After 1,450 Short Dramas a Day, the Real Moats of AI Entertainment Going Global Have Changed
Production is getting cheaper. Being seen is getting more expensive.
Until recently, mainstream platforms launched roughly 200 to 300 short dramas a day. The latest number Li Chengcheng has seen: more than 1,450.
Output has tripled or quadrupled — or more. But Li Chengcheng, CEO of Framesight Technology (VideoPlus), added the second half of the sentence at the same roundtable: distribution revenue has not grown by the same multiple. Ad spending is rising, and total market revenue has not kept up one-to-one.
AI has lowered the barrier to producing content — and raised the barrier to building a content business.
This roundtable was about AI entertainment going global. The guests came from different points on the value chain: Wu Jiabing, VP of Wondershare Technology; Yang Zhengxin, head of commercialization and marketing for Tiangong AI; Zeng Shirui, chairman of Wuxian Huanyu and Xinchuang Zhongyu; and Li Chengcheng, CEO of Framesight Technology (VideoPlus).
Some build tools, some build models, some build trading platforms, some produce video.
You might expect the answers to diverge. By the end, one judgment had become clear: production is getting cheaper, being seen is getting more expensive.
Asked to pick the single category best suited for globalization, three guests voted for short drama. Their reasons differed, but they pointed to the same word: certainty.
Yang Zhengxin gave the market result. Global micro-short-drama is already a market of roughly USD 10–20 billion, and AI short drama is still growing exponentially; DramaWave, under Kunlun Wanwei, already exceeds USD 65 million in monthly revenue. What AI has truly changed is not just how much cheaper it is to make one piece of content, but that content can, for the first time, iterate like a product — validating, optimizing and scaling in different markets, shrinking the distance between creation and market feedback. The Tiangong workbench, he said, exists to make that production mechanism run. As production stabilizes and the distribution capability represented by DramaWave keeps validating, the commercialization ecosystem around both ends is taking off in more and more countries.
Wu Jiabing pointed at the channel window. Among overseas short-drama distribution platforms, he said, 80% to 90% come from China. In the past, a content company wanting to work with Netflix faced a very high bar. Today, making contact with a platform like DramaWave might happen at an event.
Model capability has gone global; production is no longer a Hollywood-only game. And the distribution channels are held by Chinese companies. That combination is rare.
Zeng Shirui offered the founder’s judgment. Short drama and comic drama are like the restaurant business: low barrier, big market, fierce competition. Everyone — men, women, old, young — can watch them, and user-education cost is lower than for interactive content or games. For most teams, certainty is higher.
Put the three statements together and the meaning is blunt: the opportunity in short drama looks on the surface like content catching fire, but underneath, the chain of production, distribution and monetization has already been run through by someone.
Li Chengcheng was the only one who voted for interactive content and games. His standard is technical and cold: in the AI era, whichever category can more easily obtain data that drives self-evolution inside a sandbox is the one that is easier to do.
Games and interactive content have deeper user interaction. Where users click, how long they pause, which branch they choose, why they quit — all of that can be recorded. The gaming industry has long been good at event tracking, data logging and analysis. Streaming knows far less: whether a user bounced, how long they stayed, likes and comments — that is about the end of it.
This answer sounds farther from money, but closer to the future. Short drama validates today’s cash flow; interactive content bets on tomorrow’s mode of production. When content can be repeatedly trained, validated and iterated on data, entertainment products increasingly resemble an evolvable system.
So the most tension-filled moment of the roundtable: one side has already run out USD 65 million in monthly revenue, the other is still waiting for its own Sputnik moment.
Zeng Shirui used e-commerce as an analogy. People first used Taobao because the same goods were cheaper than offline. Later, looking at the bills, they were spending far more online than offline. AI content may take the same path: it starts by saving money, and the more you use it, the bigger the budget becomes, because it grows into a different kind of content ecosystem.
He offered a more positive judgment: AI’s role has moved from cost reduction to quality improvement. Creators used to have imagination but were bound by cost, afraid to attempt hard-to-shoot scenes and stories. Now, the things they dared not imagine can be produced. The teams that capture the dividend are those with quality IP and strong creative capability.
Wu Jiabing added the other side. AI lowers the entry barrier and also makes content easier to homogenize. You copy me, I copy you; basic content grows more and more abundant, and its consumption value gets thinner. Moving up, whoever can differentiate, whoever can grow original IP into big IP — that is the hard part.
Yang Zhengxin put it in more product terms. Content used to have to be completed and then thrown into the market to be validated, success or failure like opening a blind box. AI shortens the creative process and shrinks the distance between content and market.
From lyrics, melody, arrangement and mixing to vocals, music can be finished faster; short drama can also enter different markets faster — validate, adjust, validate again. Every creator has to become an aesthetician.
Li Chengcheng was the most sober. He reminded everyone that supply growth will outpace consumer growth. What ultimately consumes content is people, and people’s attention and time are limited, so the power-law effect will become more pronounced. The head becomes more head; the middle class of content struggles more.
That is the real order in which AI reshapes content supply: first lower cost, then release imagination, then push competition toward aesthetics, IP and iteration speed.
Yang Zhengxin said many teams doing AI content overseas instinctively treat overseas users as one uniform user. In reality, they are one country after another, groups of completely different people. Culture, religion, consumption habits and content soil all differ. The real differences between markets are not just genre preference, but how users spend time, where they enter content, and why they are willing to pay.
These differences directly determine a category’s acquisition approach, retention logic and monetization model, so platforms cannot run one global category strategy. When Tiangong AI built the music model Mureka, for example, no one expected AI music to grow so fast in the Spanish-speaking region. Capability kept accumulating to a certain point, and then users from South America and Latin America suddenly poured in.
They later tested the African market in Nigeria and found the local music ecosystem equally active. Yang used one phrase: the pond-effect. Growth is slow at first; then, around day 30, the lotus suddenly covers the pond.
Wu Jiabing’s reminder was more down-to-earth. You cannot sit in an air-conditioned room and remotely manage an American team. You hold a meeting at nine in the morning and the American colleague drags themselves up at midnight — that alone means it cannot last. To truly do a local market, you have to go there, and see how people eat, talk and spend.
In May in Silicon Valley, he observed more and more startups of a dozen or twenty people setting up small three-to-four-person branches in the US. That used to be a big-company configuration; now it has become a basic move for startups.
Americans eat seriously when they eat, chat seriously when they chat, and rarely pull out their phones. That habit is very different from China’s, and it shapes content forms.
India is another example. Many people assume Indian users have weak willingness to pay and love to freeload. But Wu said having little money does not equal unwillingness to pay for something truly loved. For products like music that let people express themselves, they will pay.
Zeng Shirui gave a sharper example. Journey to the West is a perennially re-told story in China; Sun Wukong can appear in countless stories. But in some Muslim-majority countries, the character Zhu Bajie may touch a taboo. The same IP, transplanted into a different cultural soil, has a completely different fate.
So the word “localization,” when it lands on execution, is very concrete: which market to pick, which taboos to avoid, what aesthetics to adapt to, who to build the local team with, and where to invest the production cost you saved.
When Li Chengcheng dealt with North American distributors, he felt something similar. They do not change their content judgment because of AI. They care more about whether the saved budget can go to music, dubbing and higher-tier actors.
AI can lower production cost. It cannot make cultural judgments for you.
Zeng Shirui cited a domestic case: the animated series Ten Thousand Demon Records has already been updated to season 12, and possibly more than a million people have reserved season 13. The content assets of the first twelve seasons are always there, continuously bringing in new users; old users may keep following the next season.
In the context of going global, this case is instructive. When short drama and AI drama first appeared, they looked like fast-moving consumer goods: watch and leave, pay for traffic and it rises, stop paying and it falls. But once content begins to accumulate IP, the logic changes. What you buy is not a single burst of traffic but an asset that can be monetized repeatedly.
This also explains why, although the guests picked different categories, they all circled back to the same thing: the team itself. Zeng Shirui said many people, hearing “which category is best for globalization,” unconsciously translate it into “which one is my opportunity.”
There is no standard answer. Personnel makeup, resource endowment and development plans differ, so answers differ.
At the close of the roundtable, one sentence was worth remembering: before you dig a category to the bottom, you do not know where the ceiling is. The ceiling you see in front of you, dig further, and there may be another layer.
By the end, the category debate had faded. Short drama has short drama’s cash flow, games have games’ data depth, interactive content has interactive content’s future, comic drama has comic drama’s growth speed. What is genuinely hard is whether you can use AI and new technology to solve a few more of the key problems in supply, distribution, localization and payment.
After the production barrier is knocked down, entertainment going global enters the stage of competing on system capability.
The full transcript of this roundtable is worth reading to the end. You will see several completely different ways of judging: some look at revenue, some at data, some at channels, some at culture. Put together, they form the real map of AI entertainment going global.
Zeng Shirui, Chairman, Wuxian Huanyu (Infinite Universe) & Xinchuang Zhongyu (New Creation Zhongyu)
Wu Jiabing, VP, Wondershare Technology
Yang Zhengxin, Head of Commercialization & Marketing, Tiangong AI
Li Chengcheng (Edward), CEO, Framesight Technology (VideoPlus)
Wang Chaochao, Partner, Unique Capital
Zeng Shirui: Hello everyone. I am Zeng Shirui, founder of Chengdu Wuxian Huanyu and Xinchuang Zhongyu. We mainly run two platforms. One is an entertainment digital-asset trading platform authorized under a cultural property exchange license. It helps entertainment project practitioners raise funds using a rights-based financing method that is neither equity nor debt.
For practitioners, this is a relatively comfortable funding channel; for investors, it also reduces the cost black box that plagued the industry. In the past, such investments lacked liquidity; now, with the exchange mechanism, investors can decide after T+5 whether to resell. That is our trading platform.
Second, we have built an AI-powered distribution middleware system to support the trading platform. It matches listed units with upstream and downstream players in the industry, forming a closed business loop from project initiation, advancement, distribution and release to capital recovery. These are currently our company’s two core businesses.
Wang Chaochao: Somewhat like drip irrigation financing plus an industrial collaboration platform?
Zeng Shirui: Yes.
Wang Chaochao: Mr. Wu, please introduce yourself.
Wu Jiabing: I am Wu Jiabing, from Wondershare Technology. Wondershare is a listed AIGC company. Founded in 2003, it has been doing overseas business for 23 to 24 years. We also have a creation tool related to short drama and comic drama called Wondershare Drama Factory. Friends in the industry may be familiar with it. Thank you.
Wang Chaochao: For those who follow the A-share market, Wondershare is the AIGC company on the A-share board, also known as the “App Factory.” Mr. Yang, please.
Yang Zhengxin: Hello everyone. I am Yang Zhengxin, Adam, from Kunlun Wanwei, in charge of Tiangong AI’s commercialization and marketing. Kunlun Wanwei is an A-share listed company. Tiangong AI builds both models and platforms. We have a self-developed text LLM and the music model Mureka, which we consider among the world’s leading AI music models. We also have the video model SkyReels and the world model Matrix-Game. Corresponding products include Skywork and the overseas AI music platform Mureka — think of it as an AI version of Spotify. On the short-drama side we have DramaWave and FreeReels, both of which have reached relatively top positions overseas.
Wang Chaochao: Multimodal — each with its own model, doing both commercialization and platforms. Mr. Li, please.
Li Chengcheng: Hello everyone. I am Li Chengcheng, CEO of Shanghai Framesight Technology. You can also call me Edward. We have been in stealth mode, fairly low-profile, but what we do is quite relevant to everyone.
We are an AI-first company. At the model-application layer, we do end-to-end video secondary-creation services. For the secondary-creation assets needed when film and TV content is distributed and promoted, we can generate them end-to-end, with generation quality and conversion results at what we call the current industrial SOTA level. Our product Smart Drama Workshop will open for free use in September, and interested friends are welcome to join.
Underneath, we have a code-based editing model, which is fairly general-purpose. Unlike pure text-to-video models, it does not focus on generation. Instead, it makes the model deeply understand shot narrative, visual material and editing logic, describing video through code, thereby enabling smooth switching and recombination of cross-style, cross-modality materials — not simple cut-and-paste on a timeline.
Right now, besides domestic short-drama and film/TV teams, we are also advancing technical integration and scenario validation with some top North American streaming-ecosystem clients such as Netflix and Google.
In China, beyond short drama and film/TV, we are also trying the code-based editing model in robotics: a domestic robot company with very large shipment volumes will embed our model to generate daily vlogs and similar content for users. Separately, we are building hardware that links micro-expressions with EEG signals. A core problem is how to conquer ultra-low-latency streaming generation (StreamDiffusion) for diffusion models and high-frequency response. We try to use hardware to capture micro-expressions and EEG, train data into models, and help robots generate better facial performances.
Wang Chaochao: “Framesight Technology,” if you only hear the pronunciation, may remind people of the “Zhenguan era,” right? Today, the biggest token consumers are coding and video. Mr. Li does both, using coding to drive video generation.
Let’s move to the first question. Everyone here sits at a different position in the drama and video value chain: digital assetization, AI creation tools, LLM and commercialization, and video production.
If you can only pick one category, which kind of entertainment content is best suited for globalization right now? Please answer in the context of your own business: short drama, comic drama, games, or interactive content — pick one of four. Edward, you first.
Li Chengcheng: While preparing, I thought this was a very good question. I want to look at it from another angle: in the AI era, whichever category can more easily obtain data that drives self-evolution inside a sandbox is the one that is easier to do.
The biggest difference between the entertainment industry and programming is that entertainment content is hard to standardize, and evaluation and verification are difficult to do in AI. To do this, you must have enough data.
Whichever category can more easily obtain data and build a sandbox is the better one to do. I am talking about the future, not the past. My direct answer: interactive content and games are better, because their data depth is deeper. Of course, forms like streaming may gain more data in the future too.
The key is sandbox-level data. Google recently published research on moving toward AGI, called Dream-RSI. It needs enough historical state and action data for AI to judge what to do next, so data matters. Realistically, interactive and gaming data depth is better.
Wang Chaochao: Can I understand it this way: games and interactive content are both cultural products and digital products, and real humans interact with them most deeply. The data you mean is mainly the depth and frequency of user-product interaction?
Li Chengcheng: Yes. First, interaction depth. Second, looking at historical accumulation, gaming has done deeper event tracking, data logging and analysis. On the streaming side, the deeper work in the past was Hollywood’s emotional-curve analysis, but sampling for that kind of analysis is hard.
New-media forms like TikTok usually only know whether a user bounced and how long they stayed. Beyond that, you have no more interaction data — just likes and comments.
Wang Chaochao: A decade ago, “interactive entertainment” was a popular concept on the A-share market. I think we are only now truly entering the era of interactive entertainment. Mr. Yang, which one do you pick? You can only pick one.
Yang Zhengxin: This is a good question. From the model side of Kunlun Wanwei and Tiangong AI, our music model Mureka, in our view, is at global SOTA level.
Logically, with a music model, we might pick music. But I personally believe short drama is better suited for going global. My judgment is based on the flywheel of content, distribution and commercialization.
If you look at the whole global micro-short-drama market — including live-action short drama — it is already very large, roughly a USD 10–20 billion market. AI short drama is growing exponentially. Short-drama content in China and overseas has already formed a system.
As AI develops, distribution systems improve and commercialization improves, the number of covered countries keeps growing. The flywheel is turning. From stable content structure to stable distribution model to commercialization ecosystem, once we spin the whole flywheel, the total market keeps getting bigger.
From my personal view, I also have an observation: our short-drama platform is called DramaWave. Our monthly revenue has already exceeded USD 65 million, growing very fast.
Short drama has a mature foundation in content and production requirements to begin with. AI short drama is not subverting a stable industry; it is growing new wings on top of the original base. That is a very important point.
Wang Chaochao: Mr. Yang is judging from results that have already run out: short drama grows fast, has high output, and monetization is relatively smooth.
Yang Zhengxin: Correct. The back-end (post-production and delivery) is a very important indicator for us. Yes.
Wang Chaochao: That is from the perspective of making and selling products. Mr. Wu, what is your view?
Wu Jiabing: Short drama, comic drama, games, and interactive film/games — all are going global to varying degrees. If I give advice, the key question is which category is easier to go global with right now.
My view is similar: short drama is currently the easiest industry to go global with. But precisely because it is easy, competition is also fierce. There is still a threshold for going overseas; competition is not as fierce as domestically. Not everyone can smoothly enter overseas markets and do well. So if you are going to do going-global now, short drama may be the optimal choice.
There are several reasons. First, models are truly mature now. In the past, competing with Hollywood on production quality was almost hopeless. Now model capability is spreading globally, not limited to Chinese models, so it is a huge opportunity for ordinary people.
Second, as far as I can observe, 80% to 90% of overseas short-drama distribution platforms come from China. That means we are in China’s best era — we have captured all the distribution platforms.
With an American platform like Netflix, the cooperation threshold may be higher. But with DramaWave, you might make contact right at a Unique Capital event. I think this is a very good window.
Third, mature models and platform-channel reach let many people with ideas enter the content track. In the past, the film/TV threshold was very high and ordinary people could hardly participate. And only a tiny group of very top stars could enter.
Now, everyone in this room has an opportunity. To summarize: based on the points above, this is a very good opportunity, so people can try it.
Wang Chaochao: Let me restate Mr. Wu’s point: Chinese people have great enthusiasm for improving their lives and creating wealth. Short drama is already a proven sub-category. In overseas markets, it may be hard to sit at someone else’s table — but we can also open our own table and build a platform. Mr. Zeng?
Zeng Shirui: The question is well designed. But for most people in the industry, I think they subconsciously turn it into another question: among these categories, which is my real opportunity? With that question, answers vary by person, depending on the team itself.
Wang Chaochao: Right, we are asking a general question; the guests can answer from their own backgrounds.
Zeng Shirui: Yes. Different teams’ personnel makeup, resources and development plans all affect the answer. Back to the original question: my conclusion is close to the previous two speakers — short drama and comic drama may be the better tracks. Their audience threshold is lower; men, women, old and young can all watch short drama.
Interactive content and games currently have more young users. As technology matures and market education deepens, other content may have bigger prospects. But looking at today’s industry entry threshold, short drama and comic drama are the higher-certainty opportunity.
Looking at overseas distribution platforms again, many are built by Chinese people. The previous speakers also validated this with business data. Combined with my own judgment, our two platforms’ data and the cases of companies we serve, short drama’s certainty is indeed higher.
It is just like the restaurant business: low threshold, big market. Done well, the market is huge, but competition is also fierce. So which track to pick as your main track comes back to the earlier question: where is the team’s real opportunity.
Wang Chaochao: The four guests gave answers in two directions. Three guests, from the content-supply and monetization angle, chose short drama, which has already run through the whole chain. Mr. Li chose interactive content from the angle of interaction-data depth and technology-driven content forms.
We just talked about category selection. The second question concerns content supply and production cost. AI can write scripts, generate images, make models and produce video.
AI’s role on the content-supply side is no longer just cost reduction. About two years ago, when people talked about AI, they mainly talked about cost reduction. Now, whether for drama or interactive content, competition has entered another level. It is also changing the logic of content production.
In this context, what is the most critical step in AI’s reconstruction of content supply? Mr. Zeng, you start.
Zeng Shirui: This resembles the development of e-commerce. Initially many people used Taobao because the same goods were cheaper than offline — maybe 70% to 80% of the price. But now, looking at the bills, people are spending far more online than offline.
Short drama and AI-derived content may take a similar path. At first, businesses and consumers use AI content because cost is lower; the more they use it, the more they may spend, and it becomes a different content ecosystem.
From my perspective, AI models are already very capable. At least by now, its role has moved from cost reduction to quality improvement. Recently I have seen many projects whose planning methods differ greatly from traditional film/TV.
In the past, people had imagination but were bound by cost, afraid to attempt hard-to-realize scenes and stories. Now, with AI’s cost advantage, the things, scenes and stories people dared not imagine can be produced.
For creators, the creative space is bigger. For audiences, content is richer; things they could not see before, they may see now. So the most critical factor, rather, is teams with quality IP and very strong creative capability. They may capture the dividend of moving from cost reduction to quality improvement. That is my view.
Wang Chaochao: It used to be said that AI frees productivity; now it also frees imagination, moving from cost reduction to quality improvement. Mr. Wu, what is your answer?
Wu Jiabing: I think AI first lowers the entry threshold. In the past, many people could not enter this industry; now most people have a chance.
But how do you create differentiation? Think back: most products were handmade at first. Then machines arrived, bringing industrialized production. Then came intelligent manufacturing, where some factories are almost entirely machine-run. After production became universal, products gradually differentiated again. Different products build difference through brands and brand propositions.
Content production works the same way. AI lets more people in and also makes content easier to homogenize. Whether original or not, you copy me, I copy you — homogenization gets worse. But on top of homogenization, the ceiling also rises.
Whoever can produce differentiated content, whoever can grow original IP into big IP — that capability is the big challenge of the future. Everyone can produce basic content, but at a certain stage basic content may stop generating much consumption value.
Eventually it moves toward high-end content, and not everyone has that capability. So I think the difficulty of content reconstruction lies in moving toward the high end.
Wang Chaochao: Hearing you, I think of dressing. Everyone can buy clothes at the same price, but the style and aesthetic of what you wear differ.
Styles differ, but sometimes they converge into trends and people borrow from each other. Still, the whole trend of beauty moves forward. Within trends there are dressing styles with many fans and leading figures that everyone recognizes — right? Mr. Yang, please.
Yang Zhengxin: The remarks by Mr. Zeng, Mr. Wu and Mr. Edward inspired me. In my view, AI reconstructs not just content cost — more importantly, content can iterate like a product.
In the past, whether short drama, music, games or interactive content, you had to complete the content first and then throw it into the market to validate. It might succeed, it might fail. AI greatly shortens the content creation process and also shortens the distance between content and market.
Take music: from lyrics, melody, arrangement and mixing to vocals, AI tools shorten the whole process. Short-drama workbenches and similar tools do the same.
We can dramatically cut content creation time. We are no longer just teaching AI how to do things; we also need to teach it to understand human aesthetics. Every creator becomes an aesthetician. Once creation time shrinks, content can enter different markets faster, validate and adapt to local needs. AI’s empowerment of content becomes a product-iteration logic.
Good content can find a more fitting market through iteration. I think that is the interesting part. My personal view: letting content iterate continuously like a product is an important change in AI’s reconstruction of the content ecosystem.
Wang Chaochao: From yesterday’s roundtable to today’s, everyone kept saying human judgment still matters. Today, from Mr. Zeng to Mr. Wu to Mr. Yang, the discussion has been moving up layer by layer: from freeing imagination, to aesthetic differentiation, to AI pushing human aesthetics and imagination forward in return. Mr. Li, what do you think?
Li Chengcheng: I look at it from another angle. Suppose AI keeps iterating and combines ordinary creators’ content with expert content better. Then supply growth will outpace consumer growth, competition intensifies, and the power-law effect may become more pronounced.
We have done a lot of distribution work and have seen short-drama output change with our own eyes. Two years ago, mainstream platforms had about 200 to 300 short dramas a day; the data I saw yesterday was 1,450 — output grew three or four times, or even more.
But distribution revenue is declining while ad spending grows. Total market revenue has not grown by the same multiple; it is not a one-to-one relationship.
In the end, people consume short drama — not machines. People’s attention and time are limited, so we return to the power-law effect.
Under the power law, the key is the team’s creativity, IP sources, and the ability to iterate human-machine collaboration. Competition will intensify — that is my judgment. At the same time, AI does give more people a chance to participate, but there will also be clear stratification.
I am not that optimistic about whether AI can achieve what Musk calls UHI — universal high income. In the end, the people who already had resources may benefit more.
People with good ideas but lacking resources to land them have gained some opportunity too. But how long that opportunity lasts, I cannot say.
So for IP owners or rights holders, from a marketing and traffic perspective, being able to lower promotion costs appropriately matters a lot — otherwise so much content cannot be seen and cannot compete.
Wang Chaochao: The 1,400-plus we just mentioned is only what mainstream platforms have already launched; there are also the ones that did not pass review.
Yang Zhengxin: If you count what was made but never launched, output is even higher.
Wang Chaochao: What ends up as waste may be even more. AI makes content supply flourish and also deepens stratification. The next question is about user demand. Supply answers whether it can be made; demand asks who it is for and how to monetize facing consumers.
From the overseas market data you observe, what are the differences in consumption habits for short drama, comic drama, games and interactive content across regions? How do these differences affect platforms’ category choices? Mr. Li, please start from your side.
Li Chengcheng: In this direction, I may have less experience than the others. But my past work touched both games and film/TV. For example, we have another company in Beijing that specializes in higher-end film/TV content. It is AI-core but also very IP-dependent.
We acquired a Hugo Award-winning IP, are co-producing a film with Wanda, plan to release it next summer, and are also negotiating overseas rights with North American distributors.
Because of this recent work, I have a deep feeling: North American distributors’ view of content has not changed much because of AI.
What AI brings is this: with production costs lower, they care more about where the saved budget can go. For example, from a content angle — music, dubbing, whether they can hire higher-tier actors, even Nicole Kidman. They consider moving money there.
So at least in the North American market I have touched, AI has not yet obviously changed the industry’s original nature. Rather, it has become a funding source for new directions of investment after lowering costs. That is my feeling.
Back to your question: at least for now, I have not seen major changes in the original industry structure.
However, some money is entering new directions that may merge in the future — like the combination of interactive content and XR — with more resources invested. Long-term, the moment of real change has not arrived. The “Sputnik moment” has not arrived.
Wang Chaochao: As far as you know, for Hollywood or other overseas content producers, a budget that was, say, USD 10 million remains USD 10 million. Only some links’ costs have fallen, and they invest the savings into other innovation or content-form innovation. Interactive content is still in its early stage.
Yang Zhengxin: Or the early stage is still ongoing.
Li Chengcheng: But early capital has begun to enter; VC money is flowing in. We see very early things — like the Roblox Agent platform and Seikai’s interactive products. These very small companies have raised significant money. At least from investment cognition, the market is already paying attention to these directions.
Wang Chaochao: The primary market thinks this may be a big market with a promising future.
Li Chengcheng: It has not fully opened up yet.
Wang Chaochao: You mentioned platforms that existed five years ago with some presence. Mr. Yang, please.
Yang Zhengxin: Edward’s sharing touched me. I think there is a small misconception in AI content going global: many people treat overseas users as one uniform user.
Wang Chaochao: But in reality they are groups of completely different people.
Yang Zhengxin: Right, different groups. We have distributed short drama to more than 170 countries. Different countries’ cultures, religions and cultural soil all differ. Some content fits North America, some fits Southeast Asia. Ecosystems differ, and content changes a lot.
At the same time, there is a “pond effect,” like CC just said: some companies may have existed for a long time but erupt at a certain point. When we built the music model Mureka, we did not expect AI music to develop so much in the Spanish-speaking region. But as we kept accumulating capabilities for the Spanish-speaking region, a large number of Spanish users suddenly poured in, and our presence in South America and Latin America rose quickly.
Not long ago we also went to Nigeria to test the African market and found the local music ecosystem very active. For the music category, we had been accumulating capability and letting it grow. We do not treat overseas users as one group; we adjust for different markets, countries and cultures — until the pond effect arrives: growth is slow at first, and then around day 30 the lotus suddenly covers the pond. That moment only comes from continuous grinding between content and different cultural soils. Like the last question, AI’s reconstruction of content is about making content more like a product — able to iterate, update, and be cultivated and developed in different cultural soils, waiting for that moment.
Wang Chaochao: Right. Some niche markets have always existed; there was just no matching product, and AI, data or technology was not enough to produce content suited to the locals, so they were ignored. Once a fitting product enters, monetization may actually become easier.
Yang Zhengxin: Exactly.
Wang Chaochao: Mr. Wu, what do you think?
Wu Jiabing: We have been doing going-global for a long time — since 2003, our first product targeted overseas. So we have some understanding of overseas markets. Before you do it, it feels far away; once you actually do it, you find that different users’ needs, habits and cultures are all different.
Let me share a few small cases. I also oversee the group’s strategic investment and go to the US at least twice a year. In May in Silicon Valley, I observed several differences.
First, to do local content and truly go global, I see more and more startups setting up branches or offices in the US. In the past, only big companies considered setting up a US branch; now startups of ten or twenty people set up a three-to-four-person branch there. Localization awareness is much stronger than before.
Second, in American coffee shops and restaurants, I observed that locals rarely look at their phones while eating.
But at gatherings in China, when not talking, many people hold their phones. Americans eat seriously when they eat and chat seriously when they chat; their phones basically stay put. This consumption habit also affects content form and content production.
Third, people may think Indian users do not have strong spending power. But for some differentiated products, they are indeed willing to pay. Not having much money does not mean they will not spend on things they truly like. Through these three points, you can see that market differences are very obvious.
So, if you really want to do a regional market well, you need enough localization — go there yourself and feel the culture. You cannot sit in an air-conditioned room and remotely manage an American team: you hold meetings at nine every morning and make American colleagues drag themselves up at midnight — that will not work.
Only by being there do you learn what people value, care about, and what must be respected. Content built on that has a much higher chance of success.
Wang Chaochao: You just made three points; let me follow up with two small questions. First, TikTok short video and short drama share the character “short” and have swept the globe. You said Americans do not look at phones while eating. As short content spreads, will that habit change? Will it become like China’s?
Wu Jiabing: It will — in fact, it is already gradually changing. The habits of the younger generation are being shaped by the current content ecosystem. This is also why TikTok-type content grows so fast overseas: the younger generation’s habits differ from the older generation’s.
Take Japan as another example. We set up a Japan branch in 2008 and hired local staff. A distinctive feature in the past: Japanese employees rarely changed jobs; they might stay in one position for life. Now the younger generation is different — changing jobs every two or three years is much more common. The older generation still keeps the one-job-for-life habit.
Wang Chaochao: China’s cultural export influence is indeed far-reaching. Second question: the European and American markets’ content — how to make, distribute and promote it — has been discussed a lot. You mentioned India; that market is relatively unfamiliar to us. Can you expand on local users and their consumption habits toward cultural products?
Wu Jiabing: I just mentioned music products; we have also studied this market. Domestic music-product makers choose very different target markets: some only do Japan and Korea, not even Europe or America; some focus on Europe and America; some specialize in India.
India is supposedly hard to do; people think willingness to pay is weak. There are also many “freeloaders” — people doing LLMs may feel this: wherever there is a discount, there they go. But we later found that for some products, users are not just freeloading. They genuinely care about music and want to use it to express themselves.
Wang Chaochao: Can you then monetize by playing along with the freeloading mindset?
Wu Jiabing: Not all products’ users just want to freeload. If your product truly fits their interests and needs, they are willing to pay. So it still depends on the specific product — do not judge the market with stereotypes. Only by trying it yourself do you learn whether users are just freeloading or can generate commercial revenue.
Wang Chaochao: Right — speaking of freeloading, we freeload quite a bit too. Mr. Zeng, please.
Zeng Shirui: The three previous speakers were comprehensive, and my views overlap heavily with theirs.
Wang Chaochao: Can you talk about a non-mainstream market you are familiar with?
Zeng Shirui: I am not especially familiar with non-mainstream markets either. Building on what was said: we have exchanged views with friends doing Belt-and-Road business. Muslim-majority countries have large populations and many countries, but because of religion, there is actually high consistency in some concepts. Entering these markets is not easy.
For example, some elements in Chinese web novels and traditional culture are not taboos domestically but can become taboos in Muslim-majority countries — like Zhu Bajie in Journey to the West. Journey to the West is a great mythological theme in China, repeatedly adapted, with Sun Wukong appearing in countless stories.
But the same IP, in some Muslim-majority countries, may touch local taboos and be very hard to enter.
This means, when doing the global market, as Mr. Yang said, you cannot treat it as one whole, but as regions each with distinct cultural features. First choose the sub-market you value most, localize around it, then accumulate the long-tail effect. When short drama or AI drama first appeared, they were generally fast-moving consumer goods.
Now, I see some domestic works beginning to accumulate their own IP assets. For example, Ten Thousand Demon Records has been updated to season 12, and possibly more than a million people have reserved season 13.
The content assets of the first twelve seasons are always there, continuously attracting new viewers; people who watched earlier seasons may keep following season 13. Its story fits the path of Chinese web novels.
By analogy, North America, Europe, Southeast Asia and Muslim-majority countries each have content types that fit local characteristics and retain users well, and are worth seeking out for teams going global. Done this way, distribution strategy is also more precise than blanket coverage of overseas markets.
Wang Chaochao: Thank you, guests. We have talked from content supply to user habits, production cost and localization. I feel there is a common thread: to earn dollars, rubles or other foreign exchange abroad, you cannot just pick the category that looks best — you must pick the category that fits your own capabilities and target market.
From commercialization, technology, content depth and data, different products have different traits: short drama has scale, comic drama has growth speed, games have a strong foundation, and interactive content has stronger willingness to pay and deeper interaction. But their thresholds differ.
So I want to close with one sentence: the priority of globalization does not depend entirely on a category’s ceiling. Before you dig a category to the bottom, you do not know where the ceiling is; the ceiling you see in front of you may reveal a new layer when you dig further.
More importantly, in the whole chain from supply to payment, how many key problems can you solve with AI and new technology and new playbooks? The more and the faster you solve, the higher your efficiency in making money is likely to be. Thank you, four guests.
Edited on October 10, 2026. Unique Research.