The 400-billion-yuan AI short drama market, "90% of the companies have already gone out of business"

"From my observation, more than 90% of companies producing AI short dramas have already closed down."

Sun Shuche, founder of Yilian Technology, told Pencil News.

He noticed that a large number of early AI short drama teams expanded production by renting computers and equipment; recently, a leading equipment leasing company told him that many teams have started to terminate their leases, and there are even cases of arrears and difficulty in recovering equipment.

On the other hand, the industry is still exploding in scale.

Data from DataEye shows that the market size of domestic AI dramas/manhua dramas in 2026 is expected to exceed 400 billion yuan, a year-on-year increase of 138%; the overseas AI dramas/manhua dramas market size is expected to exceed 4 billion US dollars, accounting for nearly 70% of the entire overseas micro-short drama market.

The market has more than doubled, yet 90% of practitioners have left. Where has all the money gone?

The first batch of profitable companies are on the verge of collapse

The first wave of profits was taken away by production companies, and also vanished the earliest.

In the early days, platforms such as Hongguo had guaranteed projects priced at S-level and A-level. Film and television companies and traditional short drama companies received orders from platforms and then subcontracted them to AI teams for production. At the peak of the market, the production end quoted 1,000 to 2,000 yuan per minute, with a cost of only 500 to 600 yuan.

"In the two or three months around the Spring Festival, the problem was not whether there were orders, but that there were too many orders to handle all of them," Sun Shuche recalled, adding that good orders could achieve a gross profit margin of 60% to 70%.

When money is that easy to make, a flood of people will rush in. Teams from IT, workflow, e-commerce and other fields all poured into this track. Sun Shuche himself used to work in the supply chain and e-commerce sectors, and he felt like he was trading stocks back then: "When all kinds of people start rushing in, the dividend is almost coming to an end."

The production capacity then exploded. In the first half of 2026, Douyin saw more than 221,900 new native AI short dramas, with an average of over 1,200 episodes launched every day, and a new drama born every 20 seconds at peak hours. In contrast, the total number of live-action micro-short dramas launched nationwide in 2025 was only 33,000.

More than 200,000 pieces of content are competing for users, and the quotation on the production side has been falling all the way: the quotation for ordinary assembly line projects has dropped from about 5,000 yuan per minute to a few hundred yuan, approaching the cost line.

On September 6, CCTV Finance reported that since the beginning of this year, the production quotation for standardized AI short dramas has dropped from about 5,000 yuan per minute to a few hundred yuan, approaching the cost line. At the same time, some high-quality customized projects can still be sold at 10,000 to 20,000 yuan per minute.

More fatal than price reduction is the change of platform rules. In the early stage, there was a guaranteed payment, and you knew how much you could get after finishing the work; later, it shifted to revenue sharing — finishing the work only got you a ticket to the game, and how much you could get depended on the play volume and platform traffic.

AI short dramas have changed from "getting paid for receiving orders" to "gambling on traffic to make money".

This just hits the soft underbelly of pure production companies: being able to generate videos does not mean knowing what the audience likes to watch. One script can be reproduced by changing the characters and pictures. AI has increased production capacity, but also brought a flood of homogeneous content.

The result is very cruel. Among the 221,900 AI short dramas in the first half of the year, only 1,055 had a play volume exceeding 100 million, with a hit rate of 0.47%; measured by the break-even line of 50 million plays, less than 1.3% of the works can recover the cost — roughly for every 77 new dramas, only one can barely break even.

The ebb Sun Shuche saw from the returned equipment leases is just the other side of this round of reshuffling. In the past, everyone competed on "whether the cost per minute can be reduced from 2000 yuan to 500 yuan", but now the question becomes: "After you finish producing the work, why would people want to watch it?"

Those who make big money are not making short dramas

"Tools will be used by everyone sooner or later. What is really scarce is good content and good IP."

Yilian Technology has been preparing for this for a long time.

In 2023, the team has started to incubate animation IP accounts, and built several accounts with millions of followers on Douyin, Kuaishou, Xiaohongshu and WeChat Channels.

At that time, content was still produced by original painters and animation directors. Later, AI video tools such as Pika, Runway and Sora came out, and they integrated AI into the original production system, reaping the first wave of efficiency dividends.

But when many teams were still recruiting people, expanding machines and competing for production capacity, they had already started to look for high-quality IP resources.

The team later introduced shareholders and partners with film and television, radio and television backgrounds, and established a joint venture with a traditional film and television company, hoping to make up for the shortcomings in content assets.

China Literature is also a typical case. After AI reduces the cost of film and television adaptation, a large number of online novels that were difficult to be adapted for screen in the past have been made into AI manhua dramas.

In the past, to shoot live-action content for a mid-tier novel, you needed actors, photographers, venues, costumes, props, makeup and post-production. After calculating the cost, many IPs had no adaptation value at all.

After AI cuts down this cost, the content that was previously shelved in the copyright library can be redeveloped.

In the past, shooting capabilities were the valuable part, but now the content assets themselves are being re-priced.

Similar changes have also taken place for traditional film and television talents.

Sun Shuche found that the industry was originally pursuing "how many minutes of content can be generated in one day", but now directors, editors and art designers are being invited back to the industry.

The reason is that the pure tool-driven production process easily hits the quality ceiling. You can produce a work for only dozens of yuan per minute, but problems such as plot holes, unreasonable camera shots, poor character performance and narrative defects will follow.

"How directors express ideas, how shots are designed, how stories are narrated — these cannot be completely replaced by a set of workflow."

This change is no longer limited to short dramas.

On August 31, the AIGC long drama *The New Journey to the West* premiered on Hunan TV's prime time and Mango TV. Its highest real-time audience rating reached 0.3384%, ranking first among provincial satellite TVs in the same time slot.

Sun Shuche told Pencil News that many short drama companies have also expanded their tracks from Hongguo to iQiyi, Youku, Tencent Video and even the cinema market.

In his view, AI short dramas are only the first test field for AI to enter the film and television industry.

If AI really enters the fields of long dramas, advertisements and even movies in the next step, the industry will no longer calculate "how much cheaper it is to produce one minute of content".

Instead, it will consider: can content that used to cost millions or tens of millions of yuan to produce be reproduced with a different cost structure?

This may be a much bigger business than mass-producing low-cost short dramas.

Making money by going overseas? That's easier said than done

Sun Shuche's current judgment is: companies that only make money by charging production fees for producing AI short dramas for others are facing very high risks. There are roughly three types of companies that can still make profits:

The first type has capital, high-quality IP and professional film and television teams, producing high-quality content; the second type holds stable long-tail IP, whose characters already have fixed audiences, so they can continuously develop sequels and derivative works; the third type does not pin all their hopes on the single income of "producing dramas".

Yilian Technology is taking the third path. Sun Shuche no longer defines the company as an "AI short drama company", but an "AI content company": in addition to producing dramas, it also produces advertisements, e-commerce materials, incubates film and television IP accounts, AI artist accounts, AI images for real artists, as well as trendy toys and e-commerce live streaming content. Its revenue is divided into six lines: drama production, agency operation, advertising, distribution, tools and computing power.

"Only with the joint support of multiple business segments can the company move forward in a relatively healthy state."

This also explains a contrast: the quotation for ordinary assembly line AI short dramas has dropped to a few hundred yuan per minute, while high-quality customized works can still be sold at 10,000 to 20,000 yuan per minute. The price difference does not lie in computing power, but in the fact that customers are willing to pay for creativity, brand understanding, directorial capabilities and final play conversion results. What is most easily reduced in price by AI is the "execution work", and what is hardest to be reduced in price by AI is the "final result".

The financial reports of leading companies have already confirmed the flow of money. Kunlun Tech's short drama business revenue in the first half of the year was 1.534 billion yuan, a year-on-year increase of 163%. More than 90% of the new content launched by its DramaWave platform in July was generated by AI; Duiba's AI short drama revenue in the first half of the year was 223 million yuan, accounting for 51.2% of the company's total revenue.

Going overseas is another big cake, but Sun Shuche poured cold water on this idea first: "Going overseas is not as simple as translating domestic short dramas into English and posting them on TikTok." When Yilian Technology produces overseas content, it basically rewrites the scripts and re-produces the works according to the local market.

The overseas market is also facing fierce competition. According to the actual situation Sun Shuche's team encountered, the revenue of some native TikTok content per 10,000 plays has dropped from about 8 US dollars to more than 2 US dollars. If you develop your own short drama app and rely on user acquisition to drive paid traffic, you will also face the rising customer acquisition cost.

But teams that have formed a closed loop of "content-traffic-payment" are indeed making money. NetShort, under Malt Media, achieved about 120 million US dollars in in-app purchase revenue in the first half of the year, doubling year-on-year, with about 130 million downloads, ranking second among overseas short drama apps; its downloads in a single month in July exceeded 23 million, topping the overseas short drama app download list.

The difference is: producing a drama for others can only get you one production fee; if you own the app and users, you can continuously earn revenue from in-app purchases, subscriptions and advertisements, and guide the same group of users to the next piece of content.

One direction is to develop long-form content, the other is to go overseas, but the profit logic points to the same direction: pure AI production capacity is getting cheaper and cheaper, and the links closer to content and users are more valuable.

Sun Shuche himself is the epitome of this round of change — from earning 60% to 70% gross profit with a cost of 500 to 600 yuan per minute, to actively leaving the business of "only charging production fees".

Of course, there will be people making money in this 400-billion-yuan big market. Since September 1, the *Measures for the Administration and Development of Micro-short Dramas* has been officially implemented, which requires mandatory labeling of AI-generated content. The era of crude mass production is over. In the next stage, money will more likely flow to people who master IP, professional content capabilities, distribution channels and user resources, rather than companies that only compete on how many more minutes of video they can generate per day.

This article does not constitute any investment advice.

This article is from the WeChat official account "Pencil News", author: Song Ge, published with authorization from 36Kr.