Two industrial bearings may look nearly identical, yet a small difference in specifications can determine whether a component is suitable for a particular machine. Identifying such differences, and verifying them against delivery documents, remains a challenge in industrial procurement and warehouse operations.
This is one of the problems that LingLong Insight, a lightweight industrial multimodal AI model, is designed to address. Developed by LingLong Intelligence and powered by Intel's edge computing technology, the model combines visual recognition with industrial knowledge to help workers identify materials, verify specifications and detect potential discrepancies.
But the story behind the technology extends beyond AI. LingLong Intelligence is a subsidiary of ZKH Group Limited (NYSE: ZKH), a Chinese company with nearly three decades of experience in industrial supplies. Listed on the New York Stock Exchange since 2023, ZKH describes itself as a leading maintenance, repair and operations (MRO) procurement service platform in China.
The company's expansion into AI raises a broader question: How can the industrial knowledge, supply chain capabilities and manufacturing experience accumulated in China translate into competitive advantages in global markets?
To explore this question, Yiran Xing, Research Director at EqualOcean, spoke exclusively with Yang Liu, Chief Technology Officer of ZKH and CEO of LingLong Intelligence, about the company's international expansion, the changing needs of industrial customers and the role of AI in cross-border manufacturing.
Before joining ZKH in 2023, Liu held technology and management positions at Alibaba and Lenovo, spanning e-commerce, cloud computing and product development. His experience across consumer internet businesses and industrial procurement offers a perspective on how digital technologies are being adapted to manufacturing—and how those capabilities might travel beyond China.
The following interview has been edited and condensed for clarity.
01. Two Paths to Global Markets
For ZKH, international expansion is already part of its business operations, rather than a distant strategic ambition.
The company has adopted two approaches. In emerging markets, including Southeast Asia, South America, Africa and the Middle East, it primarily supports Chinese manufacturers expanding their production operations overseas. In developed markets such as the United States, it is pursuing a more localized model aimed at serving domestic businesses directly.
According to ZKH's 2025 annual report, its fulfillment network supporting Chinese manufacturers abroad had reached 17 countries as of March 31, 2026.
In the United States, ZKH launched NorthSky Supply, its U.S. MRO e-commerce platform, in December 2024. Targeting local small and medium-sized enterprises, the platform offered more than 1,000 stock-keeping units (SKUs) as of March 2026 and was supported by seven local warehouses.
ZKH also reported that its international gross merchandise value (GMV) increased tenfold year over year in the first half of 2026, driven by efforts to support Chinese manufacturers overseas and expand localized operations in the United States.
Yet these two approaches require different capabilities. Following existing Chinese customers overseas allows ZKH to build on established business relationships, while serving local American businesses requires the company to develop customer relationships, distribution capabilities and after-sales services suited to the local market.
EqualOcean: ZKH serves both Chinese manufacturers expanding overseas and local businesses in international markets. What are the most significant differences between these two customer groups?
Yang Liu: We have observed that local overseas customers and Chinese companies operating abroad have different priorities.
Local customers tend to focus more on delivery speed, product quality and user experience. Chinese manufacturers expanding overseas, on the other hand, are more concerned about integration with their existing supply chains in China, the synergies they can achieve and their overall costs.
Another interesting observation is that overseas customers are increasingly interested in intelligent industrial products.
Take power tools as an example. Traditional tools perform a mechanical function, but intelligent tools can also record operational data, such as the torque applied when tightening a bolt. This information can later support product maintenance, lifecycle management and predictive services.
In that sense, intelligent functionality is becoming an important part of the value proposition for industrial products.
EqualOcean: Beyond exporting products, what has ZKH learned about building a localized business in the United States?
Yang Liu: Over the past nearly two years, we have explored a model that combines the advantages of China's cross-border supply chains with localized operations.
We have established a local team in the United States so that we can stay closer to customers and respond to their needs.
We are also developing product categories suited to local usage scenarios, including warehouses, garages, workshops and personal protective equipment.
In personal protective equipment, for example, we have accumulated capabilities covering product research, design, manufacturing and supply chains. Vertical integration allows us to compete not only on costs but also on product quality.
02. The Hidden Complexity of Cross-Border Industrial Procurement
In EqualOcean's conversations with Chinese companies expanding overseas, a recurring challenge is that relocating production capacity is only one part of establishing operations abroad. Manufacturers also need to rebuild procurement processes, establish supplier networks and adapt their operational systems to local requirements.
Industrial supplies encompass thousands of categories, from equipment components and chemicals to tools and protective gear. Differences in technical standards, product descriptions, supplier networks and delivery arrangements can make cross-border procurement more complicated than the physical transportation of goods.
EqualOcean: In recent years, we have seen more Chinese manufacturers establishing production facilities overseas. What new challenges are they encountering in supply chain management?
Yang Liu: When our existing customers establish factories in Southeast Asia or Europe, they often need to rebuild their procurement systems to suit local manufacturing conditions.
First, they need to redesign procurement processes to meet local service requirements, warehousing practices and logistics standards.
Second, they need to establish corresponding IT systems, including procurement and enterprise resource planning systems. These can be more complex than the systems they use in China.
An overseas factory may source products not only from China but also from local suppliers. It also needs alternative suppliers to ensure supply continuity.
This creates substantial challenges in aligning information across languages, standards, suppliers and product models.
In our experience, many of the problems we help customers solve are essentially problems of translation.
EqualOcean: Given ZKH's experience in China's manufacturing sector, where do you see its advantages compared with established international industrial suppliers?
Yang Liu: I think we need to look at the broader development of global manufacturing.
China has accumulated an extraordinary range of industrial scenarios, professional talent and engineering experience. These provide a strong foundation for training industrial models and developing intelligent applications.
Over nearly 30 years, most of our business experience has been accumulated in China. But as Chinese manufacturing and supply chains globalize, we are seeing something beyond the movement of physical products.
Industrial knowledge, industrial experience and industrial intelligence are also becoming global.
Consider the differences between Chinese, European, American and Japanese standards. Then add different product descriptions, currencies, pricing systems and transportation costs.
The physical movement of a component may be relatively straightforward. What makes the transaction complicated is the information and expertise associated with that component.
This is why I describe it as a translation problem. Just as people speaking different languages need translation, cross-border manufacturing requires different forms of industrial information to be understood and matched.
03. From Industrial Knowledge to AI Capabilities
The complexity of industrial procurement also helps explain why ZKH has been investing in AI.
According to the company's November 2026 announcement on LingLong Insight, ZKH has accumulated more than 27 million industrial product records and approximately 4 billion product parameters. Its AI subsidiary seeks to turn this information, together with technical standards and specialist knowledge, into structured data that can support industrial models and applications.
LingLong Intelligence was established as a dedicated AI subsidiary in 2026. Liu describes its approach as comprising four layers: data, models, AI agents and consulting services.
The data layer focuses on converting fragmented, inconsistent industrial information into structured and usable data. Specialized models build on this foundation to understand industrial products and business rules, while AI agents apply these capabilities to tasks such as materials management, product search and warehouse verification.
The fourth layer addresses what can be a harder problem: getting AI to work inside a company. Employees and established processes may resist change, even after the technology is ready. In an earlier interview with China Industry News, Liu described such organizational resistance as one of the barriers to AI adoption.
LingLong's consulting services aim to bridge this "last mile" by helping companies refine existing workflows and adapt their operations to integrate AI into day-to-day business practices. The approach is similar to the work of forward-deployed engineers (FDEs), who help customers integrate AI solutions into their existing systems and operations.
LingLong's published customer cases illustrate this approach. In one materials management project for a large bio-food ingredients company, LingLong reported standardizing 120,000 historical material records across multiple sites and reducing duplicate materials by 30%. The project also involved aligning multilingual and differently formatted records under a unified system of material descriptions and codes.
For Liu, who spent nearly nine years at Alibaba before joining ZKH, the transition from consumer internet businesses to industrial technology has highlighted a fundamental difference in how digital services create value.
EqualOcean: Having worked in consumer internet businesses before joining ZKH, what do you see as the biggest difference between applying technology to consumer platforms and industrial procurement?
Yang Liu: Internet platforms primarily addressed connectivity, making it easier for buyers and sellers to find each other and improving the efficiency of matching supply with demand.
Industrial procurement also involves matching supply and demand, but the requirements for accuracy are much higher. Decisions can directly affect production lines, product quality, equipment performance and even personal safety.
Another difference is how technology creates value. In the AI era, every interaction consumes computing resources, so the focus is no longer simply on attracting users' attention, but on whether AI can solve real problems and deliver value worth paying for.
While general-purpose AI provides powerful capabilities, applying it effectively in industrial settings requires specialized data, industry standards and an understanding of individual businesses. Without this domain-specific knowledge, general AI may struggle to deliver reliable results or justify its costs.
EqualOcean: You mentioned that incorrect product selection can affect production quality and safety. How can industrial AI address such risks?
Yang Liu: The key is accurate product selection.
In industrial environments, you need to consider the people involved, the equipment, materials, operating methods and environmental conditions.
For example, a component may need to withstand particular temperatures, pressure levels, loads or corrosive environments.
Traditionally, decisions like these relied heavily on the experience of engineers and specialists.
What we are doing is structuring industrial expertise and integrating it into our models and AI agents, so they can provide product selection recommendations supported by relevant technical knowledge.
Traceability is essential. Large language models can generate hallucinations, but industrial applications require reliable reasoning.
A model can make inferences, but the information supporting those inferences must come from trustworthy sources, such as national standards, industry standards and company-specific requirements.
That is the foundation we are trying to build.
04. Taking Industrial AI Global
ZKH's existing international operations could provide LingLong Intelligence with an initial route into overseas markets. However, the two businesses are at different stages of international expansion.
While ZKH has established overseas procurement and fulfillment operations, LingLong Intelligence is at an earlier stage of developing its international AI business. It has begun exploring opportunities with multinational manufacturers and technology partners, but has yet to build out a comparable overseas customer and service network.
EqualOcean: What gives LingLong Intelligence confidence in its potential to expand internationally?
Yang Liu: One indication of the opportunity we see internationally comes from our customer base over the past few months. The number of customers representing foreign brands and domestic Chinese brands has become roughly balanced, which exceeded my expectations.
To me, this indicates that international brands recognize the value of our industrial data, specialized knowledge and AI capabilities.
It has strengthened our confidence that industrial intelligence, much like manufacturing supply chains, can cross national boundaries.
EqualOcean: What factors will guide LingLong Intelligence's selection of international markets?
Yang Liu: We are looking at regions with an established industrial base, particularly where Chinese supply chains are expanding and where there is demand for complementary services.
As manufacturing supply chains expand internationally, companies also need digital services, intelligent services and industrial knowledge.
We want to develop our business in response to those needs.
The planned expansion also raises questions about how much of LingLong's existing industrial knowledge can be applied outside China.
Liu believes many industrial problems share common characteristics across markets, although local standards and operating conditions still require additional information.
EqualOcean: Much of ZKH's industrial knowledge has been accumulated in China. How applicable is that knowledge to overseas markets, given differences in industrial standards and operating environments?
Yang Liu: Many industrial scenarios can be generalized across markets.
For example, one of ZKH's earliest product categories was industrial chemicals, including adhesives.
If you are dealing with a damaged floor, you need to understand the material of the surface and the required strength and bonding characteristics before choosing the right product.
These underlying engineering problems are often similar across different regions.
China's extensive manufacturing scenarios provide a strong foundation for training models. That gives us a useful starting point when applying the technology elsewhere.
Of course, different markets have their own requirements and conditions. Those are the missing pieces we need to understand and gradually incorporate as we serve customers.
EqualOcean: How could LingLong Intelligence use ZKH's overseas network to reach international customers?
Yang Liu: In China, we have been able to bring intelligent products and services to industrial customers through ZKH's established sales network.
We believe a similar approach can work overseas.
ZKH already has local service capabilities in markets such as the United States. Its current overseas business is mainly focused on industrial supplies, which requires a different type of after-sales support from AI solutions.
As LingLong develops its international business, we can build on ZKH's existing customer relationships and sales channels, while developing the additional capabilities needed to support AI deployment.
Industrial supplies can be sold through established distribution channels. Industrial AI, by contrast, often requires integration with customer systems, specialized support and ongoing adaptation to local data and operational requirements.
ZKH has already begun building the local infrastructure required to compete in overseas industrial procurement markets. LingLong's international business remains a plan for the next stage, with its commercial model and scale still to be tested abroad.
For Chinese industrial companies seeking to expand globally, the opportunity may increasingly extend beyond exporting physical products. Their accumulated manufacturing experience, technical knowledge and digital capabilities could also become part of what they offer international customers.
The more demanding test will be whether those capabilities can solve practical problems for customers operating under different standards, systems and expectations.
For ZKH, that is where the next chapter of its international expansion is likely to be written.