---
title: "Not Designing Your Product for Globalization on Day One? In the AI Era, the Cost of Catching Up Will Be High"
author: "Unique Research"
sourcePublication: "Unique Research Substack"
originalPublishedAt: "2026-04-16T10:01:48+00:00"
canonical: "https://ffcap.cn/en/research/src-20260416-02html"
source: "https://uniqueresearch.substack.com/p/src-20260416-02html"
language: "en"
---

# Not Designing Your Product for Globalization on Day One? In the AI Era, the Cost of Catching Up Will Be High

_Original · Unique Research · 2026-04-16_

_Editor's note: The first-person report and its judgments belong to the original Chinese author. This English rendition retains the opening essay and full panel, including every speaking turn. Claims about investment, markets, companies, products, and globalization strategy are source or speaker claims, not independently audited findings. Company, fund, and personal names are transliterated where official English forms remain unverified. The OpenClaw download and user figures, Indonesia e-commerce penetration data, and all market-size claims are preserved as speaker statements without independent verification._

Unique Awards

Capital Perspective: Who Is Defining the Next Generation of Globalization Champions?

AI-era globalization champions will not simply be companies that "win in China first, then replicate overseas." They are more likely to be teams that design their products, organizations, and growth as global players from day one.

Not all AI companies are suited for globalization—this is of course correct. But if a company truly wants to become the next AI champion, it can almost certainly not confine itself to a single local market. Especially for software-, application-, and Agent-leaning companies, many times from day one you are already not competing in a local track, but in a global tournament.

At this "Unique Awards · Hangzhou AI WEEK" trends roundtable, several investors discussed precisely this matter.

I originally expected to hear many familiar talking points—like the market is huge, there are many opportunities, globalization is the inevitable path, and so on. But as I listened on, Iinstead felt that the most valuable part of this Panel was that it explained very specifically the difference between AI-era globalization and the previous generation of Chinese enterprise globalization.

It is no longer the old logic.

In the past, Chinese enterprises doing globalization often followed a three-step process.

First was channel going-global: finding locals to sell. Then came product going-global: making localized versions for different markets. Reaching a more mature stage, it was the entire organization going global, truly laying R&D, sales, channels, and operations capabilities across different regions of the world.

Everyone is familiar with this path.

Anker, Shein, and many previous-generation cross-border companies basically fought their way out this way. Their success is remarkable, but the underlying logic is still closer to "starting from China, then extending outward."

But in this AI wave, many things are no longer in this order.

AI First Is Not a Slogan

Zhou Qi from Jinqiu Fund used a term: AI First.

This term is actually somewhat overused now—many companies wish they could hang it in their introductions. But what she meant was not the vague "we also use AI," but two more specific layers.

One layer is whether the company internally has truly been restructured by AI. For example, why can a very small R&D team achieve efficiency far exceeding its headcount? It is not that everyone works harder, but that the entire company has absorbed Generative tools, automated workflows, and per-capita efficiency leverage.

The other layer is more critical: it is not adding an AI feature to a product, not "the thing we used to make now has an AI feature added," but truly treating AI as the core differentiating capability of the product experience.

When these two layerssuperimposed, many companies will become completely different from before.

You will find that traditional globalization companies value scale, processes, and the ability to manage large teams. But many AI companies today are not that large in size, their organizations areinstead lighter, with fewer people, yet they can still run fast globally. Because their production method has been changed, and their growth method has also been changed.

Luo Wei put it very directly.

They started investing in going-global in 2021, went through brand going-global and service-provider going-global, and by 2023 began focusing on AI. His feeling is that many previous-generation projects were labor-intensive. Once GMV reached a certain scale, you basically had to stack hundreds or thousands of people, competing on the CEO's ability to manage large teams and refine operations.

But this generation of AI companies is completely different.

Even if their GMV today may not yet be as large as previous-generation giants, the entire company's architecture isobvious lighter. Fewer people, but much higher efficiency. The so-called "one-person company" may be a bit exaggerated, but one person equaling ten people, or even one hundred people—this trend is becoming increasingly clear.

Behind this statement actually lies the most fundamental layer of change in AI globalization.

In the past, globalization was more like the expansion of organizational capability.

Today, globalization is increasingly like the amplification of technology leverage.

In other words, whether you can become a globalization champion no longer depends only on whether you can lay out a large organization behind you, but on whether you can use a lighter organization, stronger products, and faster growth to break through the global market first.

Products Must Be Global from Day One

This is also why several investorsin unison mentioned "products must be global from day one."

This sounds like a correct platitude, but if you think about it carefully, it is actually very specific.

For software products, for example, you do not wait until going global to add multi-language support—from day one you must build in multi-language, global deployment, compliance, and privacy. Zhou Qi gave a very good example: in the Middle East market, Arabic is read right-to-left. This is not something that can be solved by patching a translation pack before launch; it affects your earliest product design. Another example: GDPR, data compliance, deployment requirements in different countries—these are also not homework to make up after the product grows big, but must bepre-embedded into the architecture from day one.

Needless to say for hardware.

Supply chain, tariffs, certification, policies—many things even need to be prepared earlier than the product itself. Because you are not moving an already-established business overseas, but accepting the constraints of the global market from the very beginning when the business is just starting to grow.

This is really different from before.

In the past, many companies first ran the product and model mature locally, then did internationalization. Now, if many AI companies do not design for globalization from the start, it will be very painful when they try to catch up later.

Of course, having a "globalized product architecture" alone is not enough.

Because globalization, at the end of the day, is not about making one globally unified big product, but about making many localized small details.

On this point, Louis Liang Sicong spoke very interestingly. He used two terms: one called "breaking walls" (breakthrough), and one called "reconstruction" (reconstruct).

So-called "breaking walls" refers to AI, especially at the software and content level, beginning to break many things that used to stand in between. In the past, for hardware going-global or supply-chain going-global, there were many bottlenecks in thepathway. But today, if what you do is content, Agents, or software services, many thresholds that used to be uncrossable suddenly become lower.

For example, short dramas, games, content production—these tracks can all use AI to directly serve veryfine cultural needs and local needs in a specific region. What you might previously have relied on large local teams of writers, translators, and production teams to accomplish is now drastically reconstructed by Agentic-First workflows.

And so-called "reconstruction" refers to the production model of content and services being completely换了一套.

In the past, to serve a local market, you had to hire many people who spoke the local language, understand veryfine cultural differences, and build a heavy human system. Today, AI certainly cannot eliminate all these problems, but it has indeed changed the way production and delivery work. You can do content testing faster, run local versions faster, and match minority-language users faster.

But there is a particularly subtle point here.

AI appears to have democratized capability, but the results actually produced with AI will not become the same.

This is also why Liang Sicong emphasized that what truly matters is who understands that market better. For example, who best understands the Brazilian market, who truly knows what Portuguese-speaking users want to watch, what they care about, what they are willing to pay for. AI has only lowered the production barrier; it has not completed market understanding for you. On the contrary, because tools are more widespread, it ultimately tests even more whether your insight into the local market is deep enough.

This judgment, I think, is particularly important.

Because now when many people talk about AI globalization, they easily fall into an imagination—as if with models, translation, and content generation, the global market is suddenly flattened. But reality may be exactly the opposite. AI has indeed helped you open the door, but behind the door, each market's culture, language, consumption habits, compliance requirements, and channel structure are still vastly different.

Globalization Is Not a Buzzword, but a Set of Capabilities

What Zhou Qi said later about Southeast Asia also left a deep impression on me.

Many people talk about Southeast Asia as a single whole, but anyone who actually does business knows it is not a single large market at all. It is more like six or seven completely different markets pieced together—different languages, different political and economic environments, different consumption habits, different channel strategies. So if you have not built localization capability, do not have a local team, do not have locally-adapted sales and product strategies, relying solely on "I validated it in China, so I can also do Southeast Asia" is basically very dangerous.

And precisely because of this, this roundtable has been repeatedly reminding us of one thing.

Globalization is not a "buzzword."

It is a very specific set of capabilities.

You must be able to design a global product architecture, you must be able to build a global organization, you must be able to grow with different strategies in different markets, and you must truly understand what local users are thinking.

Put simply, globalization is not sending products out; it is driving understanding in.

Globalization is not sending products out; it is driving understanding in.

What Capital Is Looking at Now

From a capital perspective, the tracks worth watching next are actually becoming increasingly clear.

Hu Yanjun mentioned two directions they value highly, which I think are quite representative. The first is AI hardware, because this wave of interaction is changing. The previous mobile internet entry point was the screen—the phone, the tablet. Today in the AI era, human-computer interaction is gradually no longer relying solely on screens; voice, ambient sensing, continuous companionship, screenless interaction—all are催生 new hardware carriers.

This is also where China's supply chain has particular advantages.

Many software-application entrepreneurs, if they can find a suitable hardware entry point early on, mayinstead find it easier to build genuine product barriers in the global market.

The second is Agent-related infrastructure and interaction systems.

This part sounds more futuristic, but several guests actually mentioned similar directions. Zhou Qi talked about Build for Agent, mentioning underlying products like transactions, identity, and Memory. Hu Yanjun mentioned that in the future one person might maintain a dozen or dozens of Agents, and globally there might be tens of billions of Agents—then围绕 Agent interaction, settlement, and organizational methods, entirely new infrastructure opportunities will surely emerge.

If you put these statements in today's context, many are still just雏形.

But capital has actually already started looking along this path.

One of the most typical examples is the OpenClaw ecosystem.

Liang Sicong mentioned that OpenClaw has developed very fast, with GitHub downloads already reaching over 2 million; if you count the cloud version, actual users are probably between 10 million and 20 million. Although it still has many practical usage problems, what truly interests him is no longer OpenClaw itself, but the industrial-chain opportunities growing around it.

For example, doing identity authentication for OpenClaw.

This sounds a bit convoluted, but it is actually very much like a demand that will appear in the next stage. Because in the future, if Agents are not just chatting with you, but startmanaged wallets, executing tasks, consuming on your behalf, and calling services—then their identity, security, and trustworthiness will become new infrastructure problems. In Liang Sicong's words, it is a bit like giving an Agent a CPA certification to prove it is secure and verifiable.

Going further, it may also衍生 out communication systems, collaboration systems, and permission systems between multiple Agents.

These things look very early right now, but once Agents truly move from tools to "work-capable digital labor," they willinstead become very critical new infrastructure.

Don't Just Stare at Old Money

This is also why several investors ended by talking about one thing: don't just stare at old money.

Jichuan said something similar in a previous roundtable: old money is certainly very crowded. In this capital roundtable, the meaning is similar. Whether in mature European and American markets, or emerging markets like Southeast Asia, the Middle East, and Latin America, what is truly worth doing is not necessarily the old businesses that everyone already understands and that are already crowded, but the new opportunities where AI can truly change production methods, interaction methods, and organizational methods.

And Luo Wei's final reminder is actually something everyone who wants to go global should remember.

He spoke very practically: going global must have a sense of敬畏. Because besides the parts that can be made more efficient with AI, there are stilla large number of links that require physical contact, heavy operations, and strong execution that simply cannot be replaced by AI. Pre-sales, after-sales, logistics, warehousing, taxation, certification, testing—you cannot escape any of these, and many times they are harder overseas than in China.

But precisely because these things are hard, they will in turn become your barriers.

For example, if your hardware can smoothly pass Brazil's tax inspection and hardware testing, and sell well offline—this kind of seemingly clumsy, unsexy capability may in the end be precisely the most solid moat. Because the truly complex, troublesome, dirty and tiring part will scare off a bunch of people who only want to come arbitrage.

This point I think is particularly real.

Today many people talking about AI globalization easily place their imagination on "can do global business faster, lighter, smarter." But what capital will持续 bet on is often not those who only tell stories about speed and imagination space, but those who both know how to use AI to amplify efficiency and know which dirty and tiring work must be honestly tackled.

Conclusion

At the end of the day, although this roundtable discussed a capital perspective, what I heard it truly define is not who will become the "next globalization champion," but what underlying capabilities capital is actually looking at.

Not who is better at telling global stories.

Not whose TAM on the PPT is bigger.

Not even just whether you have AI.

But whether from day one you have designed your products, organization, and growth as a player in the global market. Whether you understand both technology leverage and local markets. Whether you dare to use AI to break walls and reconstruct, while also being willing to maintain敬畏 for the most complex, least sexy realities of going global.

If I had to summarize this roundtable in one sentence, I would say: AI-era globalization champions will not be companies that "win in China first, then replicate overseas."

They are more likely to be the kind that from the start did not define themselves as "a Chinese company doing overseas," but as "a company solving problems for global users."

These are two completely different starting points.

And once the starting point changes, the products, organization, financing, and market choices that follow will almost all change along with it.

More Conversation Details

Unique Awards · Hangzhou AI WEEK Trends Roundtable Panel

"Capital Perspective: Who Is Defining the Next Generation of Globalization Champions?"

Guests:

Well-known USD Investor — Liang Sicong (Louis)

Yijian Investment Founder — Hu Yanjun

Jinqiu Fund Managing Director — Zhou Qi

Yingdong Capital Cross-Border Lead — Luo Wei (Wayne)

Moderator: EPIC Connector Hangzhou Lead — Shiyin

Shiyin: Hello everyone, today's roundtable theme is "Capital Perspective in Defining the Next Generation of Globalization Champions." In the past couple of years everyone was discussing whether AI enterprises should globalize, but now the answer is actually quite clear—not all AI companies are suited for globalization, but the next AI champion must face the world from day one. Now let me introduce the guests. First is well-known USD investor Mr. Liang Sicong (Louis), who has long focused on early-stage investment in intelligence and internationalization, covering AI and robotics quite heavily. He has a well-known investment case and has very profound insights into both the globalization and localization of tech enterprises. The second guest is Mr. Luo Wei (Wayne), cross-border investment lead at Yingdong Capital. He is mainly responsible for the full-cycle investment layout of Yingdong Capital's cross-border going-global track. He has reviewed over a thousand going-global startup projects and has very deep frontline insights into the new paradigm of Chinese AI globalization and the globalization growth path of early-stage projects. Third is Mr. Hu Yanjun, founder of Yijian Investment, who has deep experience in investment and industrial layout in both the US and China markets, and has successfully invested in multiple listed enterprises both online and offline. Fourth is Ms. Zhou Qi, managing director of Jinqiu Fund. She previously served as global channel general manager at a leading enterprise, building overseas channel business from scratch, with deep insights into Chinese enterprise globalization and her own very closed-loop underlying logic. Now moving to our second segment, let's discuss the core markers of the next generation of globalization champions. I would like to ask everyone: in today's AI-restructured globalization, what are the core essential differences between the AI industry and traditional industry globalization? How do you judge the markers of whether an AI enterprise can become a globalization champion?

Zhou Qi: Very happy to have the opportunity to交流 with everyone here. First, I feel a great sense of affinity here, because when I was doing investments before, I invested in Anker and witnessed Anker's journey all the way to listing. Returning to this topic—actually doing business and later joining investment—was mainly because I saw this AI wave coming, which actually gives many entrepreneurs and international leading enterprises the opportunity to compete on the same stage and directly do overseas Global markets. If we look at globalization, actually many Chinese enterprises have gone through three major stages: first is channel going-global, finding people to sell; second is product going-global, making localized products; third may be the entire company building a globalized organization. Personally, I care about three points for doing globalization in the AI era. First I call it "AI First"—one aspect is the application of AI technology inside the company. The simplest example is that recently we have seen many startups whose R&D teams are very small, but the entire company uses a lot of Generative tools to help improve per-capita efficiency and achieve higher growth. The second aspect, in the company's products—whether software or hardware—it is not adding an AI feature to the product, but how to use AI well to make products with differentiated experiences. I think this is a very core point of globalization in the AI era. The second more important point is globalized organization. It may sound a bit abstract, but once you do overseas business, you will inevitably encounter many cultural issues, as well as productization issues in different regional markets, because doing globalization is essentially doing localization. Then you will also encounter that each region, each market's GTM (Go-to-Market) commercialization logic path and strategy are different. So how to have a globalized organization that allows us to do efficient growth因地制宜 globally is very important. Third, I think in doing globalization in the AI era, the so-called "product is a globally-deployed product from day one." For hardware, the supply chain may need to go global first, because policies in many countries and regions must be studied well. For software, it means your product can be deployed in multiple global locations from day one, and multi-language and Compliance issues are considered from day one. For example, if everyone does the Middle East market, Arabic is right-to-left—should these things be considered in product design from day one? And there are many GDPR and compliance issues. This kind of product architecture being global from day one, in my view, is also a change very different from before.

Shiyin: I would like to ask Mr. Luo Wei again. You have been deeply involved in the cross-border track for many years. I would like to ask: what is the core generational difference between the AI era and traditional cross-border tracks now?

Luo Wei: I am Luo Wei. Our fund is in Hangzhou and has always invested in early-stage projects. We started investing in going-global in 2021, mainly investing in brands, products, and various service providers at that time. In 2023 we began paying attention to AI, because we felt that the AI track from day one must compete in a global tournament, whether it is various software-leaning applications or various hardware-leaning AI hardware. We observe what is different between current companies and previous companies—I think it can be seen from three directions. The first is from the labor-intensive perspective: in previous projects, headcount was basically tied to GMV, and once reaching a certain scale it was basically hundreds or even thousands of people. What they competed on was actually the CEO's ability to manage large teams and refine operations, because they might be composed of many Amoeba organizations. But now it isobvious different when doing software and hardware through AI. Although the GMV of this wave of AI companies may not yet be as large as that brought by previous-generation Shein or Anker, it isobvious felt that the entire company's architecture has become lighter, with not as many people. It may be a bit exaggerated to encounter a "one-person company," but I think each person being equal to ten or even a hundred is possible. The second angle is the company's core assets. In the past, when looking at products, people might feel the product defined a certain track, or seized a certain webdevice Listing and first captured the dividend. But in the current track, the core value is no longer like that—it is the overall understanding of AI, the overall understanding of the track and product, and then creating new species with AI assistance; this is the true core value barrier. The third I want to look at from the growth perspective. In the past, companies captured traffic dividends, such as Google's SEO dividend, or Facebook, Instagram, Snapchat, or even TikTok traffic dividends. But the profits of this wave of AI companies rising may require founders to have the ability to operate on Twitter, or have strong appeal on Kickstarter, relying on their own content power and internal drive to achieve growth. So from these small angles, the current generation of companies is very different from before.

Shiyin: I would like to ask Mr. Hu again, because you have been doing venture capital in both China and the US. Will the paths of US本土 enterprises and Chinese本土 enterprises doing globalization be different?

Hu Yanjun: Let me briefly introduce—we benefited from the previous wave of mobile internet development, so for the current wave of AI development, it feels very similar to the mobile internet of 10 years ago, in a state of "empty streets" where everyone talks about AI, even surpassing the previous wave. Second, I personally understand AI as equivalent to the previous wave of mobile internet, but the previous wave was an information revolution, while this wave is a productivity revolution. Its力度, depth, and breadth of change across all industries will be greater. So we are very optimistic about this wave of disruptive innovation. A few years ago when looking at AI, I was thinking: to look at AI you only look at China and the US, nothing else needs looking at—so our fund has always shuttled between China and Silicon Valley. Which Chinese and American enterprises are we optimistic about? Returning to what was just said, the biggest change in this wave is AI, so we are more optimistic about AI-native enterprises. Especially those rooted in industry scenarios, with great potential to disrupt and redo the entire industry and original service model in the future. Especially enterprises with differentiated advantages in the industry competitive landscape, able to leverage the differentiated advantages of China and the US well—this is our view.

Shiyin: Finally, I would like to ask Louis (Liang Sicong). You have always invested in relatively early AI projects. If an AI enterprise wants to do globalization and do it well, what is a very underlying trait it needs?

Liang Sicong: I have two keywords combined with the big trend to talk about: one called "breaking walls," one called "reconstruction." In the past, when we looked at hardware or supply-chain going-global, there were many bottlenecks in thepathway. But today at the AI software level or Agent level, from the content production perspective it has broken the points of separation in between—so this is a brand-new model of "breaking walls." For example, the short-drama track and game track in content going-global all have the opportunity to use the breaking-walls model to serve localized cultural needs or users' nuanced needs, which pure hardware很难integrate. Second, why is it called "reconstruction"? Because it reconstructs the entire content production model. In the past, you had to hire many writers, screenwriters, even local translators—like India has over 20 commonly used languages, with bottlenecks in between. Today it has become an Agentic-First workflow, reconstructing the production model. Software-leaning AI enterprises actually all build new business architectures from these two points. Based on these two points, the so-called globalization champion is a result—we look for what these enterprises do differently. For example, the team that best understands the Brazilian market, knows what content and cultural needs Portuguese-speaking users want, and then uses AI for mass production. It seems AI has democratized, but actually the people using AI and the results produced will be very different—like the currently popular Prompt Engineering, all depend on how these AI companies specifically operate.

Shiyin: We have discussed some standard-type questions. Next, I would like to ask everyone about the tracks and markets you are optimistic about going forward. How to choose markets like the Middle East, China-US, Europe, Southeast Asia based on your own situation? Feel free to speak.

Hu Yanjun: Let me first share my feelings. Globalization is not a question of whether to do it—it must be done. Under the extremely competitive situation in China, whether based on the original system or a newly built AI business system, one must find differentiated advantages and entry points from a global perspective. It is best to "be born global, grow global," becoming a global champion in asegmented field. Now AI iterates too fast, everyone is very anxious (FOMO) afraid of missing out, but entrepreneurs should look at problems from the middle game. Just like the AI development path OpenAI previously talked about—the agent stage is accelerating, followed by innovators and organizers. Deriving from the middle game how we should do business systems, how to innovate. Following the big trend at your own pace will be calmer. The previous wave of mobile internet only provided information value; this wave is a productivity revolution, and many industries need to be redone. I don't really distinguish whether you are a Chinese or American enterprise—I value more whether you can, from the perspective of future AI transforming industries, find a reasonable model to redo things; this is the most core trait.

Zhou Qi: Let me briefly share my personal thoughts. I very much agree with what Mr. Hu said—AI brings the reconstruction of productivity. One big trend is that many enterprises have the opportunity to do global markets from day one. If doing enterprise services or applications in China is very difficult, because large enterprises compete fiercely. But in this AI wave, many enterprises do overseas global markets from Day 1. For markets, there are two categories: one is developed-country markets (Europe and America, Australia and New Zealand, Japan), which can commercialize well; the other is emerging markets (Middle East, Southeast Asia, Latin America), although relatively smaller in scale, user spending power and penetration rates are much better than imagined in China. Everyone should find a suitable market based on product form. Specifically, there are three directions I am optimistic about: first is application-related—with Agents, what is delivered is results, which requires you to build barriers and differentiated positioning, and design new business models based on Token pricing; good applications doing PLG growth will be extremely fast. Second is AI Infra (infrastructure)—Build for Agent is the most core, involving underlying technical products like transactions, identity, and Memory, and open source is a very good promotion channel. Third is AI consumer hardware—large models bring more possibilities for hardware forms, using computing power to bring better experiences, and the business model can become hardware + subscription services.

Shiyin: Next, I would like to ask Mr. Luo Wei: what is the difference between doing European and American markets versus doing Southeast Asia, Central Asia and other markets?

Luo Wei: I think the biggest difference is how large you want to make this company, because AI truly means you do as much as your ability allows. Sharing two observations: last year when we went to Vietnam for research, the largest local MCN company told us they use DeepSeek, and directly type in Vietnamese inside it—not Chinese or English—and the feedback was quite good. But this is limited to the largest companies using it. This year OpenClaw is very popular, but when I asked friends in Vietnam and Indonesia, almost no one was talking about it. If you are in China you may have blind spots—you not only need to understand AI, but also understand those countries. For current small and medium developers, a clever arbitrage opportunity is not to only stare at Europe/America or Southeast Asia where everyone is in "hand-to-hand combat." You can consider doing Japanese and Korean markets. Because the Japan-Korea region has been validated by the gaming industry as a high-payment region, naturally has language isolation, but is very willing to accept good products. If you see a certain AI product in Europe/America with particularly high ROI, you can consider quickly launching Japan-Korea versions to arbitrage and earn the first bucket of gold, then with cash consider the next step.

Shiyin: I would like to ask Louis: recently OpenClaw is quite popular. What products extending from OpenClaw are you optimistic about, or what traits do you value when investing in such projects?

Liang Sicong: OpenClaw has developed very fast, with verifiable downloads on GitHub at over 2 million; if adding various cloud versions, actual users are estimated at between 10 million and 20 million. Although there are still problems like "lobsters are easy to raise to death," there are several categories that genuinely produce value in practical scenarios. One is B2B-leaning—for example, using it to replace a certainaspect in cross-border e-commerce or content production, directly calling local software. The other, which I as a VC am more interested in, is treating OpenClaw as new opportunities generated by the industrial chain. For example, I saw a Hangzhou companyspecifically doing identity authentication for OpenClaw. Because future Agents have independent personalities and economic sovereignty (such asmanaged wallet consumption), they need something like giving them a CPA to prove their security. Based on this, complex communication systems between multiple OpenClaws can also be derived—infrastructure like this I pay great attention to. Also I want to cue Wayne: globalization is not just Europe and America. Before I graduated, I invested in the Indonesian market, which was non-consensus at the time, but in the past couple of years it has changed greatly—e-commerce penetration has gone from 2–3% to over 10%. Large base, fast growth, and after AI arrives combined with local language, cultural attributes, and habits, there will be many new models that can be done locally.

Shiyin: Since everyone is from a capital perspective and has invested in many enterprises, I would like to ask if you have any impressive cases to share about doing globalization?

Liang Sicong: One suddenly comes to mind. Every market is very localized; I pay great attention to category innovation based on something that does not exist locally. For example, pool robots—initially many Chinese investors could not understand them, because China generally does not have the living模式 of houses with pools, and cannot understand the high cost of local pool cleaning. But a group of people used China's supply chain advantages, combined with AI autonomous navigation to complete the cleaning model, creating a brand-new category. This kind of category innovation combining local demands with China's supply chain background advantages is the opportunity I am watching closely.

Hu Yanjun: Let me share my feelings. Globalization is not a question of whether to do it—it must be done. Under the extremely competitive situation in China, whether based on the original system or a newly built AI business system, one must find differentiated advantages and entry points from a global perspective. It is best to "be born global, grow global," becoming a global champion in asegmented field. Now AI iterates too fast, everyone is very anxious (FOMO) afraid of missing out, but entrepreneurs should look at problems from the middle game. Just like the AI development path OpenAI previously talked about—the agent stage is accelerating, followed by innovators and organizers. Deriving from the middle game how we should do business systems, how to innovate. Following the big trend at your own pace will be calmer. The previous wave of mobile internet only provided information value; this wave is a productivity revolution, and many industries need to be redone. I don't really distinguish whether you are a Chinese or American enterprise—I value more whether you can, from the perspective of future AI transforming industries, find a reasonable model to redo things; this is the most core trait.

Zhou Qi: Let me quickly share myuser experience of doing business in Southeast Asia over the past two years. Many Southeast Asian countries are still in the early stages of digitalization, and unlike China and the US which are single large markets, Southeast Asia is six major markets, six languages, and completely different political, economic, and cultural systems. So building new-market localization capability and expansion capability is very important. Second, whether enterprise services or consumer goods, channels are very important—cooperating with local cloud vendors and large channel players can add leverage to growth. Third is localization—you need a local team to understand the local market, because the pricing and sales systems of overseas products are completely different from those in China.

Shiyin: Mr. Luo Wei has actually been relatively quiet. I would特别 like to hear his insights and sharing.

Luo Wei: I think if we talk about going global, I still hope everyone can maintain a sense of敬畏 about this matter. Because besides the parts of the going-global track that can be made more efficient with AI, there are still many links that require physical contact, such as pre-sales, after-sales, first-leg and last-leg warehousing and logistics. Many processes cannot be replaced by AI, and doing these abroad is even much harder than starting a business in China. But rest assured, these difficult links will in the future be your barriers. Imagine if your hardware can pass various tax inspections and hardware testing in Brazil, and sell well offline—then a Huaqiangbei product might sell for five or six hundred in Brazil. Precisely because the process is complex and distant, many people who want to come arbitrage get cold feet, and at this time it can instead solidify your moat. So going global has two sides, requiring everyone to do it very seriously.

---

Original publication: https://uniqueresearch.substack.com/p/src-20260416-02html
On-site reading page: https://ffcap.cn/en/research/src-20260416-02html