Most people’s impression of the MRO platform is that of an “e-commerce company that sells screwdrivers and lubricants” and relies on scale to make price differences.
However, in the first quarter, which is the traditional off-season, the data delivered by Zhenkunxing are very eye-catching: GMV reached 2.45 billion yuan, a year-on-year increase of 12.9%, and adjusted net profit surged 103.4% year-on-year, achieving first-quarter profits for the first time.
If you include the more than 6-fold increase in overseas revenue, the value of this company is obviously not in the transaction volume, but in a technology it has just released – the industry’s first vertical AI large model for industrial supplies and the supporting petabyte-level data engine.
This is a clear identity switch: Zhenkunxing is no longer a procurement platform with SKU richness as its core selling point. It is becoming an infrastructure that provides intelligent decision-making for the industrial supply chain.
Full-stack AI cannot just do selection, it must start from the data base to the last mile.
The key to determining the value of a company’s AI is not whose model it uses, but whether it has its own closed-loop data source.
Zhenkunxing’s “Domino” data engine is the industry’s first petabyte-level industrial supplies data engine.
Its underlying data composition is the real barrier to competition: in addition to the product parameters of 27 million SKUs, it integrates the purchasing behavior of 180,000 companies, the performance records of 32 general warehouses, and the real-time receipt data of more than 6,000 production line smart small warehouses.
This multi-dimensional data base allows the training accuracy of the “Expert Linglong” large model to be 47% higher than that of the general large model.
Compliance advantages constitute an entry barrier to external competition.
“Expert Linglong” is the first large-scale vertical model of industrial supplies in China that has been registered with the Cyberspace Administration of China.
This means that while industry rivals are still applying for registration, Zhenkunxing has completed complete commercial deployment from model to scenario.
Its supporting visual search engine “Expert Eyes” can compress material identification time to seconds, and the accuracy of complex selection reaches 92%.
The purpose of this AI layout is not to make cool product demonstrations, but to allow manufacturing customers to go from “spending several days to find a screw” to a second-level response.
When a platform company announces technological transformation, the market is usually worried that high investment will erode profit margins.
But Zhenkunxing’s first-quarter data points to another path: AI not only did not eat up profits, but became the core driver of profit reversal.
In the first quarter, there were 66,000 customers, a year-on-year increase of 11%.
The strongest growth is not a certain type of customer, but all customer groups: the GMV growth rate of small and medium-sized customers exceeds 20%, state-owned enterprise customers return to double-digit growth, and the growth rates of the four core tracks such as electrical manufacturing all exceed 20%.
The logic behind this is that AI has filled the biggest efficiency black hole in the MRO industry – the strong reliance on manual experience in the procurement process.
In the past, purchasing an industrial automation component required engineers to compare parameters repeatedly, communicate between different departments by email, and waste several days in one selection.
Zhenkunxing’s “AI Material Butler” turns experience-driven into data-driven. More than 60 AI agents launched in the first quarter released more than 2,000 manpower hours per month. The goods-to-person picking efficiency of Qingpu smart warehouse is more than three times that of the traditional model.
Intel China General Manager Wang Jingjia pointed out a trend at the meeting: the focus of competition in the industrial supplies industry is shifting from SKU quantity and price to technical capabilities and scenario development.
Zhenkunxing’s full-stack AI capabilities are essentially reducing the marginal cost of serving customers on a large scale – the more customers there are, the denser the data, the more accurate the model, and the lower the service cost.
This is a moat that traditional dealers relying on manual experience cannot build.
The weakness of general-purpose large models is that they are general but not precise, while Zhenkunxing places its bets on the professional MRO field.
The growth of professional categories on the platform in the first quarter proves this judgment.
The GMV of factory automation, industrial lubricants, and chemical reagents all achieved double-digit growth, and the proportion of private brands rose to 9.7%.
This set of data shows that customers do not come to the platform to compare prices, but to find a set of “knowledgeable” solutions – this is where the generation gap between large vertical models and general models widens.
The explosion of overseas business provides another layer of value distribution.
The 6-fold growth rate does not come from going overseas at low prices, but from two parallel strategies: one is to serve Chinese manufacturing companies to go global, and the other is to deploy business in the United States, with the goal of achieving overseas breakeven in 2026.
This means that Zhenkunxing’s AI capabilities are not just “dialects” that can only adapt to the domestic supply chain, but are forming replicable global delivery standards.
Xiao Yongbo, a professor at Tsinghua University, proposed at the meeting that companies should build a trinity supply chain of “efficiency, resilience, and ESG.”
Zhenkunxing’s AI layout answers this question exactly: efficiency comes from data-driven, resilience comes from global distributed fulfillment, and the human cost released by AI itself is a kind of organizational carbon reduction.
Zhenkunxing is essentially becoming a data service company for industrial supplies, and transactions are only the final form of data output.
When manufacturing companies’ purchasing decisions increasingly rely on intelligent selection rather than manual experience, this full-stack AI infrastructure will become an underlying capability that they cannot replace.
