWave shipped about 980,000 ad creatives in 2025 — the most of any overseas short-drama app — while FreeReels added 150,000. FreeReels' Q1 2026 downloads were estimated above 120 million, topping the free short-drama download chart.
In other words: AI lowers the cost of making and localizing a show, and a firehose of paid ads drives it to foreign users.
The uncomfortable caveat
This growth is real but expensive. Kunlun Weiwei remained loss-making in 2025, and its selling expenses jumped 81.5% to about ¥4.18 billion, largely to fund short-drama and AI-platform user acquisition. The model leans on heavy ad spend and an overseas market (over 94% of company revenue) exposed to platform-policy and geopolitical risk. "AI-built" also does not yet mean "AI-only" — human writers and editors still steer the hits.
Honest limitations
- The ¥1.617 billion, +864.92% and US$400M-ARR figures come from Kunlun Weiwei's own 2025 annual report and are company-reported, not independently audited market shares.
- Sensor Tower and DataEye numbers are third-party estimates with their own methodology; treat them as directional, not exact.
- "AI production" here means AI-assisted scripting, translation, dubbing and editing at scale — not fully autonomous show creation; we did not verify the share of any single drama that was machine-generated.
- The business's dependence on paid-user acquisition and on overseas app-store policies is a documented risk in the company's own disclosures; sustainability is unproven.
- We did not assess content-quality or regulatory questions (e.g., local content rules) in individual markets.
What readers can do now
- Watch the ARR-to-profit gap — a US$400M run-rate means little if user acquisition costs keep the business loss-making; track whether DramaWave's margins improve.
- Test the localization claim — watch a DramaWave or FreeReels show in a non-Chinese language and judge the dubbing and cultural fit yourself.
- Follow the AI share of production — the real story is whether AI moves from assistant to author; watch for disclosures on how much of a drama's script and edit is machine-generated.