For most of the history of video marketing, there has been a fairly obvious constraint: making good video is expensive.

You need a script, actors, locations, cameras, lighting, editors, production crews and time. Even a relatively simple campaign forces a brand to decide which ideas are worth producing before the cameras start rolling.

China’s booming microdrama industry is beginning to show us what happens when that constraint starts to disappear.

Microdramas were already designed for speed. They are short, mobile-first, heavily serialised stories built around immediate hooks, dramatic turns and cliffhangers. Now generative AI is pushing the economics even further.

Around 220,000 new AI-generated microdramas appeared on Douyin in the first half of 2026, collectively generating roughly 500 billion views. According to one Chinese production company, a five-person team using newer AI video tools could produce a 120-minute project of ordinary quality in about a week for roughly RMB15,000, or US$2,200. That is reportedly less than one tenth of the lower end of traditional live-action production costs.

Those numbers matter. But the more interesting question is what comes after them.

Because if producing video becomes dramatically cheaper, the competitive advantage shifts somewhere else.

And that shift could have important consequences for brands marketing in China.

China had already industrialised short-form storytelling

AI didn’t create China’s microdrama phenomenon.

The industry was enormous before the current generation of video models arrived.

Microdramas are typically made as series of very short episodes designed primarily for smartphones. The stories move fast. Romance, revenge, family conflict, fantasy, historical settings and sudden reversals are common because there is very little time to convince someone not to swipe away.

This sounds like disposable entertainment, but the commercial scale is difficult to ignore.

iiMedia Research estimates China’s microdrama market reached RMB67.79 billion in 2025, up 34.4% year on year. Its consumer research also found Douyin was the leading platform for watching microdramas among respondents, ahead of Hongguo Free Short Drama (formerly Tomato Free Short Drama) and Kuaishou.

Other industry estimates put the market considerably higher, which is worth noting because definitions and methodologies vary. What is much less debatable is the direction of travel. Microdramas have moved from internet curiosity to mainstream entertainment format surprisingly quickly.

Then AI arrived.

And microdramas turned out to be unusually well suited to it.

Microdramas may be almost the perfect format for generative AI

AI video still has weaknesses.

Long scenes can expose inconsistencies. Characters can subtly change appearance. Physical movements can look strange. Continuity remains difficult. A beautiful ten-second clip is much easier to generate than a coherent two-hour film.

Microdramas work around many of those weaknesses.

Scenes are short. Cuts are frequent. Plots move quickly. Visual spectacle can sometimes matter more than realism. Audiences already expect heightened characters, dramatic situations and improbable plot twists.

One of the breakout AI microdramas reported earlier this year was Master of Fengshui, a 61-part series involving a Chinese feng shui master battling a Japanese Shinto priest on a snowy mountain. Its opening episode reportedly received 100 million views within 12 hours of release. No stunt team or mountain shoot was required. The scenes were generated using AI.

It is easy to dismiss something like that as novelty.

That would miss the point.

The important thing is not whether an AI-generated fantasy series looks as good as a premium television drama. The important thing is that somebody can now make, publish and test an idea like this without accepting anything close to the production risk that would once have been involved.

That changes the economics of creativity.

When production becomes cheap, experimentation becomes cheap too

Imagine a brand has ten ideas for a video campaign.

Under a traditional production model, it probably doesn’t make all ten.

People debate the scripts. Budgets are allocated. Concepts go through rounds of approval. Eventually one or two ideas survive and production begins.

AI creates the possibility of reversing that logic.

Instead of:

Idea → production → distribution → results

the process starts looking more like:

Many ideas → rapid production → controlled testing → audience data → scale the winners

That distinction matters more than the fact that an AI video costs less.

Cheap production means cheap failure.

A creator can test one protagonist against another. Change the first five seconds. Try a different storyline. Alter the setting. Rework the ending. Produce another version for a different audience.

Most will probably fail. That is fine if failure is inexpensive.

China’s digital platforms are particularly interesting in this context because the infrastructure around the content already exists. Douyin isn’t simply somewhere to upload a video. It combines recommendation algorithms, creator ecosystems, advertising, livestreaming, stores and behavioural data.

Add dramatically cheaper content generation and you begin to get something closer to a creative testing machine.

Not one campaign with three variations. Potentially dozens of stories competing against each other in real time.

This matters because Chinese brands were already turning advertising into entertainment

There is another part of the story international brands shouldn’t overlook.

Microdramas were becoming a marketing format before generative AI seriously disrupted production.

China’s National Radio and Television Administration launched a broader “Microdrama+” initiative in 2025. One of the specific programmes was “Microdramas Showcase Brands”, encouraging companies, producers and advertisers to use the format to tell brand stories and connect with consumers.

Brands took the idea considerably further than product placement.

Data covering Douyin and Kuaishou found that of 315 new non-paid microdramas released in the first half of 2025, 185 were brand microdramas. Food and beverage, beauty, personal care, automotive, finance and tourism brands were among the categories participating.

KFC China has produced fantasy microdramas. McDonald’s has experimented with its own fictional series. Beauty brands have built products directly into romantic and dramatic storylines.

This is important because a microdrama does not have to behave like a conventional advertisement.

A viewer might discover the first episode because the story is funny, strange or genuinely entertaining, then keep watching as the brand becomes part of the narrative rather than interrupting it. And on platforms like Douyin, the path from watching the story to discovering or buying the product can be incredibly short.

Entertainment, discovery and conversion start occupying the same space.

AI could accelerate that model considerably.

The opportunity isn’t simply to make cheaper ads

This is where brands could easily take the wrong lesson from what is happening.

If generative AI reduces video production costs by 80 or 90%, the obvious response is to produce the same advertising more cheaply.

That may be the least interesting use of the technology.

A more significant change would be using AI to explore narratives that would never have survived a traditional production budget.

A skincare brand might test five completely different recurring characters rather than committing to one campaign ambassador.

A tourism organisation could create different storylines around food, romance, family travel and adventure, then watch which narratives actually build an audience.

A retailer could turn a product category into an episodic story instead of interrupting someone else’s entertainment with an advertisement.

And international brands could potentially experiment with China-specific concepts without constantly recreating expensive global campaign assets.

The key word here is experiment.

Because nobody should assume that putting an AI-generated character next to a product suddenly makes good marketing.

Infinite content creates infinite sameness

There is an uncomfortable consequence to making content much easier to produce.

Everyone else gets the same tools.

If every creator can generate cinematic landscapes, beautiful faces, fantasy kingdoms, explosions and impossible camera movements, those things quickly stop being impressive.

We may already be seeing the beginning of that problem.

The supply of AI microdramas is exploding, but supply isn’t the same thing as demand. Even in a market producing hundreds of thousands of titles, audience attention remains finite.

In fact, iiMedia’s consumer research suggests something surprisingly conventional about the supposedly disruptive microdrama audience: story and production quality still matter. Plot was the most commonly cited content element respondents cared about, followed closely by production quality.

Technology can make a video.

It cannot guarantee that anyone cares about it.

And that may be where the value starts shifting.

When production was difficult, production capability itself was valuable.

When production becomes abundant, taste becomes valuable. So does cultural understanding. Good writing matters. Characters matter. Knowing why something is funny in China rather than merely translating a joke matters. Knowing where a product belongs naturally inside a story matters.

AI may actually make those differences more visible, not less.

There is another problem brands can’t prompt their way around

AI-generated characters also bring copyright, likeness and regulatory questions with them.

China is already dealing with disputes over AI-generated faces that resemble real actors, while a new market is emerging in which people explicitly license their likeness for use in AI-generated dramas and advertising.

Regulation is moving quickly too.

China’s AI-generated content labelling rules have been in force since September 2025, requiring various forms of explicit and embedded identification for AI-generated or synthesised text, images, audio and video.

Microdramas are getting their own clearer rules as well. China’s new Microdrama Development Management Measures, which took effect on September 1, 2026, specifically state that microdramas generated or produced using AI must carry a prominent notice in every episode.

For international brands, the opportunity is to experiment at a much greater scale, but that experimentation still needs clear guardrails around copyright, likeness rights, AI disclosure and brand safety.

AI may be lowering the cost of production, but it doesn’t remove the need for good judgement.

The bigger lesson from China’s AI microdrama experiment

It is tempting to frame what is happening in Hengdian, , one of China’s biggest film and microdrama production hubs, as another story about AI replacing people.

There is certainly a human cost. Actors and production workers interviewed by CNN describe falling wages, disappearing shoots and, in some cases, leaving the industry altogether.

But for marketers, there is a second story happening underneath it.

China is showing us what happens when the cost and effort of turning an idea into a finished video suddenly becomes much smaller.

That changes the question.

For years, brands have asked:

What content should we produce?

Increasingly, the question may become:

What should we test, and how quickly can we learn from what audiences choose?

That is a very different marketing model.

The brands that benefit probably won’t be the ones generating the most videos. They will be the ones that understand which stories are worth telling, which cultural signals matter, where commerce fits naturally and when technology is helping rather than simply making more noise.

AI can make the production line much faster. It still can’t decide what people will care about. And in China’s increasingly crowded content economy, that may become the most valuable skill of all.