A Chinese AI short drama is made every 36 seconds?

Published : 21/08/2026

AI short dramas｜70,000 new released in one month

Initially, AI short dramas were predominantly 2D animated series, before 3D versions slowly gained ground.

The usual plotlines revolve heavily around high fantasy, immortal cultivation, time slips, and riches-to-rags reversals, where protagonists power through challenges with a non-stop barrage of crowd-pleasing moments.

Given their demand for massive visual set-pieces that would prove prohibitively costly to film on location, these genres play directly to AI's strengths.

As technology advances, human-like AI dramas have emerged. Industry standards still classify them under the broader banner of animated serials. Though AI-generated, their aesthetic closely mirrors live-action filming.

Consequently, classic live-action genres like contemporary romance, dynastic family drama, and reincarnation revenge plots are increasingly being rendered in this format.

In other words, AI short dramas have evolved from animated-style content to a human-like form that is close to live-action, and are gradually entering the short drama market that was originally dominated by live-action actors.

According to a DataEye report, the popularity share of AI live-action dramas on short drama app Hongguo Manju (紅果漫劇) has remained at about 60% since April 2026.

But what really shook the industry was not the genres, but the production volume.

Statistics from Chinese media The Paper on newly launched short dramas on Douyin (Chinese local version of Tik Tok) each month found that: in August 2025, there were only 655 new AI dramas; a year later, this number soared to 74,313, an increase of over 100 times, accounting for 75.4% of new dramas that month.

In July 2026, the monthly release rate had reached an average of one AI short drama every 36 seconds.

AI short dramas｜5 people can produce 60 episodes in 1 week

How can it be so fast?

Filming a live-action short drama usually requires the collaboration of a team of at least 40 people.

The industry's proud "7+7+7" model, consisting of seven days each for preparation, filming, and post-production promotion, is already considered high-speed production.

AI has accelerated this pace even further.

In February 2026, ByteDance launched the video generation model Seedance 2.0, after which, according to industry sources, a team of five people with a few computers can produce a 60-episode AI short drama within a week, at less than one-tenth of the cost of a live-action short drama.

What is rewriting the rules of the short drama industry is not just the low cost and high efficiency brought by technology, but also a new profession with a name that sounds like a gaming term: the "card puller" (抽卡師).

On the production line for AI short dramas, after the team buys a script, the "card puller" is responsible for inputting the script content, visual requirements and prompts into an AI model to generate a large number of clips, from which they "pull" usable shots.

The material is then handed over to editors and post-production staff for processing and assembled into a complete short drama.

But "card pulling" is not always a success on the first try.

Each generated result has an element of randomness, and pulling clips where the characters are not distorted, the movements are natural, and the expressions are just right tests both prompting and operational skills, as well as relying on a bit of luck.

Therefore, the "card puller" must clearly specify the time of day, scene, camera distance, character's actions and expressions, lighting, and camera movement. The prompt for a single piece of material can be up to 5,000 characters long.

The main focus for live-action short dramas is "traffic acquisition", which means spending money on advertising to promote the drama.

AI short dramas, on the other hand, have an additional "card pulling" process: no matter how detailed the prompts are, one still needs to repeatedly select from a large number of generated clips to filter out usable footage.

AI short dramas｜Mass production does not equal a hit

AI has lowered the barrier to entry for drama production and has also led to an explosion in the number of works. So, AI short dramas must be very profitable, right?

Data from DataEye shows that in the first half of 2026, 221,900 new AI short dramas were added to Douyin, of which only 1,055 surpassed 100 million views, accounting for 0.48%; if 50 million views is taken as the break-even line, works that crossed this line accounted for only 1.3%.

In other words, for approximately every 77 new dramas, only one barely breaks even.

The other side of this explosion in production is that it is harder for a drama to be seen.

When more and more people can produce a drama in a week, the scarcest resource is no longer production capability, but the few minutes that an audience is willing to stop and watch.

Mass production does not guarantee a hit; instead, it makes it easier for good works to be drowned out by homogenised stories, crude visuals and sensationalist plots.

From a live-action crew of at least 40 people to an AI team of five people and a few computers, and from shooting a drama in real life to piecing together a drama with prompts and generated visuals, AI short dramas have indeed rewritten the production process.

But it has not made it easier for creators to succeed.

It has merely shifted the competition to a different battlefield: previously, the competition was for actors, funding and ad placement; now, it is also about who is better at mastering the AI models and who can retain an audience amidst a sea of content.

AI short dramas have turned drama production into a larger-scale card-pulling game; the hardest thing to "pull" is still the audience's time.

Read more:

Shooting short dramas in old Ming and Qing Dynasty houses?

Hengdian becomes "Shudian": How do hit series drive industrial development?

A market worth tens of billions with more users than food delivery?

Do north American audiences also love China's "overbearing CEO" trope?