---
title: "For AI Companies, Going Global Is Not Plan B—It Is the Only Growth Curve"
author: "Unique Research"
sourcePublication: "Unique Research Substack"
originalPublishedAt: "2025-11-09T09:32:00+00:00"
canonical: "https://ffcap.cn/en/research/unique-research-2025-11-09-04"
source: "https://uniqueresearch.substack.com/p/unique-research-2025-11-09-04"
language: "en"
---

# For AI Companies, Going Global Is Not Plan B—It Is the Only Growth Curve

_Original · Unique Research · 2025-11-09_

_Editorial note: This historical article preserves Li Shouguo’s company figures and market views as reported at the time. “Last year” is retained from the interview; the source gives no separate interview date. The discussion of local hosting and government certifications reflects the interviewee’s experience, not a universal legal requirement. In Q10, “status” translates an unspecified term in the source and does not establish a particular immigration status. Beta Data, AI Sales Coach, and Digital Avatar are source-based English renderings. The omitted ordinary stage portrait is labeled CAAIEC: China (Guangxi)–ASEAN Artificial Intelligence Enterprise Conference and carries no additional business facts._

While many AI founders are still preoccupied with whether they should build large models, Li Shouguo, founder of Beta Data, has already moved past that question and chosen to head straight overseas.

This interview has neither dramatic twists and turns nor a dazzling parade of valuation figures. It feels like a calm retrospective, yet it leaves much to reflect on. That is because it reveals a mature entrepreneur’s profound understanding of speed, globalization, and standardization after navigating product transformation, the trap of customization, industry cycles, organizational contraction, and finally a determined push overseas—an understanding more worthy of attention than ten thousand stories about AI funding.

I. The Pain and Insight of Moving from Projects to Products

When asked whether he regretted taking on customized projects, Li Shouguo did not hesitate. His answer was: “If I could do it again, I would commit firmly to standardized products.”

In its early days, however, Beta Data took on plenty of project work. This was especially true with state-owned and centrally administered state-owned enterprise clients: once the company was on their procurement lists, it felt wasteful if that access did not lead to project work. Unfortunately, things never unfold as imagined. Projects pursued for sales revenue usually generated little profit and even disrupted the organization’s rhythm and underlying DNA.

The observation that stayed with him most was: “You used to think you could make everything work by designing a perfect solution. Now you have to validate quickly—and shut it down if it does not work.”

This is the fundamental change of the era. The emergence of large AI models has reduced the cost of validation and compressed iteration cycles. Lengthy polishing and validation are no longer necessary; every product now requires rapid judgment and decisive life-or-death calls. The mindset of slowly perfecting an ideal system is obsolete. Today’s product playbook is to identify quickly the core that can stay alive, then build around it and survive.

II. Going Overseas Is Not the Icing on the Cake, but a Second Life

For Beta Data, going overseas is not a Plan B that might be worth trying. It is a strategic restart that has to happen.

The domestic market remains in a cost-cutting cycle, industries are contracting conservatively, and traditional sectors such as finance are especially cautious. Overseas markets have become one of the few remaining arenas for incremental growth.

The company chose highly portable AI products: an AI sales coach that helps sales teams improve how they communicate, and a digital avatar product that clones a salesperson’s likeness to generate videos. Both are highly generalizable and directly address overseas pain points of high labor costs and heavy dependence on marketing.

Li Shouguo’s thinking is clearly not confined to figuring out how to export Chinese experience. He is rebuilding a product system designed natively for overseas use cases. From day one, for example, the team considers whether the aesthetics and interface interactions fit European and American markets; the front end centers on minimalism and single-purpose functionality; and language is not merely a translation issue, but a reconstruction of user habits and psychological expectations.

As he put it, “Some products can even target overseas markets exclusively from day one. Validating them in China would simply add unnecessary cost.”

This is a dilemma for many companies expanding abroad: should they build in China first and then go overseas, or go all in globally from the outset? Li Shouguo’s answer is that it depends on the market and payment habits the product is inherently built for. If it is a PLG product and its users’ willingness to pay is already concentrated overseas, why take a detour?

III. PLG Is Not a Master Key, but It Gives Chinese Founders a Downhill Path

In China’s predominantly enterprise-software startup ecosystem, PLG—product-led growth—was marginalized for a long time. Overseas, however, it has instead become a way for many Chinese teams to break through with less baggage.

Why? Because PLG lets teams avoid an initially expensive, slow, and culturally misaligned sales-led process. Allowing customers to use a product first and then decide whether to pay is a culturally more acceptable way to test the waters.

Li Shouguo does not mythologize PLG. He points out with great clarity that, for founders from China, PLG merely lowers the entry barrier; it is not the final destination for growth. But outside Chinese-speaking markets, PLG at least lets a company find an initial usage signal and verify whether anyone is genuinely using the product.

Once a company truly needs to serve local customers in depth, Beta Data’s approach is equally clear: its own team goes in first to validate the market; as a deal approaches completion, the customer is handed to a partner; and channel-based delivery follows later. This is not about saving money, but about developing a firsthand feel for the market.

The warning that “you cannot build an overseas B2B business through business trips from China alone” should stay with every company venturing abroad.

IV. On Implementing AI, He Used Just One Word: Process

When discussing the path of AI’s evolution, Li Shouguo offered a simple but fundamental sequence: Coach → Copilot → Agent, or replacement.

Many companies skip the first two stages and try to move directly to replacement, only to trigger customer resistance and organizational rejection. He cautions that technology changes quickly, but customer mindsets migrate slowly. At large enterprises in particular, procurement often originates in a business unit. It is acceptable to talk about cutting costs and raising efficiency, but the moment employees hear that AI will replace them, the project falls apart.

Instead of frightening customers, use AI first to improve efficiency, then to assist people, and only eventually, perhaps, to replace some work.

This also reflects the mindset AI founders themselves need: do not expect to become successful in a single leap.

V. If He Could Start Again, He Would Study Abroad First

This was the most moving answer in the entire interview.

Asked what he would do if he could start over, Li Shouguo said, “I would study abroad first.” Not for the credential, but to become familiar with the environment and build a globally oriented mindset and team structure from the very beginning.

He knows that native-born Chinese founders who want to build global products cannot rely on logic and technology alone. They also need time to understand how the world works, the language of markets, and the rhythms of different cultures. The simplest way is to live there—to immerse yourself first.

This is not merely preparation for going overseas. It is a way to become an entrepreneur of the world.

Conclusion: Going Overseas Is the Romance of the Solitary

Many people say going overseas is a business. In Li Shouguo’s account, however, it sounds more like a posture and a choice of how to live.

You can neither rush for quick results nor stop moving forward. You must understand both the universality of technology and the specificity of culture. You must experiment quickly while respecting the process.

Beta Data may not have been among the first companies to go overseas, and it may not yet command the high ground in the market, but it is mapping its own path step by step in a thoroughly professional way. That quiet yet determined rhythm is precisely one of the qualities in shortest supply among today’s AI founders.

Going overseas is not merely a business decision. It is also a process of redefining the boundaries of both oneself and one’s company.

The Chinese companies that ultimately win the world may not be the loudest, but they will certainly be the clearest-eyed.

Selected Interview Q&A

Q1: What is the current scale of Beta Data’s business and workforce?

Li Shouguo: The company currently has about 160 employees. In terms of business scale, last year’s revenue was approximately RMB 100 million, down somewhat from before. This was mainly because we proactively adjusted our business mix and cut several semi-customized projects with relatively low gross margins so that we could focus more closely on standardized products. The market environment also had an impact: some of the financial-institution clients we serve are in a cost-cutting cycle, which meant that certain projects could not continue.

Q2: What were the biggest challenges or traps you encountered as the company developed, and what lessons did you learn from them?

Li Shouguo: I think there were two main traps. The first was taking on a large number of customized projects in pursuit of sales revenue. At the time, we believed client resources—especially relationships with state-owned and centrally administered state-owned enterprises—were hard to obtain, so we accepted a great deal of work. We later discovered that the company’s DNA, team, and processes were not actually suited to it, resulting in low gross margins and high management costs. If I could do it again, I would commit to standardized products from the outset. The second lesson is that your validation mindset must be fast. Especially today, with large-model technology becoming widespread, the cost and time required to validate a product idea are falling sharply. Rapid experimentation, rapid decisions, and shutting something down immediately when it does not work matter far more than repeated hesitation and excessive validation.

Q3: Why has Beta Data made globalization and going overseas a core strategy for the next 3–5 years?

Li Shouguo: We firmly believe globalization, or going overseas, represents the market’s largest source of incremental growth today. The domestic market—particularly the state-owned and centrally administered state-owned enterprise clients we know well—is in a cost-cutting cycle, and the room for future growth is uncertain. By contrast, overseas markets, especially in AI applications, still offer enormous opportunities and growth potential. Looking to the world is therefore an inevitable choice for us over the next several years.

Q4: Which two AI products is Beta Data currently promoting in overseas markets, and what problems do they solve?

Li Shouguo: The two products we are promoting overseas both belong to the digital-marketing field:

“AI Sales Coach”: This is an enterprise product aimed mainly at companies with large numbers of frontline communicators, such as salespeople, store associates, and customer-service representatives. By using AI to simulate conversations with real customers, it helps these employees train and improve their skills, addressing the low efficiency and high cost of training new hires.

“Digital Avatar”: This product is oriented more toward prosumers—individual consumers who use products professionally. It enables users such as real-estate agents, insurance agents, and medical-aesthetics consultants to generate short marketing videos quickly using their own likeness. A user only needs to enter a script, and the system clones a digital-human version of that person to deliver it, greatly reducing the time and effort required for filming with a real person. It is especially useful for managing audiences through a business’s own customer channels (“private-domain traffic”).

Q5: How do you see AI evolving in enterprise applications, and how do Beta Data’s products reflect that path?

Li Shouguo: I believe AI adoption in enterprise scenarios will pass through three stages, and our product portfolio follows the same logic:

The first step is Coach: AI serves as a coaching tool that helps people grow and improve their skills. That is the positioning of our AI Sales Coach product.

The second step is Copilot: AI serves as an assistive tool that helps people within their workflows and improves efficiency.

The third step is Agent: AI ultimately replaces human work in certain parts of a process. We believe that if you jump directly to the third step, customers—especially business units—will have very low acceptance because they will worry about being replaced. Beginning with coaching and assistance therefore offers a smoother path that better matches both today’s technological capabilities and the market’s willingness to accept them.

Q6: How do the target customers and go-to-market models differ between AI Sales Coach and Digital Avatar?

Li Shouguo: Their customer profiles and go-to-market models are completely different.

AI Sales Coach is better suited to medium-sized and large enterprises because they usually have dedicated training departments and scaled sales teams of 50–100 people or more. They have both the budget and the willingness to pay for employee training, so the product uses a traditional enterprise sales-led model.

Digital Avatar is better suited to individual professionals. It follows a PLG—product-led growth—model, and either individuals can pay for it themselves or companies can purchase it for their employees.

Q7: What market-entry and channel strategy do you use when expanding into overseas markets such as Southeast Asia?

Li Shouguo: We use a strategy of proving the model through direct sales and delivering through channels. When we first enter a new market, our own team initially works with several direct customers so we can experience the market firsthand, validate the product, and look for PMF—product-market fit. When a project is close to being signed, we hand that customer and subsequent service to a local partner we have identified. This approach lets us maintain a frontline feel for the market while giving partners concrete success stories that strengthen their confidence. Ultimately, local service and delivery still need to rely on local channels because they understand the culture better, while doing everything ourselves would also be too expensive.

Q8: What specific localization and compliance challenges do Chinese AI companies face when going overseas?

Li Shouguo: The challenges are enormous. First, customer trust and data compliance are the biggest barriers. Overseas customers, especially enterprise customers, care deeply about data security. You must deploy data services on local servers in their own country or region and ideally obtain relevant certifications from the local government, or it will be difficult to win their trust. We chose to expand overseas alongside major cloud providers, which can address this issue to some extent, but the entire process is extremely time-consuming. Selecting and deploying cloud services alone took us six months. The second challenge is product interaction and aesthetics. Overseas users’ UI/UX preferences differ greatly from those in China. They favor products with a single function and a clean interface, so the front end must be localized thoroughly rather than copied directly from the domestic version.

Q9: Why do you recommend the PLG—product-led growth—model more strongly for Chinese companies going overseas?

Li Shouguo: Because a traditional enterprise sales-led model is extraordinarily challenging overseas, especially for a Chinese team unfamiliar with the local culture. You need to build local sales, presales, and customer-success teams. That process is extremely long and expensive, and it severely tests a founder’s patience. A PLG model has a relatively lower entry barrier: the product itself is the best salesperson and can reach a broader audience through online channels. Users can begin by trying it, and the value of the product then drives growth and payment. That approach is better suited to Chinese teams testing a market with limited resources at the outset.

Q10: If you could return to the beginning of your entrepreneurial journey, what advice would you give yourself in today’s environment?

Li Shouguo: I would study abroad first. This would not be merely for status; most importantly, it would let me become deeply familiar with the overseas business environment, social culture, and user habits. Then, from the first day of the company, I would assemble a global team and design and build the product directly to global standards, rather than making a Chinese version first and an overseas version later. In terms of product direction, I would lean more toward PLG or consumer utility products because they are highly generalizable, can charge both enterprises and individuals, and are a better fit for Chinese founders.

Q11: How do you view the future development of AI application markets in China and overseas?

Li Shouguo: I believe overseas markets, especially at the AI application layer, will maintain steady growth over the next few years and offer a great many opportunities. The domestic market is harder for me to predict because many companies today—particularly the state-owned and centrally administered state-owned enterprises I know well—are in a cost-cutting cycle and will be more cautious about investing in new technology. It remains unclear when the next growth cycle will arrive. Overall, then, making a determined move overseas is a very clear and important direction for Chinese AI companies today.

---

Original publication: https://uniqueresearch.substack.com/p/unique-research-2025-11-09-04
On-site reading page: https://ffcap.cn/en/research/unique-research-2025-11-09-04