Original AI short drama’s myth of getting rich, the fever subsided in half a year

“When the production quality returns to benign, your production accuracy will become a scarce resource that others value.”

A numbness and an expectation. Behind the two opposite expressions is the same wave of price cuts.

The free fall of a price curve

Forest Pictures, a company with a traditional film and television background, will take over a national AI project with a historical theme in October 2025.

“The quotation at that time was very impressive, close to 10,000 to 15,000 yuan per minute.” During the same period, the common quotation in the market was 3,000-5,000 yuan per minute. The reasons for the high price are obvious: historical themes, 1:1 reproduction accuracy, higher than the level of ordinary AI comics.

However, this price system did not last for one year.

At the end of 2025, the employer approached us to make a 2D comic series, and the quotation had dropped to 1,000-1,500 yuan per minute, and foreign suppliers offered 400-1,000 yuan; the 3D and simulated human projects dropped from 3,000-5,000 yuan to 1,000-2,000 yuan.

“By this year, my friends told me about the automatic flow workflow platform, which can be generated with one click and even quoted dozens of yuan per minute – which is already outrageous.”

Outrageous, but true. Li Juncheng, the founder of Yujin Wuliang, also had the same experience: at the beginning of the year, the market quotation was still 2,000 yuan per minute, but now it is 1,000-1,500 yuan—halved in half a year. “There was no particularly severe fall. It continued to fall, falling by several hundred each time.”

Please pay attention to the wording of the founder of the leading label: it is not “shocked”, but an almost dull calm, which is exactly the same as Chang Wenbo’s feeling.

This numbness is not dullness, but muscle memory. They have all been working in the live-action short drama era. They have seen Douyin’s policies repeated, they have seen production fees cut in half overnight, and they have seen their peers build and collapse high-rise buildings in half a year.

But it was precisely this kind of numbness that allowed them to see the end result earlier than others: because “anyone can do it”, “the cake will definitely be divided into many small pieces”; the threshold is so low that anyone can start their own business, and companies that rely solely on production to make a living are destined to have no way of survival.

“The current quotation, I thought it would take a year, but I didn’t expect it to drop to this level in half a year.” Regular fulfillment does not constitute news to an insensitive person.

Why can the quotation fall all the way below the cost line? The answer lies in the rewrite of the cost structure.

Feng Shao has dismantled the real accounts of a 2D comic series: the cost is 80,000-100,000 yuan, the duration is about 120 minutes, the script accounts for 10%-15%, the computing power accounts for about 30%, and the remaining 60% is manual-video generation, post-production, packaging, and dubbing are all included.

The structure of batch dramas given by Li Juncheng is another world: a script costs several thousand yuan, computing power accounts for 50%-60%, and manpower is compressed to 10%-20%.

When the two sets of figures are put together, the essence of this round of price reduction becomes clear: when the automatic flow connects the workflow: the novel generates a script with one click, the script generates storyboard prompts with one click, and then converts it into pictures and videos. The screenwriting, storyboarding, and post-production processes that originally accounted for 60% of the budget were completely swallowed up by a canvas-style platform. The remaining major part of the cost was computing power.

Therefore, the sharp drop in the price of the AI ​​short drama is not vicious bidding, but the anchor of pricing has been switched from “human hours” to “model token”. Prices can only sink along with model capabilities, which also explains why the cost line itself is constantly moving downwards.

Not only that, the 9.9 yuan AI course and the 3-day guaranteed teaching and monetization course have sold tens of thousands of copies on major platforms. With a computer and a set of AI tools, short drama production that once started with a million yuan can now be completed by ordinary people.

DataEye Research Institute predicts that the market size of AIGC video models in the field of AI dramas and comics will exceed RMB 20 billion in 2026. In the first half of the year, the total number of AI drama and comic-related practitioners and individual creators has exceeded 500,000. “Everyone can create” is rewriting the boundaries of the film and television industry.

A 120-minute AI short drama can be delivered for as little as a few thousand yuan, which is equivalent to dozens of yuan per minute: it’s not that the price is crazy, it’s that the value of “people” has been evacuated by technology.

The other side of 90% loss: the money doesn’t disappear, it just changes pockets

Although prices have dropped by 90%, demand has not disappeared.

QuestMobile data shows that the number of users in the short drama industry has reached 851 million in May 2026, of which Hongguo’s free short drama monthly active users are 356 million, nearly double year-on-year, and the average daily usage time per person per month exceeds two hours, leaving traditional online video platforms behind.

The users are still there, the attention is still there, so where did the money go?

Let’s look at the production side first. In the first half of the year, 221,900 AI short dramas poured into Douyin, and only 1,055 of them exceeded 100 million views. Calculated based on the break-even line of 50 million views, less than 1.3% could make back the money – about every 77 new dramas, only 1 barely broke even.

Chen Wei, head of AI technology at China Literature Group, said similarly: Only AI short dramas that have been played more than 40 million times have a chance to make a profit, and such “hot hits” account for less than 1%.

Li Juncheng’s judgment is more direct: “There are hundreds of thousands of AI dramas on the market now, and 90% of them are losing money.”

Feng Shao has a sample that condenses the industry cycle: a partner launched a 2D drama in April with a cost of less than 100,000 yuan, and the revenue was more than 200,000 yuan, and the profit margin doubled; so it increased production and expanded production, and by June, the new drama did not even have a few thousand yuan in revenue.

It only took two months from huge profits to losses – the speed of supply influx was faster than anyone could learn arbitrage experience.

Looking at the distribution side, money mainly flows into two pockets: the short drama platform and computing power.

The steering on the platform side is particularly interesting.

Feng Shao recalled that when we were producing high-quality live-action dramas in April last year, Hongguo still fully financed the production fee, and the volume of a high-quality drama was close to 2 million yuan. In the first half of this year, the “Thousand Production Plan” was launched, and the guarantee for a single film was reduced to 200,000-300,000 yuan. In the latest matching plan, the guarantee was further reduced to 130,000-220,000 yuan, accounting for less than 10% of the production cost of a high-quality drama.

“The policy is still there in name, but there are fewer and fewer projects that can pass the meeting. This amount is friendly support.” Feng Shao said.

Blue Whale Technology’s previous investigations corroborate this: In July, Hongguo switched the Qianbu plan from a guaranteed mode to an investment matching mode. Under the regular matching mode, the platform no longer provides guaranteed deposits for production, and the producers only receive 25% of the turnover income.

The accounting side is equally opaque.

Feng Shao told Business Data School that the corresponding account share for 10,000 plays of a high-quality drama is only about 5-10 yuan, and the calculation formula is the “black box” of the platform. “After a play is finished, if you can’t get back your original work, you really don’t have a clue in your heart.”

According to the Securities Times citing data from DataEye Drama, a cruel scissor gap has taken shape: production costs have dropped by about 90%, but traffic costs have increased by more than 100% year-on-year, and revenue per thousand plays has dropped from about 60 yuan in the second half of 2025 to 15-30 yuan.

It’s cheaper to make and more expensive to be seen. This wave of price cuts has washed away companies that “only know how to produce”.

As Chang Wenbo judged: “In the era of AI short dramas, production companies have no way to survive.”

But there is always another side to the coin. There is an easily overlooked detail in CCTV’s financial report: while the standardized quotation has fallen below the cost line, the customized short drama business for specific needs can still be sold stably at 10,000-20,000 yuan per minute, and the quotation for brand-customized works of about 20 minutes can reach 200,000-300,000 yuan; the annual salary of compound talents who understand both AI and film and television is close to one million yuan.

Feng Shao also observed that the salary of AI directors in the industry is rising against the trend.

The pricing of Ice and Fire shows that the market does not refuse to pay for AI content, but only refuses to pay for “production without scarcity.”

This also explains a common move of the two interviewees: neither of them nailed themselves to the status of “controller”.

Where does the skit go after the shuffle?

Whether it’s Li Juncheng or Feng Shao, the bulk of their company’s profits is not short plays, but IP.

Feng Shao positioned Forest Pictures as “long-term operation of IP”, and the short play was just a stepping stone: after winning multiple awards within the State Administration of Radio, Film and Television with a short play on the restoration of intangible cultural heritage and being translated into 8 languages ​​and promoted to more than 200 countries, the company successively took over judicial and cultural protection projects from the Supreme People’s Procuratorate Film and Television Center and cultural tourism projects in Datong, Shanxi. The next step is Jingdezhen Ceramics and the ancient village of Songyang in Lishui, Zhejiang.

In itself, a high-quality drama with a sales volume of 2 million can make back its investment in one or two months if it is a hit. If it is not a hit, the results will be known in the first month when the data is released.

Therefore, the real value of high-quality dramas is not in the division of accounts, but in awards, labels and subsequent business. In a market where 10,000 views are only worth 5-10 yuan, this may be the most rational self-rescue for the producer.

Li Juncheng turned around earlier and more thoroughly.

Today, Wuliang still presents itself as a production label to the outside world, but at its core it is a copyright company – it holds its own IP such as “What a Good Girl” and mainly engages in copyright cooperation rather than purely accepting orders.

He also predicted something further: As the quality of AI generation approaches that of real people, the licensing of portraits of purely new actors has lost its meaning. “The agency wants to license new actors to us for free, but we don’t need it – why don’t I generate a new actor myself?”

In his opinion, species such as actors, card draw artists, and pure production companies will disappear one by one.

Certainty is emerging both on the policy side and on the overseas market side.

On the policy side, the “Management Measures for the Development of Micro-Short Dramas” was officially implemented on September 1, and the State Administration of Radio, Film and Television’s “Management Tips (Classification and Stratification Standards for AI Micro-Short Dramas)” was implemented in July, with hierarchical management based on the amount of investment.

Administrative power is drawing the boundaries for barbaric growth – this is precisely the way to support quality.

On the market side, data from the China Network Audio-Visual Association shows that the number of live-action dramas released during the 2026 Spring Festival is about 1/50 of that of AI dramas, but the total broadcast volume is 25 times that of AI dramas. The premium of “liveness” cannot be replaced by AI in the short term.

At the same time, the floodgates to the sea have been opened. The industry predicts that the scale of overseas micro-short dramas will exceed US$4 billion in 2026, with AI content accounting for nearly 70%. A joint report between Adjust and Sensor Tower shows that the number of overseas short-form drama app installations increased by 238% year-on-year in the first quarter, and the number of sessions surged by 679%.

However, the golden age of going overseas may be shorter than imagined.

Feng Shao went out and watched it, and found that a large number of overseas AI short dramas “still have the same content in Chinese, but they are placed overseas.” There are even arbitrage strategies of directly converting live-action dramas into AI dramas.

Li Juncheng pointed out another layer of reality: “Seven to eighty percent of the short drama platforms in the world are Chinese companies. If you go around and around, you will still be involved in the channels of Chinese companies.”

Although the industry’s income ceiling has been raised, the distribution structure has not changed – computing power and platforms are still the two most profitable ends.

結語

Feng Shao gave a vivid summary: long-form dramas and movies have developed for a hundred years, live-action short dramas have completed the same band in two years, and AI comic dramas have completed their development in less than a year.

One conclusion is that the acceleration of technology has compressed the content industry’s “upslope-peak-homogenization-shuffle” cycle into shorter and shorter periods. But why not, technology has made the original content “fast food” more “fast” and “cruel”.

This is also the most essential coordinate to understand this round of price drops: the price has dropped by 90% in half a year, and what has dropped is not the industry, but the premium of the three words “know how to produce”; what has been washed away by the 90% loss rate are players who use AI as a traffic arbitrage tool.

When the supply is no longer scarce, the scarcity returns to where it should have been – good scripts, good aesthetics, IP that can operate in the long term, and the platform ecosystem itself that holds the right to distribute traffic.

Related Reading

• More people are sending AI to fight their customer service battles2026-09-25
• Why “exclusive franchise” is the choice of the times2026-09-25
• Who will stop the cyber flood from flooding the world?2026-09-25
• Who needs a pair of AI glasses during Golden Week?2026-09-25
• Original Muxi shares staged a tens of billions of hands changing drama2026-09-25