Original · Unique Research · 2026-06-22
Editor's note: This is a deep report on AI short drama going global, based on a panel discussion at the Unique Bloom 2026 Singapore AI Agent Summit. All speaker names, company affiliations, numerical claims, and case studies are preserved as source attributions. Industry data (Sensor Tower figures) are cited as reported. Speaker statements are their own, not independently verified findings.
AI Industry Observation
After AI Short Dramas Hit 1 Billion RMB in a Single Month Overseas,
the Industry's Biggest Anxiety Is No Longer Technology
Is the "Gacha Pull" Era Over?
"
Last entire year, Chinese short dramas going global earned only 1 billion RMB. This year, a single month broke 1 billion.
This isn't growth—it's explosion. And when an industry starts expanding at 10x speed, the most dangerous thing is thinking you understand it.
At the Singapore AI Agent Summit, four industry heavyweights took the stage to discuss "AI short drama going global."
David Wang, FizzDragon Founding Team Member. The word "short drama" doesn't exist in his world—he just put the world's first AIGC film, Madam Cheng I-sao (郑一嫂), into theatrical release. He's playing a different game. A content producer—the kind who isn't afraid to burn money or make feature films.
Sitting next to him is Jony Deng, Minipix founder, former iQiyi overseas business head, a 15-year going-global veteran. David manages "making it"; Jony manages "selling it." He knows all too well: the story of good content dying in distribution happens every day in this industry.
Liang Wei (梁巍), MovieFlow founder. Making films for 20 years, now doing AI video tools with 2 million users. He's the one selling shovels to the gold rushers—and he truly understands gold mining. He didn't force his way in from a tech background; after 20 years in film, he discovered technology can change fate.
Tang Xiangyang (唐向阳), ShanJian Intelligence (闪剪智能) founder. 13 years in short video tools, 300 million cumulative users, now focused on B2B AI digital humans. Note this path: C-end tools → B-end services → AI digital humans. He's looking for the next scaled deployment outlet; short dramas are just one possible destination.
Four people, four coordinates. Content, distribution, tools, platform. Every key link in the industry chain has someone present.
But the problem is: when David talks about "artistic breakthroughs in AIGC film," Tang Xiangyang is thinking about "how digital humans can batch-replace actors to cut costs." When Jony analyzes Southeast Asian market payment habits, Liang Wei's mind is turning over "can my tool let one person achieve what a team does."
Everyone is talking about AI film and video, but everyone sees a completely different landscape.
This is actually a signal of sector fissure. Production capacity is no longer the problem—AI has cut video production costs to the ankle. But after capacity is released, then what? Who pays? What content keeps users? Who holds distribution power?
This quiet fissure is tearing "AI short drama going global" from a single trend into four different battlefields.
"Gacha Pull" Becomes "Stable Output"
A particularly obvious feeling is that the hottest keyword in this industry is completely different from last year.
ShanJian Intelligence CTO Tang Xiangyang summarized this year's biggest change in three words—"determinism." It sounds like the restraint of an engineer, but behind it is the entire industry's leap from "gambling" to "mass production." He recalled that last year, making AIGC content was like pulling gacha—the same prompt run ten times produced ten different results, and the team simply couldn't promise delivery standards to clients. Now? Sora, Kling 3.0, Seedance 2.0, Luma, Haiper—the various foundation model capabilities have converged closely, and "gacha pulls" have become "stable output."
The most direct consequence is that the threshold has collapsed. UGC users can train for a few days and produce finished content; it's no longer the exclusive domain of professional teams.
MovieFlow has been live for only 8 months and already has 2 million users, generating over 5 million minutes of video. This number would have been unimaginable last year.
But after the technical problem is solved, new anxieties immediately emerge.
Production Capacity Is No Longer the Problem—Distribution Is
Minipix founder Jony—who previously ran overseas operations at iQiyi—said the most frequently used word in the conference hall this year is "distribution." In the past two years, the industry was still arguing about "can this thing even be made" and "can production capacity keep up." Now production capacity is clearly not the issue. The problem is: massive content is produced wholesale, and then what? Who watches it? Where? How much do they pay?
Jony raised a thought-provoking point: Netflix's biggest moat was never content production capability, but distribution channels covering 300 million users worldwide. Every language, every country, every cultural taste—it delivers with precision. The production side has been disrupted by AI, but the industry has only just started catching up on the distribution side. Sensor Tower data is telling: in 2025, global short drama app in-app purchase revenue was $2.8 billion, up 116% year-over-year. Sounds lively, right? But the flip side is—80% to 90% of projects can't recoup their costs, and 90% of revenue is concentrated in the single North American market. Content has exploded, but money hasn't rained down evenly.
So the real watershed for the industry may not be technology maturation, but who can fill the gap between "making it" and "selling it."
Inside this gap, new things are emerging.
AIdeo: AI Video May Be an Entirely New Species
MovieFlow co-founder Liang Wei (梁巍)—who has 20 years in the film industry—discussed the concept of "AIdeo" (AI+Video). His point is that AI video is not a derivative of short drama; it's an entirely new species.
Behind this judgment is a complete logic of media evolution: the camera invented film, digital shooting spawned TV variety shows, smartphone ubiquity gave birth to short video and live streaming—every time the underlying technology is restructured, an entirely new content form is born. The new species brought by AI multimodal generation capability is what Liang Wei calls AIdeo.
Whether this framing is completely accurate is too early to judge. But the direction he points to is correct. Deepfake-level image quality has made it impossible for people to distinguish between real filming and AI generation. And when "shooting that looks real" is no longer a problem, the only remaining limit is imagination.
"
AI's powerful capability has for the first time given people around the world 'creative democratization.'
You don't need to be a director, you don't need to be a cinematographer, you don't need to blow money renting equipment and assembling a crew—you just need an idea, and you can turn it into moving images. This change sounds romantic, but for the entire industry, it's both an opportunity and a ruthless reshuffle.
What Cards Do Chinese Players Hold
Jony said something blunt: "In terms of content combat power, Chinese teams are probably the strongest in the world."
This isn't boasting. In production speed, cost control, and adaptability to overseas markets, Chinese content teams have exported their "involution" skills worldwide. A fact many people haven't noticed: the most popular AI short dramas on North American platforms—werewolves, Greek mythology, domineering CEOs—are mostly operated behind the scenes by Chinese people. Chinese teams leverage domestic production capacity while cross-border tapping into other countries' language and cultural resources. Production capacity and demand are, for the first time in a real sense, globally connected.
The most convincing evidence is two wildly different cases.
One is Madam Cheng I-sao. The world's first AIGC theatrical film, from 13 countries, over 130 creators collaborating remotely, many who have never met in person. Completed in 6 months, cost only several hundred thousand USD. In 2025, it was released in theaters in Singapore and Malaysia. This is what "Chinese production capacity + global collaboration" means.
The other is Zombie Scavenger. In Kunming, Yunnan, a 29-year-old guy named Liu Ziyu, vocational school graduate, non-film-trained. 3,000 RMB cost, completed independently in 10 days. Over 70 million views on Douyin. Hollywood producers personally posted saying they wanted to collaborate.
Two extreme cases point to the same signal: what AI gives is not tools, but the right to experiment.
Previously, making a film required tens of millions in investment and film-school credentials; now several hundred thousand USD, or even a few thousand yuan, can produce something memorable. The threshold has collapsed—anyone can enter. And the people best at "making flowers bloom within low thresholds" are Chinese teams.
But holding good cards doesn't mean winning.
Sensor Tower data looks good: 2025 global short drama in-app purchase revenue of $2.8 billion, up 116% year-over-year.
But another set of numbers isn't funny: 80%-90% of projects can't recoup costs, most platforms "increase revenue but not profit." More painfully, 90% of revenue is concentrated in the US, emerging market penetration is less than 10%.
Jony dropped a soul-searching question at the panel: "Everyone making AI short dramas now is probably Chinese teams burning tokens. What's spent is money, what's invested is also money. How do you ultimately monetize and recover?"
This is the classic "production capacity trap." Too many people who can make content, too few who know how to sell it. You have a dragon-slaying blade, but you find there are barely any dragons in the market. More realistically, all the dragons are in North America; other places don't even have a dragon's shadow.
More Dangerous Than Not Being Able to Sell: Thinking You Can Sell
More insidious than not being able to sell is "thinking you can sell."
Tang Xiangyang's judgment is crisp: "Can you just copy domestic hit-making logic overseas and succeed? Absolutely not."
Jony's experience in Malaysia is the strongest footnote to this statement.
Malaysia is an interesting market—Malaysians, Chinese, Indians, and indigenous peoples, four major ethnic groups mixed, official religion is Islam, but English, Mandarin, Malay, and Tamil are all spoken. You take the domestic "domineering CEO chase-wife crematorium" logic and translate it directly? It dies quickly. Jony spent five years doing overseas expansion at iQiyi; he knows this lesson all too well.
So he tried a different approach: not translating Chinese content, but remaking it locally.
He led his team to solidly produce 12 series in Malaysia—all with Malaysian actors, all in Malay, all centered on family relationships, religious beliefs, and community ethics in Malay culture. The themes weren't decided on a whim; they were developed through conversations with locals: what conflicts do Malaysian audiences resonate with, what emotional rhythms they accept, what taboos absolutely cannot be touched. Chinese teams brought production capability and industrialized processes—how to write scripts efficiently, control costs, and ensure consistent quality. But the "DNA" of the stories all came from locally. Result: out of 12 series, 3 were bought by Netflix.
This ratio is high for any content market. Netflix has invested in over 180 local original titles in Southeast Asia; their eye is famously picky. Being selected by them shows that Jony's "Chinese production capacity + local cultural DNA" playbook works.
The logic behind this, as Jony himself summarized clearly: a domineering CEO that keeps domestic audiences binge-watching 50 episodes can become a chicken rib in Southeast Asia; conversely, the generational conflict and community bonds in Malay family dramas don't match Chinese templates at all. The key isn't translation, but "retelling using local cultural DNA."
So Tang Xiangyang said next year's core opportunity is "AI content team localization"—not product going global, but service going global, team going global. You have to be on the ground, or let local people use your productivity tools to tell their own stories.
Where Is the Real Opportunity for Chinese Teams?
So where is the real opportunity for Chinese teams?
Jony and Liang Wei directly said at the panel "we can collaborate." This seemingly polite statement actually releases an important signal: solo content production models are becoming obsolete; systematic global distribution capability is the next moat.
Liang Wei's judgment is bolder: "In the future, most of those truly disrupting the global AIdeo entertainment market should be Chinese people." He even predicts that in the AIdeo era, there will be hundreds and thousands of people who want to be Netflix—China may have 40, the world may have 1000. Because productivity has risen, rich creative output across different cultures will explode at scale.
What does this mean? It means the opportunity for Chinese teams is not making more content, but using China's production capacity advantage combined with deep understanding of overseas localization to upgrade from "content factory" to "global distributor." Whoever can sign local creators in Malaysia, whoever can build local operations teams in Brazil, will survive this elimination round.
Anyone can make content. But delivering content precisely to the right people—that is scarce globally.
What Is the Industry Waiting For?
David Wang said one sentence:
"
The global content industry is waiting for the next Toy Story.
This metaphor is spot on. 1995's Toy Story was humanity's first full-CGI feature film. It wasn't a tech demo—it had Woody's jealousy, Buzz Lightyear's confusion, two toys going from mutual dislike to mutual acceptance. Pixar used a pile of code and pixels to tell a story about "who am I." At that moment everyone understood: the real victory of new technology isn't how realistic it is, but whether it can make you cry.
Thirty years later, the AI video industry is going through the exact same juncture. There are plenty of AI films on the market—dozens of seconds of cool shots flying everywhere—but close your eyes and think back: is there one AI feature film whose character names you remember?
No. Not a single one.
This is why David and his team are making their next feature film, King Gesar. Tibetan epic, the world's longest heroic legend, cost several hundred thousand USD, targeted at global distribution. This isn't a passion project; it's a bet on one thing: using AI to make Chinese people's own stories into good films that the whole world can understand and engage with.
"We hope to create a good story that resonates with audiences worldwide," he said calmly, but I heard the ambition.
What truly moved me was what he said next.
"AI tools have for the first time given ordinary people, or non-film-trained people, the chance to experiment."
The weight of this statement is severely underestimated. In the past, making a feature film required hundreds of millions in capital, film-school connections, and a decade of industry immersion—the threshold high enough to block out 99% of people. In early 2024, the entire industry was still saying "AI can't make feature films at all." David and his team gritted their teeth and made Madam Cheng I-sao, carving a path out of impossibility.
A photographer from a Yunnan town, an epic from the Tibetan grasslands, a group of people who "don't qualify"—these stories strung together point to an increasingly large and increasingly real proposition.
What truly matters in the AIdeo era was never the technology itself. Technology updates, models iterate—today's Sora could be surpassed tomorrow. What truly changes the rules of the game is that the right to create is no longer held hostage by budget and background.
Good stories have never been scarce. What's scarce is the opportunity for good stories to be told.
Now, the opportunity is here.
More Panel Details
Panelists: David Wang (FizzDragon Associate VP of Global Distribution); Jony Deng (Minipix Founder & CEO); Wei Liang / Liang Wei (MovieFlow Co-Founder); Xiangyang Tang (ShanJian Intelligence CTO)
Host: Abner Zhao (UniqueResearch Partner)
Abner Zhao: Hello everyone, I'm Abner. We're sitting here today to discuss the theme of this panel: AI short drama going global, from production workflow to global commercialization. First, please each guest give a brief self-introduction, about one to two minutes. Introduce yourself, your company's products and business, and briefly about going global. Let's start with Jony.
Jony Deng: My name is Deng Liang, Jony. I spent 15 years at iQiyi, and the past five years running iQiyi's overseas business, headquartered in Singapore, focused on Southeast Asian market development. I recently left, also seeing some new opportunities, especially AIGC growth, so the company just launched a month ago. But recently, including these past two days at these AI events in Singapore, and exchanges with peers, my confidence and certainty about AI growth have been getting stronger. So I hope to exchange more with peers going forward. Because my direction is still AI film and video—of course in this video space I'll also make various connections, with producers, content creators, and technology providers. Not just short dramas; more likely IP and interactive products in the future, so I hope to exchange more with everyone.
Wei Liang: Hello everyone, I'm Liang Wei. Our product is called MovieFlow. We're old friends of Unique, very happy to meet everyone in Singapore. I was originally a filmmaker; currently in the AI video field, I'm probably the only one who's spent 20 years in the film industry as a number-one person doing this. MovieFlow has been live for over eight months now; we've been doing going-global for about a year. The product launched overseas from day one; currently we have nearly 1.5 million overseas users, with the major language regions being Brazil, India, Indonesia, and the US. Very happy to chat here about the next major phase of the entire video landscape and how the sector will develop.
Xiangyang Tang: Hello everyone, I'm Tang Xiangyang, co-founder of ShanJian Intelligence. Currently responsible for our overseas product Muta AI. ShanJian has always been doing short video creation tools—like DouPai, ShanJian, FeiTui, Faceplay, etc.—these are all C-end short video tools, with cumulative global users now over 300 million. In the past two years we've started focusing on B-end video marketing scenario creation tools, mainly using AI digital humans as the video creation foundation. Our overseas expansion is currently relatively simple: mainly combining our past C-end video tool and content experience, then finding overseas channel partners to build our overseas product together. That's roughly the logic.
David Wang: Hello everyone, I'm David. I'm at FizzDragon, responsible for overseas distribution of IP. Our company made the world's first AIGC film to enter theatrical release, called Madam Cheng I-sao. Currently we're a company doing AI film, AI advertising, AI education, and creator ecosystem.
Abner Zhao: The four guests just gave brief introductions; you can see their backgrounds are very diverse. In the entire AI film field, David you do content, Xiangyang does tools, Mr. Liang started in content and now makes tools for the entire content industry, Jony does platform and distribution. Next we can discuss AI film going global from everyone's different perspectives.
First question, I'd like all four guests to answer—it has two layers. The entire AI film sector has probably been one of the fastest-changing areas in AI over the past six months. Like from the second half of last year, everyone felt it clearly: initially discussing models, discussing Sora, then Shengshu had a new model PixVerse, now various 2.0 models came out, followed by Grok a few days ago, and Haiper, etc. The entire industry has changed enormously in underlying capabilities, and business models and platform content policies have been constantly changing.
So first question: what do you think is the keyword you most frequently discuss these days? What topics are peers around you talking about?
Second question: going-global short dramas just broke the news that last month's revenue reached 1 billion RMB, while last entire year was only 1 billion. Facing such numbers, what does this mean for industry practitioners? Everyone discuss from their own angle, combined with the first question. Let's start with Jony.
Jony Deng: From a platform perspective, I think the most discussed should be "distribution." Because in the first two years everyone was still discussing production—can this thing be made, is capacity enough. Now it's clear that capacity is completely sufficient. Production isn't the problem; massive content is made wholesale. Then everyone hits another dilemma: made it, how do you monetize? Who watches, where, how much do they pay? Does it have IP value? Or is it discarded after viewing? This is what I see people paying more attention to in the industry. Look at Netflix—its biggest strength is 300 million global users, very broad distribution channels, all languages, countries, cultures. When we're not worried about production capacity, then it's about who can truly, from an IP angle and distribution angle, get this content better in front of consumers. Because consumer attention is limited; ultimately how do you let them filter from massive content to find quality content they like, that suits their taste, even fresh and creative? I think that's what matters.
Then the second question, short drama revenue breaking 1 billion—I think this 1 billion probably mostly came from China. Everyone doing AI now is probably Chinese teams burning tokens. To me, this shows a very massive production capacity; models have no threshold now, basic capabilities are about the same, including Haiper gradually iterating and strengthening. And the whole workflow is getting simpler and more mature. More and more UGC, maybe trained for a few days, can make a finished piece. With this much content being produced, from Sensor Tower data, everyone is consuming—what's burned is money, and what's invested is also money creators and companies put in. How do you ultimately monetize and recover? This connects.
Abner Zhao: Jony's summary keyword is distribution and commercialization. Mr. Liang, I don't know if you agree—is the keyword you discuss also distribution? Any different views?
Wei Liang: Let me combine our MovieFlow itself. We launched last October; I'll combine three product forms to say what the keyword is.
First, facing the C-end market. Before last October, we called overseas models more, including Runway, Pika, etc.—after they appeared in October-November there was a big wave of growth. Entering December, Midjourney started updating v5.2, v6.0 one after another. Entering January-February, Sora-level models started showing signs, then erupted in March, then Luma, Kling 3.0, etc. launched one after another. Over the past eight months, from our C-end MovieFlow.ai (currently over 2 million users, generated over 5 million minutes of video), we found C-end completion is getting higher and higher. So for C-end, I think the most critical word is "improvement" and "rapid realization." We focus on one-click generation for global C-end users, orchestrating it well so ordinary people can quickly make finished films. This is what changes fastest for C-end.
For B-end, we officially launched MovieFlow Studio professional version on May 9th. With this year's Sora-level models appearing, realistic parts are getting closer to live-action. Currently the Shanghai Film Festival is being prepared; we already serve over 50 traditional film and entertainment directors, film companies, and crews who are accepting AI throughout the entire process from prep, shooting, to post-production for cost reduction and efficiency gains. So the real keyword for AI this year, for traditional film, is "entry," "acceptance," and "beginning to realize application." Previously we said this thing can't replace, but now for example Chen Sicheng's "Detective Chinatown" starting shooting next month, and films releasing this month, are all accepting AI use as comprehensively as possible. Because 4K has arrived; when these films complete production in the second half of this year, cinema-level imagery will be achievable. Traditional film大幅 cost reduction and efficiency gains are already happening, including the domestic film companies we serve and Hollywood film companies, are gradually making contact. AI creation is starting to enter professional sectors.
Third, actually following up on what you said about "distribution." We're launching a pure distribution app at the end of this month and early next. We're going global with AIdeo (AI+Video). Because the term AIdeo was also proposed by me at a Unique event last December. AI video—I want to give it a new species name, call it AIdeo. Today when we say AI short drama, it all evolved from Hongguo short drama, Douyin drama. But nobody says AI video must look like that. Just as the camera represents film, digital shooting invented TV and variety, the smartphone brought short video and live streaming, the new species brought by AI multimodal generation is AIdeo. AIdeo definitely isn't only AIGC short drama. The arrival of the AID era means truly entering a new entertainment video era. Now Deepfake makes it impossible to distinguish real filming from AI generation; AI's powerful capability has for the first time given people worldwide creative democratization.
This also connects to the second question—the short drama revenue burst actually means everyone is rapidly entering and accepting AIdeo creation and production. Next entering the AIdeo era, there will be hundreds and thousands of people wanting to be Netflix. Ultimately there won't be just one Netflix; Chinese streaming won't only have iQIYI, Youku, Tencent, Mango. When everyone enters the AIdeo era, China may have 40, the world may have 1000. Because productivity has risen, it brings rich creative output across different cultures.
Abner Zhao: The short drama explosion brings people into the content competition era; the warring states melee era may be here soon. Next please Mr. Xiangyang.
Xiangyang Tang: The first two spoke well, I agree with most viewpoints. Standing in the tool-making sector, let me share my understanding. My three words are "determinism." Before Sora and other new models came out, we did C-end products or AIGC with a core play called "gacha pull," because we couldn't guarantee the final generated quality met user expectations. But today, after model capabilities improved, the results we deliver to users have strong determinism. If you want a team to sustainably maintain generation quality daily and monthly, previously it was hard to answer this question. From a technical perspective, determinism has indeed improved significantly.
Also, for operations teams, there's been a very strong confidence boost. In the past when communicating with B-end clients, it was hard to guarantee the video made yesterday goes viral and the one made today still has traffic. But today, after the model ceiling rises, when we deliver results to users, user confidence is greatly boosted. This is my understanding of "determinism."
Regarding the short drama explosion, for our tool-making sector, opportunity and crisis coexist. If our AI-generated video tool is just an API-wrapping tool, it has no moat. That's also why starting in the second half of last year, we wanted to consolidate the product from pure AI tools into a core product that can continuously and stably deliver results to users like a team.
Abner Zhao: I have a question I'd like to discuss with you, because you just mentioned that the strong momentum might be opportunity and threat coexisting for you. We also saw news that Jianying (CapCut) might add AI video creator tool capabilities, maybe appearing in the next iteration. What does this mean for your relatively traditional tool-making company?
Xiangyang Tang: Good question, this is a bit beyond my lane. I can answer briefly, only my personal view. Our company is about 13 years old now. When Douyin came out, we were also anxious. We originally always did C-end entertainment-type video experiences; when we found Douyin had cooler experiences, for a tool product the ceiling might suddenly hit. But I always say, where there's danger there's opportunity. We flipped the thought—what opportunity does Douyin's explosion mean for us? Maybe this is our space.
Answering your question, regardless of whether Jianying does it well, I think it really is done well, whether for professional creators or novice users. But this product will always have places it can't cover, and that's our opportunity: consolidate our industry experience, serve people with information gaps who find AI hard to use. Our ShanJian was originally called Datao Brothers; our company mission is "make video creation simpler." Even today, with various 2.0, 3.0 models so strong, for novice users—for example, we launched a product last year where you can say a sentence directly to it like using Doubao and it generates a video. But we found users still open it and ask: what should I say? So there's still a threshold in prompt wording.
Abner Zhao: Next please David.
David Wang: The first three teachers spoke very practically; my answer will be a bit more abstract, sharing my views or vision.
Starting with the second question, short drama monthly revenue reaching 1 billion—how do I see it? I think this is just a start, maybe just the tip of the iceberg. Because the market is gradually generating new demand toward premium AI video, AI film. Like our FizzDragon made the world's first film entering theatrical release, Madam Cheng I-sao; we think in the future there will be more and more AI-made long-form dramas, films, documentaries. So current tool consumption is far from peak; there's still huge climbing space.
Saying something more abstract, I think AI tools first reduce costs and increase efficiency for traditional industries; second, for ordinary people or non-film-trained people, they've for the first time given them the chance to experiment. Like the Zombie Scavenger video made by the Yunnan guy, AI technology gave people more opportunities to try. Including our team's earliest core creators weren't film majors either, but we decided to make such a film. In 2024 everyone thought it was completely impossible, but we gritted our teeth and did it.
I think the global content production industry is waiting for the next Toy Story. Why this example? Because Toy Story was the first truly full-CGI well-made work that ran through and was remembered. Currently in AI video, everyone is waiting for the next memorable good work. It must be a feature film, with complete narrative, character arcs, and character stories—not flashy short, fast, cheap videos.
The first question's keyword, I think it's still "content" or "IP"—this is always the most core point that moves us. I'm often moved by good stories; I think the industry is waiting for these special, warm good stories. We're now also building our next feature film, King Gesar, about the Tibetan epic scroll, hoping to create a good story that resonates with global audiences.
Abner Zhao: So is this King Gesar something you're doing with ambition? Because you just mentioned, the industry may soon see the first AI feature film truly remembered. Do you think it could come from you?
David Wang: Let me say a bit more. When people mention epics, many think of Homer's epics, the Odyssey. But actually the longest and relatively most complete story in the world is King Gesar. It's also a wonderful story in our Chinese cultural treasures, with intricate plot; we think it'll be a good subject.
Abner Zhao: Is this targeting domestic or overseas markets?
David Wang: Currently we're mainly targeting overseas markets.
Abner Zhao: Can you share roughly what level the production cost of this work is?
David Wang: If just production, maybe several hundred thousand USD. Because compute costs and human costs together are roughly that number now.
Abner Zhao: Understood. Actually my conversation with David has already entered the second segment; everyone is very interested in David's situation. When Madam Cheng I-sao just released, it received huge attention in the industry. What symbolic significance do you think it has for AI overall? Also, there may be several AIGC films releasing in the domestic summer season; from your perspective, what's your view on the upcoming domestic AIGC theatrical market?
David Wang: We're very much looking forward to the several AIGC films preparing for domestic release. Madam Cheng I-sao's biggest significance has two: first, it achieved "cross-border," with over 100 creators from 13 countries and regions around the world making it together. This was a completely unimaginable decentralized creative process in the past—like an American producer plus Chinese voice actors, Indian composers all collaborating. Second, it proved that AI film can enter theatrical release; this truly has epoch-making significance. If you Google "world's first AIGC film" now, Madam Cheng I-sao comes up.
Abner Zhao: Next I want to talk with Jony. From your perspective: first, why did you choose to start your own business this year? What signal made you feel it was time? Second, you do AI film content overseas distribution; what type of content do you think is currently popular in overseas markets? Is it AI long dramas like David just mentioned, or the very popular AI short dramas now?
Jony Deng: Me starting my own business actually was because I felt two opportunities or windows had arrived. First, AIGC large models and their production capacity truly reached the level from "usable" to "commercially viable." It's actually proven that AI-generated human-like characters are popular at home and abroad, and people are willing to pay for them, which proves the commercial logic works.
Second, I saw many cross-border opportunities for AI content. Previously, film and TV content was strictly speaking not a global industry. Although there are film industries everywhere, everyone does their own thing; besides Hollywood exporting worldwide, Indian dramas are watched by themselves. Chinese content has platforms exporting to Southeast Asia, but more is still closed within its own culture. The AI era broke the production problem—previously requiring location shooting, finding directors and actors, hard to assemble cross-border. Now many AI short dramas on North American platforms (like werewolves, Greek mythology, etc.) are actually made by Chinese people. It both uses China's strong production capacity and cross-border taps into other countries' language and culture, truly internationalizing production and demand.
China has the strongest combat power, lowest production cost, and strongest adaptability overseas. But the problem is, this much content needs to go global, needs distribution—how do you find orders and distribution channels overseas? Many domestic companies don't understand this well. For the past five years I've been doing Southeast Asia ground expansion; we have sites and personnel in Thailand, Indonesia, and Malaysia. Over the past year-plus I made 12 Malay dramas in Malaysia—using Malay actors, telling Malay stories in Malay, and ultimately three dramas were bought by Netflix. This showed me that when you truly invest in understanding local ethnicity, language, and culture, then use Chinese teams' creativity and production capacity to produce, you can get good results. These two together make a great opportunity. Minipix hopes to be the link, connecting Chinese production capacity with overseas demand, releasing greater energy.
Abner Zhao: Understood—combining Chinese production capacity advantages with emphasis on localization work can create a huge market. So you started your own business to take this cake. But competition is fierce; the Mr. Liang on your left is about to do distribution too.
Wei Liang: We can collaborate.
Jony Deng: Yes, win-win cooperation. That's the biggest energy aggregation. AI is no longer a threshold; everyone has special experience in each vertical field. The best way is to collaborate everyone together.
Abner Zhao: Next let's talk with Mr. Liang. As a traditional film industry entrepreneur, share your journey—why did you choose to fully commit to AI?
Wei Liang: Why? Definitely because film didn't work out. But I was lucky; at that point in time we All-in'd on AI. Now a year later we're serving the film industry back; I think it's an inevitable path. The core is truly to quickly enter the construction process of a global AID market. In this process, like Jony who deeply cultivates the Southeast Asian market, I truly feel we can collaborate. He just started his own business, and we currently cover eight major language regions: India, Indonesia, Middle East, Brazil, Spanish, Portuguese, Japan-Korea, Europe-America. Everyone collaborating together to do AIdeo market production, investment, marketing, distribution, and screening. This speed closed loop can start immediately. I also believe that in the future, most of those truly disrupting the global AIdeo entertainment market should be Chinese people.
Abner Zhao: Because of time, I actually prepared many questions for everyone, but may not be able to discuss in detail. My last question I'd like to leave for Mr. Xiangyang. I just asked Mr. Xiangyang an off-syllabus question; the answer was excellent. Finally, please summarize for us: what do you think next year's biggest opportunity, biggest threat, and trap in the AI content going-global direction might be? Please share with everyone.
Xiangyang Tang: I think next year's biggest opportunity might be "AI content team localization." At the beginning of the year we talked about tools like OpenClaw, Hermes (Agent), and today we talk about AI employees, AI teams. There's an important thing—with so many smart heavyweights here today, everyone must realize that besides China there's a broader overseas market. But can we directly copy domestic hit logic overseas and succeed? This is worth thinking about. My view is definitely no direct copy.
So can this thing be team-localized? Single capability alone can hardly serve or satisfy B-end and C-end user needs; it definitely requires team-based capability. This is the most basic team service. Second is localization. With AI and No-code, making products is actually very simple; the hard part is service and localization. So I think for next year, AI capabilities will definitely be stronger. For us doing AI team localization, there may be some challenges, which are also new opportunities. This is my understanding.
Abner Zhao: Then our panel ends here. Thank you very much to all guests for the wonderful sharing, and we look forward to more opportunities to exchange next time.