(This article is authored by Reliable A-Xing and published with authorization from Titanium Media.)
Article by | Reliable A-Xing
At midnight, you decide to finish this episode and then go to sleep:
The groom skipped the wedding to head to the airport to pick up his long-lost love, while the bride, in a huff, sent the delivery guy who’d come to run errands to fetch the groom—and even dragged the fake groom along to register their marriage. Little did anyone expect that the delivery guy’s hidden identity was actually the long-lost heir of the Pei Group. Before you knew it, the story had stretched over 80 episodes—turns out they were two childhood friends who’d been separated for years.
In another drama, the girl starts out as a poor student while her older sister is a top scholar. Unfortunately, the whole family dotes on the sister and mocks the younger one—little do they know that the younger sister actually scored 708 points and secretly enrolled at the National University of Defense Technology. Amidst the family’s ridicule and scorn, the girl even loses her boyfriend to her sister...
Scrolling through it still wasn’t enough—a kind-hearted woodcutter went up the mountain and rescued a white dragon. In gratitude, the dragon gave the young man a piece of jade and told him to buy the distant mountain that had already been hollowed out by quarry workers. The young man exchanged 500 taels of silver to renovate his house, purchased the mountain, signed the land deed, and thus began a career that would make him as wealthy as an empire.
......
I couldn’t help but binge-watch all three short episodes; it’s already 2 a.m.
You know full well the plot is melodramatic and it’s time to go to bed, yet your fingers keep sliding uncontrollably to the next episode—each one even forcing you to endure an 8-second ad.
Now, short-form dramas are indeed very different from what they used to be.
From the overbearing boss falling in love with me, to being picked by a wealthy man after losing my job and raising a child, to the shocking mysterious figure—Master—who is actually just some despised son-in-law—these outlandish plots and generic templates have gradually begun to finely segment their audiences, crafting tailored character-driven narratives for different types of ordinary people. Though most of them still follow the same formula: the protagonist has been bullied, oppressed, wronged, subjected to injustice and discrimination, then stages a comeback, hides their true identity, and finally delivers a jaw-dropping twist that blinds all those who looked down on them—this remains the defining hallmark of short-form dramas.
The audiences drawn to these viral short dramas cannot fully adopt a critical perspective—simply because the audience is so vast.
According to data released at a September 17 press conference hosted by the State Administration of Radio, Film, and Television: In the first eight months of 2026, 430,000 micro‑short dramas were launched nationwide—13 times the total for all of last year—with AI‑generated series accounting for over 90% of the total. The national user base for micro‑short dramas has reached 851 million, representing 66.7% of the overall internet population. By 2025, the micro‑short drama market is expected to officially surpass RMB 100 billion in size.
During this year’s Spring Festival movie season, from Little New Year to the seventh day of the first lunar month, the Hongguo Short Drama app recorded a cumulative total of 600 million views, contributing 19.3 billion total views—by comparison, nationwide movie attendance during the same period was 187 million. Hongguo Short Drama’s daily active users have surged to 168 million, up 107% year over year, with an average daily session length of 125 minutes per user.
Thus, short-form video content has become a nationwide pastime: young people use it to pass the time during their commutes, middle-aged individuals rely on it to unwind after work, and seniors fill their post-retirement afternoons with it.
This marks an explosive surge, coming after AI short-form dramas have already decisively outperformed the entire live-action short-drama industry chain in terms of consistency. Meanwhile, at the other end—driven by the satisfying tropes of AI short dramas and the very soul of their narratives—are real people who create addictive content and find themselves deeply caught up in the industry’s whirlpool.
During this period, the WeChat official account “Reliable A-Xing” has met with friends working on AI short dramas and conducted on-site visits to AI short-drama teams of various sizes. I’ve found that AI has indeed lowered the barrier to entry for creating short dramas and, at its core, has either upgraded or reshaped the entire industry chain.
I. Feng Ge’s blockbuster training course: Each episode is an “emotional shock”
I was referred by my friend Feng, who’s a rising star and the head of a hot‑trending AI short‑video production company. At first, he thought I was in the same industry, but I told him I’m just really curious about this field because my parents have become obsessed with it. As for me, I don’t have a strong opinion—nothing against it, but I’m certainly not yet hooked either.
Brother Feng saw that I was straightforward and that we were around the same age. He used to work as a TV director, then moved into short-form video, and now he’s started his own business in the AI‑driven short‑drama space—and he’s already reaping results. He even showed me some of his hit projects, one of which is the story I shared at the beginning of this article. So he invited me to attend one of his new‑employee training sessions, which also doubled as an interview for me—saving time, since he’s really swamped these days.
I arrived at his company’s office building in Hankou’s Wushang Plaza as agreed. After exchanging a few words, I sat in the corner and listened to his training session. There, he was drawing densely packed flowcharts on the whiteboard, with the title “Script Breakdown.”
Brother Feng, wearing a white baseball cap, spoke in an uncompromising tone: “Remember this—short-form dramas are different from TV series or movies. If we tell too long a story, viewers will spend no more than one minute on us. What I aim to do is manage their dopamine levels.”
Brother Feng tapped the whiteboard with a marker. “A long-form series has 80 episodes, each 90 seconds long, totaling 120 minutes. What you need to do is break those 120 minutes down into 80 ‘emotional current points’ right on the script.”
He wiped the outline on the whiteboard and drew a wavy line, marking each key milestone:
Episode 1 (The Golden Three Seconds): Right from the opening, it delivers explosive revelations and intense conflicts—public annulment of an engagement, humiliation by a wealthy family, a terminal diagnosis, a scene of catching someone in the act, and a spirit‑summoning séance that unveils hidden conspiracies. If the first seven seconds fail to deliver a “suspense hook + emotional climax,” user retention plummets to 41%—the fate of the episode hinges on this moment! It must evoke powerful empathy.
Episodes 3–5: The First Twist—Have the user follow the protagonist’s inner monologue, choosing to either awaken, have the female or male lead’s identity exposed, witness a benefactor suddenly appear, or be framed. At this point, draw the user into the story arc.
Episode 10: The First Minor Climax — The villain gets their first taste of defeat, but immediately plants the seeds for an even bigger crisis.
Episodes 20–30: Mid‑story tension—misunderstandings, separations, suppression, third‑party interference, and the introduction of new variables continuously build a sense of frustration, pushing viewers’ emotions to their lowest point.
Episode 50: The moment the truth is about to be revealed cuts the scene; the antagonist begins to face punishment, but how exactly that punishment unfolds is left for the protagonist to decide. After all, the hardest thing for humans to resist is the state of “already knowing half.”
Episodes 75–79: Foreshadowing the ultimate twist, each episode ends right at the tipping point—just as the door is about to open or just as the words are about to be spoken.
Episode 80: The series finale has aired, but it had to leave an open-ended cliffhanger to set the stage for Season 2.
“At the end of each episode,” Brother Feng said, “the audience must feel a kind of physical discomfort—like when you’re halfway through a sneeze and someone covers your mouth. They have to click on the next episode just to ‘finish that sneeze.’”
An audience member asked: “What if viewers lose interest halfway through and exit without scrolling down? How do we handle that?” Brother Feng smiled and replied, “That’s why we have a strict requirement for ‘peak‑moment density’—there must be an emotional high every 15 seconds, and each episode must deliver at least one twist. This pacing precisely triggers the brain’s reward mechanisms; watching continuously for 30 minutes boosts dopamine levels by 47% compared to finishing an ordinary movie. You’re not fighting against your audience’s willpower—you’re actually working with their neural circuits.”
Someone whispered, “It doesn’t sound like creation—it’s more like neuroengineering.” Brother Feng tapped the whiteboard. “To be precise, it’s emotional engineering. That’s the industry’s standard answer; you may not agree, but the data doesn’t lie.”
Feng Ge went on to say, “We spend every day immersed in short-form dramas—especially those that everyone finds particularly compelling. We carefully analyze them and then consider how we’d write the script ourselves. To help everyone better understand this industry, we upload screen recordings of popular short dramas directly to our company’s shared cloud drive, and we transcribe all the dialogue, meticulously studying and analyzing it line by line. Once we’ve collected enough similar scripts, we can start identifying recurring patterns.”
Finally, Brother Feng said that short-form dramas are even more entertaining than TV series and movies, emphasizing that we’ve caught a great era—this is a golden opportunity. As long as you master this skill, finding a job afterward will be a breeze! This is what truly allows ordinary people to turn their lives around in today’s world. At this point, six or seven students burst into enthusiastic applause—and out of politeness, I joined in too!
I took a quick tour of the office with Brother Feng. The space is about 40 square meters, and it’s mostly equipped with desktop computers—actually, the setup is even simpler than that of a livestreaming team. Clearly, the focus here is on the AI‑powered content industry. Brother Feng emphasized that the heart of short‑form dramas lies in story selection, and he hopes we’ll have the chance to collaborate and refine some great scripts together. I admitted my sense of online trends isn’t strong, but he replied, “You’ll pick it up as you go!”
II. A Song: A Screenwriter Who Has Shifted from “Writing Scripts” to “Building with Blocks”
A Song works as a planner at a content studio in Guanggu. At 28, he graduated from the Chinese Department of Central China Normal University. He had originally hoped to become a teacher, but after being assigned for several years to teach in a remote rural area—where the salary was extremely low and he remained unmarried—he decided to switch careers. His writing skills are truly excellent, and he has published several online novels.
Last year, an editor at Changjiang Literature and Art Publishing House—though without a formal position—saw an opportunity to switch careers and entered the AI short‑drama industry, where he now writes scripts. I heard that he was introduced by other editors at Changjiang Literature and Art, and he even published a book. Because I became interested in this field, he took me fairly seriously. I invited him for coffee at the Starbucks downstairs from his company, and we ended up talking about his work.
I was pretty direct and asked about his income right off the bat. To my surprise, Asong’s base salary isn’t high—just 8,000 yuan per month. He earns a commission of 10 yuan per episode once each episode surpasses 100,000 views, and if a single series hits over 10 million views, he receives an additional 500-yuan creative incentive bonus. It left me utterly stunned. So it turns out that even with huge traffic, short-form videos don’t pay much at all!
Last month, a “reincarnation novel” he casually rewrote unexpectedly became a minor hit, earning him 12,000 yuan in base salary plus bonuses—his highest monthly income since entering the industry, which kept him thrilled for an entire week.
But then he plunged right back into the next “remake script,” telling me, “Blockbusters are a matter of luck; the assembly line is where the real money is.”
This is indeed understandable, as he doesn’t come from a screenwriting background and has never even worked in the film or television industry. His grasp of content creation relies entirely on the traffic generated by several of his online novels, which was decent enough to get him hired for a trial period. However, he’s also desperately short of money, eager to save up so he can get married sooner, and determined to prove himself.
Every day at work, he basically uses DeepSeek and DouBao to write AI short dramas. The keywords he inputs are usually related to the AI short dramas he’s been following lately. After distilling these keywords, he changes the character names, settings, identities, historical periods, as well as the sources of tension, grievances, and the dynamics of dominance–submission, then asks the AI how to add more satisfying plot twists and narrative tropes.
Ten minutes later, a 10,000-word draft of the script appeared on the screen. Clearly, he had completely forgotten his dream of becoming a writer after graduating from the Chinese department. He said, “Those who dive into AI short dramas have no dreams—only an insatiable hunger for traffic!”
The rest of his job is to “trim” the AI-generated draft within this week into a format that meets the standards for 80‑episode hooks, which means he has to write scripts for at least 20 episodes every day—already pushing the limits of collaboration and fine-tuning with the AI. He’s essentially a paying member of most domestic large‑scale AI models. The deadline is Friday. A Song calls this “module assembly”:
Module A: Frustration (being humiliated, misunderstood, or abandoned)
Module B: Reversal (Identity Revealed, Abilities Awakened, Benefactor Appears)
Module C: A Stunning Reversal (The Villain Kneels, Everyone Is Shocked, and the Ex Regrets It)
Module D: New Suspense (Greater Secrets, More Dangerous Enemies)
“Today we swap ‘the richest man’s daughter’ for ‘the god of war returns’; tomorrow we replace ‘mother-in-law’ with ‘a vicious best friend’—the core remains exactly the same,” said A-Song. “This is essentially assembly; you could also think of it as skin-changing. Users don’t care about the skin—they care whether the payoff lands on time.”
A Song has to produce four scripts a month, but his pace is already quite slow—since the team can only release one script per month, his income remains relatively low. If he were to ramp up his workload and generate three storylines at once, assigning each to a different large model for processing, then painstakingly crafting hooks and satisfying plot points one by one, he’d need to deliver more than three scripts per month. Moreover, if the content were too heavily plagiarized, it would certainly fail review, wasting the company’s investment in post-production. How can he strike a balance between quality and quantity? “Right now,” he says, “the only option is to practice more and get better.”
I asked him whether he wanted to create works with high literary merit, and he replied with a mix of mockery and disappointment: “That’s because you don’t understand this industry! Short-form dramas are like fast‑grown chickens—faster even than KFC’s standardization. The kind you’re talking about is more like raising a phoenix.” He went on to add,
“To create a high-quality AI-driven series, the script itself requires painstaking refinement, demanding immense creative effort. What’s more, post-production costs are exceptionally high—whether it’s art design or animation production, the expenses for tokens and computing power are ten to twenty times those of a typical binge‑watchable drama. Even more critical is that no one can guarantee viewership; time waits for no one, and everyone is racing to meet deadlines—otherwise, competitors will leave them far behind!”
III. Lele: Six faces, two thousand dramas, a monthly salary of five thousand
Lele works as an editor at an AI short‑video workshop in the Hanyang Cultural and Creative Park. He has a college diploma and earns 5,000 yuan per month. The workshop employs just three people, including the owner, and focuses on an ultra‑low‑cost model. As a post‑2005 generation entering the workforce, he doesn’t really have much career planning—low pay isn’t a big concern since he’s just graduated. He simply appreciates the convenience of working close to home. However, he didn’t expect that his previous mentor, who had trained him, quit because the salary was too low. As a result, Lele now earns only a modest monthly income and often has to work overtime. After four months of grueling training, he has already become an experienced short‑video editor.
Rather than editing, it’s more like AIGC work; the main editing tool is CapCut. The footage left by the previous editor is mostly stored on the computer, and no matter how much he cuts and reassembles it, it always features the same “old actors”—that is, those ubiquitous AI-generated faces commonly seen in short dramas, each corresponding to one of six classic character archetypes: the domineering boss, the warrior god, the innocent young beauty, the villainous supporting character, the adorable child, and the nurturing mother.
These six faces “live” in different scripts every day—on Mondays, they play the jerk; on Tuesdays, the deeply affectionate lead; and on Wednesdays, they become the elder. “Together, these faces have collectively ‘acted’ in over two thousand short dramas.”
Currently, production is driven by cost control, so they don’t use high‑precision models or custom facial designs; instead, they rely on six fixed faces that are cycled through. All scene assets are sourced from public libraries, and for an 80‑episode series, computing costs are kept under 5,000 yuan. The bulk of the expenses, however, comes from token fees—essentially, the cost of topping up credits across various large language models.
Lele’s workflow is to feed the finalized script, word for word, into the Jiemeng AI survival animation tool, then use CapCut for editing. Once an episode is edited, it’s sent straight to the copywriting team for verification; after they confirm there are no issues, it’s published right away. In fact, this “quality-check” step is pretty much optional, Lele says:
“Sometimes I find that AI can’t even tell the front from the back of a face; in some cases, the text in the image is completely wrong, and the emotions are off as well. At times, it’s enough to make me want to cry—but then I think about it: no one really cares anyway.”
I asked Lezai, “Don’t you think that’s a bit hasty?” He replied, “With 20 episodes released in a single day, as long as the audience keeps watching, that’s good enough. Right now, the boss prioritizes efficiency and volume—fine-tuning is tough.” He added, “Besides, the boss only gives me five thousand yuan a month, and he doesn’t cover much of the token expenses either; I still have to reimburse those myself!”
IV. How can short-form videos generate revenue?
Of course, the platform is a surefire money-maker! Once it captures attention and drives traffic, advertisers will pour in. Strictly speaking, short‑form drama creators are essentially employees of the platform.
After conducting some research, I’ve found that AI short-form dramas are currently benefiting from the traffic dividends previously enjoyed by the live-action short-drama industry.
Live-action short dramas have already mapped out the target audience, narrative tropes, traffic‑driving strategies, and distribution platforms—but the costs are prohibitively high. Producing a live‑action short requires paying actors, hiring makeup artists, lighting technicians, and post‑production teams, plus covering venue fees and script‑rights expenses, bringing the cost per episode to around 100,000 yuan. By contrast, AI‑generated short dramas can be completed with just a single script and basic editing; in some cases, a single creator armed with large‑language‑model tokens can handle the entire process. This slashes production costs to roughly 10,000 to 100,000 yuan per episode—down by a factor of ten to a hundred. As a result, only the leading live‑action short‑drama studios that have already proven their business models can survive, while many practitioners in the short‑drama industry have been forced to shift to AI‑powered short‑drama production.
The biggest advantage of AI short dramas over live-action ones is that they’re free, and high-quality short dramas now actually rival the visual quality of feature films.
Meanwhile, the primary revenue for AI short-form dramas is actually sustained through platform‑based ad‑revenue sharing: the higher the traffic, the greater the ad‑revenue per thousand views, which can fully offset the production costs of these shorts and, on this basis, generate additional income from ad‑revenue sharing.
Of course, by 2026, the platform’s policies had already undergone a subtle shift. Douyin Group allocated over RMB 1.5 billion in guaranteed‑support funding for live‑action short dramas, raising the natural‑traffic revenue share for paid live‑action content from 70% to 80%, while lowering the revenue‑share ratio for purely AI‑generated short dramas to 40%. Meanwhile, Hongguo Short Drama also abolished its series‑production scale coefficient in October, no longer encouraging the “relying on sequels to boost views in exchange for subsidies” workshop‑style approach.
Mid-tier AI short-form dramas are already feeling the chill, as the low barrier to entry has led to intense competition among a vast volume of content. Much of this content can no longer rely on organic traction; to become a viral hit, it must invest in paid promotion to break through the algorithmic layers. Otherwise, without sufficient visibility, it’s extremely difficult to achieve breakout success.
“If a series can’t make money and we even have to pay for ad spend, it’s like we’re just working for the platform,” said Feng. “No, to be precise, we’re working for the platform’s algorithm.” Of course, AIGC costs are themselves a form of expense; Feng put it even more bluntly: “We’re essentially working for the GPU.” The computational costs of AI video generation continue to rise. If earnings remain meager, at worst we’ll just work alone—and if that doesn’t pan out, we’ll simply switch careers and take on short‑form content production jobs at companies.
So, right now they’re still racing against time! I hope they’re lucky enough to come up with a few more hit products!
That's all!
(Note: The aforementioned Feng Ge, A Song, and Lele are pseudonyms, as requested by the interviewer.)
For more exciting content, follow TMTPost on WeChat (ID: taimeiti), or download the TMTPost App.