Key takeaways
- Roughly 20,000 AI comic-dramas were uploaded in January 2026; by March the figure reached 47,000 in a single month. Supply more than doubled.
- The hit rate has not moved: industry estimates put it at 0.16% — fewer than two in a thousand break through.
- One breakout title, The Treasure Basin: The Mixed Spirit Root Is the Real Boss, has passed 10 billion cumulative views on Hongguo's charts.
- SenseTime's Seko platform passed 100,000 users two months after launch and now exceeds 1 million. Overseas, Kunlun Wanwei said on 9 September that its AI micro-drama platform generates over US$65 million in monthly revenue, with more than 90% of new titles in July AI-generated.
Supply doubled. Attention did not.
Three months ago, one person producing one short drama a month counted as fast.
Now the industry uploads 47,000 in a month.
Supply rose by more than a factor of two and the money did not follow. Behind those numbers is a quiet transfer of where the value sits.
The number that sobers you up
Start with the hit rate, because it frames everything else.
Earlier this year a fantasy comic-drama titled The Treasure Basin held the top of the Hongguo charts and passed 10 billion cumulative views. By rights there should be a crowd of such titles. The industry's own estimate says otherwise: 0.16%.
Out of a thousand series, fewer than two surface. The other 998 finish their run and are forgotten.
Output has multiplied several times over and the hit rate has not budged. That tells you one thing clearly: when the barrier to entry is levelled, attention does not increase to match.
What is actually stuck is not the model
That billion-view series was produced jointly by Chen Ke's studio and Seko, SenseTime's AI video platform. Chen Ke's observation is one every AI micro-drama maker recognises: a large model can generate content, but it cannot yet carry a complete creative production.
The pain points are specific. Across a season of a hundred-plus episodes, the grinding work is shuttling assets between several different models. Change the camera angle or the lighting and the same character's face changes. Reusing a character across tools is close to impossible. Every new episode means rebuilding the asset library from scratch.
Models handle "generate a frame." Nobody handles "string a hundred episodes together."
That seam is currently the most valuable position in the industry.
Who is earning first: the people selling the pipeline
The position has a name. The pipeline.
Concretely, it is a system that bundles model orchestration, asset reuse and team collaboration. Feed in a script, get storyboards out. Lock a character, and every subsequent episode can reuse it. Multi-speaker lip-sync and cross-episode character consistency resolved in the background.
Domestic players include TapNow, Seko and Flova, with ByteDance internally testing its own industrialisation platform. The numbers speak: Seko went from launch to 100,000 users in two months and now exceeds one million.
The same is happening abroad. On 9 September, Kunlun Wanwei told investors its AI micro-drama platform has monthly revenue above US$65 million, with over 90% of new titles in July AI-generated; FreeReels was the only product in the overseas short-drama category estimated to pass 200 million downloads in the first half.
The model is the craft. The pipeline is the productivity.
Where an ordinary person still fits
There is a position. It has moved.
If you plan to earn money from "I can generate video with AI," that is a red ocean — the skill was scarce three months ago and now everyone has it. What is genuinely still scarce is three things:
Script and structure. Whether a series makes people click through episode after episode is mostly story. It is the weakest link for AI today and the hardest to replace.
Character and visual asset libraries. Once you have accumulated a stable, reusable set of character looks and scene styles, that is an asset you can sell repeatedly rather than a one-off film.
People who understand pipelines. Single-tool operators are saturated. Someone who can wire three or four tools into a stable production line and run a small team on it is the role studios are most short of right now.
What to do now
Do not start episode one from scratch — read the charts first. Supply is in surplus; platforms do not lack content, they lack structures that travel. Breaking the first three episodes of a comparable hit into a shot-by-shot rhythm table is worth more than studying tools in isolation.
Make character consistency your first priority. The number one reason viewers abandon a series is "the face changed," not "the picture is not pretty." Test that before anything else when choosing a tool.
And if you are in this for the long run, move upstream or move to the pipeline. Upstream means script, storyboard, character design. Pipeline means building the process. Staying in the middle — pure generation — is where prices only go one way.
When everyone can paint, the valuable thing is no longer the brush. It is the assembly line that turns paintings into a product.
Honest limitations
Upload counts and hit rates are industry estimates, not platform disclosures, and platforms count "uploaded" differently. Monthly revenue figures for overseas platforms are company statements made in investor Q&A. The 0.16% hit rate depends heavily on how a "hit" is defined.
Sources: industry statistics reported September 2026; SenseTime disclosures; Kunlun Wanwei investor response (9 September 2026). Information only — not investment advice.