Roundtable Dialogue: Beyond Borders, Within the World: The AI Process of Enterprises Going Global | 2025 Going Global Conference

On July 25th, guided by the Department of Commerce of Zhejiang Province, the Secretariat of the China Cooperation Center for Special Economic Zones of BRICS countries, the Hangzhou Bureau of Commerce, and the Qiantang District Bureau of Commerce, and jointly hosted by 36Kr and Qiantang Construction Group, the 2025 "From 'Ingenuity' to the 'World'" Overseas Expansion Conference will be grandly inaugurated at the Grand Mercure Hangzhou Qiantang. As a brand - new IP event of 36Kr focusing on the fields of globalization and overseas expansion, the conference will set up a main venue and a sub - venue, the "Investing in BRICS" - Country - Specific Cooperation Matching Meeting. The main venue of the conference will be divided into two chapters: "Finding Certainty in Uncertainty" and "Doing Business Globally", focusing on popular overseas expansion fields such as consumption, technology, e - commerce, finance, and new energy. It will cover more than 10 keynote speeches, 5 round - table discussions, and the release of the East Forward 2025 Overseas Globalization Innovation List, decoding the certain logic of the coordinated growth of "product - technology - ecosystem" and providing a referable global development path for enterprises to break through the fog of globalization and build sustainable overseas expansion capabilities.

36Kr has specially planned a round - table discussion on technological innovation and global competitiveness (AI) at the overseas expansion conference, with the theme of "Beyond the Boundaries, Within the World: The AI Process of Enterprises' Overseas Expansion". Jing Xun, the host and the host of a business column, Cao Fangning, the founding partner of Multidimensional Capital, Guo Weike, the founder of PETKIT, Zeng Yiqian, the president of the intelligent consumer business department of Yizhi Intelligence, and Ren Hangbo, the person in charge of AI innovation hardware at Tuya Smart, several industry experts will jointly discuss the "certainty" of overseas expansion.

The following is the content of the round - table discussion, edited by 36Kr:

Jing Xun: Hello everyone, I'm Jing Xun. In the past year, the AI wave has swept across the globe, and China's technological progress is obvious to all. However, overseas expansion is never just about technology. Every step, including landing scenarios, user understanding, brand building, and ecological adaptation, is a tough battle. The global market is becoming increasingly competitive, with both opportunities and thresholds. Judging from the global popularity of Deepseek, the world is looking forward to the story of Chinese AI, and even a new paradigm. Today, we will discuss with several guests: How can China win in the global AI wave in overseas expansion? First, let's invite the guests to introduce themselves.

Cao Fangning: I'm the founding partner of Multidimensional Capital. We are a well - known investment bank in the primary market. We have completed nearly $50 billion in transactions in eight years and have served many leading overseas expansion enterprises, including PETKIT here today. I'm very glad to participate in this forum.

Guo Weike: I'm the founder of PETKIT. We are a pet technology company focusing on serving cats, dogs, and pet owners. Thank you all.

Zeng Yiqian: I'm the president of the intelligent consumer business department of Yizhi Intelligence, with the nickname "Youqian". We are an AI human - computer interaction technology company, focusing on three major interaction modes: text, voice, and video. We have launched products such as AI voice calls, AI digital human live - streaming, and short - video generation, serving industries such as beauty, footwear, clothing, and food, with a leading market share in the industry. Our team has more than 300 people, and our headquarters is in Hangzhou.

Ren Hangbo: I'm the marketing director and the person in charge of AI innovation hardware at Tuya Smart. Tuya is a global AI cloud platform service provider. It has been established for 11 years. From AIoT solutions (modules, APPs, cloud) to scene intelligence, and then to AI applications, it provides one - stop AI productivity tools for developers to help AI hardware land globally.

Topic 1: From Product Innovation to Structural Capability —— How Does Technology Build Global Barriers?

Jing Xun: PETKIT won the Top Tech Award at this year's CES with its intelligent cat litter box. Could Mr. Guo share what experience upgrades AI has brought to the overseas expansion of pet products? How to form a "breakthrough force" by combining hardware technology? What are the difficulties?

Guo Weike: Although the pet industry is booming, its scale is relatively small. It's not easy to commercialize top - notch technologies. First of all, consumers' expectations for AI have exceeded the current capabilities of enterprises. Those who don't embrace AI will be eliminated. Secondly, AI brings the challenge of "increasing quantity but not price" - the cost rises, but it's difficult to increase the selling price. For example, the gross profit of our products has dropped from 45% to 24% - 25%.

The key to breaking the situation lies in "value transfer": shifting from hardware - based profit to software - service - based profit. This year, our revenue from pure software services reached 30 million, with a gross profit of 90%. This is the core of our exploration. However, in the process, we need to balance technological investment and commercialization. A small - scale industry can't expand at a loss like the new - energy industry. We must find a clear profit path.

Jing Xun: From a capital perspective, what is the biggest challenge for AI products to expand overseas? How to judge whether an enterprise has global competitiveness rather than just replicating the domestic path?

Cao Fangning: The challenges vary in different tracks:

AI Infra enterprises: They need to focus on specific regions, balance geopolitics and data security, and plan long - term market strategies;

AI Agent enterprises: Born global, they need to clearly define their user positioning, make a choice between short - term interests and long - term ecosystem, and adapt product design and payment logic to the target market;

AI + intelligent hardware enterprises: Chinese enterprises have strong supply - chain and product - definition capabilities, but they need to be vigilant against "replicating the domestic path" and adjust according to overseas users' habits.

The core of judging global competitiveness is whether an enterprise understands the real needs of overseas users rather than simply transplanting technology. For example, pool robots have become popular in Europe and America in recent years because the supply side has cultivated users' habits, and the cost and pricing have reached a critical point.

Jing Xun: As a platform - type enterprise, how does Tuya help customers build AI competitiveness in different markets?

Ren Hangbo: More than 80% of Tuya's "Powered by Tuya" devices are for overseas markets. The value of its core capabilities is reflected in three aspects:

Compliance first: It covers regulations such as the EU's GDPR and the US's CCPA, so customers don't need to make additional investments in privacy compliance;

Interconnectivity: Devices of different brands and categories can interact through a unified platform, enabling users to operate without perception;

Agile development: Standardized AI tools help customers quickly implement projects. For example, an AI hardware product can be developed within one month to meet the speed competition in globalization.

Topic 2: Beyond Technology —— The Global Market's Acceptance and Matching Mechanism for AI

Jing Xun: As a technology service provider, how do you judge the "feasible implementation window" of AI in different markets? How to help enterprises localize?

Zeng Yiqian: The global acceptance of AI is generally open, but adaptability is the key:

Developed markets (Europe, America, Japan, and South Korea): With high labor costs and a strong willingness to pay for SaaS, brands attach great importance to membership operations, and the demand for AI to improve efficiency is clear. For example, for L'Oréal and Estée Lauder we served, AI outbound calls reached hundreds of millions of users, and the call duration and the rate of adding fans were both better than those of manual calls;

Emerging markets (Southeast Asia): With low labor costs, the initial advantages of AI are not obvious. It's necessary to combine with local teams to reduce service costs.

The core of localization is "localizing the team": Just as overseas enterprises use local teams when entering the Chinese market, we deploy local teams in the target markets to adapt to language, culture, and compliance requirements.

Jing Xun: Could Mr. Ren give an example of the "pitfalls" in enterprises' overseas expansion? How to help customers quickly adapt to different markets?

Ren Hangbo: A common problem is "over - self - research". For example, a foreign customer built its own team for a non - core category of smart heaters. They couldn't solve the software - hardware adaptation problem in 3 - 4 months, but we solved it within one month on - site. Small and medium - sized customers should make good use of platform capabilities to avoid reinventing the wheel.

Another pitfall is "function redundancy": A desktop robot team spent 4 months polishing 5% of niche functions, which delayed the product launch. AI hardware should "take small steps and run fast", first meet the core needs, and then iterate and upgrade.

Our solution is the Tuya.AI platform: Customers can seamlessly switch between domestic and foreign large - scale models (such as DeepSeek and Doubao in China, and OpenAI and Gemini overseas) with one - time development, reducing the cost of multi - market adaptation.

Topic 3: Improving Efficiency in R & D and Operation —— Where is the Leverage Point of AI Value?

Jing Xun: Mr. Guo, how is AI applied in PETKIT's operation and R & D? What changes has it brought?

Guo Weike: We applied the AI large - scale model in our customer service system. We input product problems and user concerns into the model to assist front - line customer service staff in responding quickly and empower the dealer team. However, we found that users (especially high - end customers) still rely on manual service. Therefore, AI is positioned as a "tool to serve customer service" rather than a replacement.

The effect is remarkable: the pressure on customer service has decreased, the response efficiency has increased, and at the same time, the training cost and emotional fluctuations of pure manual service have been avoided. This is the core value of AI in operation - not to pursue "subversion" but "efficiency improvement".

Jing Xun: What "ineffective cases" of AI deployment have you observed? Where does the real cost - reduction and efficiency - improvement of AI lie?

Zeng Yiqian: The core of ineffectiveness is "being divorced from business scenarios" - only emphasizing the model's capabilities but being unable to quantify the effects. Valuable AI needs to meet the following requirements:

Controllable and supervisable: Avoid "large - scale model hallucinations". For example, AI outbound calls focus on welfare notifications (orders, logistics) rather than after - sales disputes;

Data quantifiable: For example, the AI call duration increases from 30 seconds to 45 seconds, the private - domain fan - adding rate increases by 20%, and the cost is reduced by 30%;

Implementing in specific scenarios: Digital human live - streaming reduces the cost of a single event from 2 - 5 million to 100,000, and at the same time reaches tens of millions of users, which is impossible for traditional models.

Topic 4: The Underlying Logic of the Next Round of Competition —— How Can "AI +" Produce Globalized Enterprises?

Jing Xun: Mr. Cao, do you think that "AI +" can produce real globalized enterprises? What are the key elements?

Cao Fangning: AI mainly solves the problems of efficiency (toB) and experience (toC). Chinese enterprises have great opportunities:

On the toB side: Enterprises need to choose the right paying industries (such as industry and energy), control the performance cost, and avoid "shell - style innovation";

On the toC side: Enterprises need to deeply cultivate user experience and design products in combination with local culture.

The key to globalization lies in the talent echelon: Chinese and overseas teams need two sets of product logics, and the founder needs to have cross - cultural management capabilities.

Jing Xun: Mr. Guo, what are PETKIT's key points for overseas expansion in the next stage? How to balance "increment" and "stock"?

Guo Weike: Overseas expansion is the first step towards globalization, and a global brand needs "cultural integration" and "increment creation":

Understand cultural differences: Overseas users are skeptical about "free services" and prefer the privacy protection and long - term service brought by paid services, which is different from the domestic logic;

Create market increment: Focus on pet health AI technology, don't disrupt the existing rules, and avoid direct competition with local enterprises. For example, create new demands through health monitoring instead of squeezing the stock market with supply - chain advantages.

Jing Xun: Mr. Ren, how does Tuya help customers transform AI into increment? What kind of intelligent manufacturing enterprises can succeed in the future?

Ren Hangbo: The key to customers' success is "differentiation": For the same category, it's necessary to accurately position (such as AI toys targeting 6 - 8 - year - old children vs. Generation Z women). We help customers quickly iterate on featured functions through the TuyaOpen open - source solution to avoid homogenization.

Facing the challenges of AI - native applications, the platform needs to maintain an "open ecosystem": Connect large - scale models, developers, and scenarios, so that customers can focus on innovation rather than repeatedly solving technological infrastructure problems.

Today's discussion reveals that overseas expansion of AI is not just about technology. It also requires understanding culture, compliance, and the real needs of users. From hardware to service, from R & D to operation, the opportunity for Chinese enterprises lies in transforming AI into an "increment tool" rather than replicating the past path. We look forward to more Chinese AI enterprises standing on the global stage as "innovators" in the future. Thank you, dear guests!