Kingdee International (00268): Subscription revenue increased 20.4% in the first half of 2026, and commercialization of AI native products accelerated

In the past ten years, Golden Butterfly (00268) solved “how to move software from local deployment to the cloud”; now, the problem facing the company has become “what else is ERP in the AI era”. In the first half of 2026, the answer given by Kingdee began to become clear: traditional SaaS continued to grow, subscription revenue continued to grow by more than 20%, while AI-native products began to enter customer procurement decisions and grow at a much higher rate than traditional businesses.

In terms of numbers, Kingdee is experiencing two things at the same time: on the one hand, the subscription business left over the past ten years of cloud transformation continues to grow and continues to release operating leverage; on the other hand, the AI-native business is growing at a much higher rate than traditional SaaS and is beginning to actually contribute revenue.

This is also the key to understanding Kingdee's 2026 interim results. Instead of looking for a new growth curve, the original SaaS curve is getting steeper, and AI is becoming the new variable driving this curve to the next level.

From cloud cashing to AI cashing

The Zhitong Finance App learned that in the first half of 2026, Kingdee handed over a good report card: total revenue was 3.625 billion yuan, up 13.6% year on year, and the growth rate was a bit faster than 11.2% in the same period last year. Among them, cloud service revenue was 3.116 billion yuan, up 16.6% year on year, accounting for 86% of total revenue; subscription revenue was 2,028 billion yuan, up 20.4% year on year, accounting for more than half of the year to 55.9%; subscription ARR reached 4.413 billion yuan, up 18.3% year on year.

What is more noteworthy is that AI native product revenue reached 296 million yuan, up 189% year on year, of which AI native subscription revenue reached 146 million yuan, up 282.9% year on year; related contract amount reached 605 million yuan, up 159.2% year on year.

At the same time, the company's net profit to mother reached 54.5 million yuan in the first half of the year, with a loss of 97.7 million yuan in the same period last year; adjusted net profit to mother reached 116 million yuan, with a loss of about 51 million yuan for the same period last year. Net operating cash flow also changed from negative to positive, reaching $142 million. Gross margin increased to 67.7%, up 2.1 percentage points year on year, and per capita income increased 20.8% year over year, and is expected to exceed 700,000 yuan for the whole year.

Judging from these numbers, Kingdee is experiencing two things at the same time: on the one hand, the subscription business left over the past ten years of cloud transformation continues to grow and continues to release operating leverage; on the other hand, the AI-native business is growing at a much higher rate than traditional SaaS and is beginning to actually contribute revenue and profits.

Among them, the company was able to turn a loss into a profit in the first half of the year, and AI contributed greatly. According to management, in the first half of the year, about two-thirds of new subscriptions in large and medium-sized markets chose original AI products, which means that AI is no longer an option, but one of the core decision-making factors for new customers to choose Kingdee.

As Kingdee's previous cloud transformation path continued, what the market saw at the time was that cloud transformation had entered a harvest period, but by 2026, what the market saw was that the cloud business began to become an infrastructure for AI commercialization. In other words, AI was not inserted into Kingdee's original business system out of thin air.

Today, AI is entering the business scenario, essentially using business data, management processes, authority systems, and customer relationships accumulated in the past. Without this foundation, AI can easily stay at the level of chat, question and answer, and office support; with ERP and SaaS as entrances, AI can further enter the core business processes of enterprises such as finance, procurement, supply chain, manufacturing, and manpower.

This acceleration in growth is essentially due to the magnitude leap in value density created by AI for enterprises.

Reiki: the fulcrum for AI implementation

The core of the original AI product revealed by Kingdee this time is the “Lingee (Lingee) Enterprise Intelligence Operating System”, which was officially released on May 20, 2026. It is positioned as a unified entry point for enterprise-level AI, transforming the business data, management knowledge, organizational memory, and business rules accumulated by the enterprise into an enterprise-level context that can be governed and reused.

According to the Zhitong Finance App, Lingji's billing model is “seat fee+usage points”. The seat fee ranges from 58 yuan/year to 1,168 yuan/year. The usage includes not only the large model's token consumption, but also the engineering capabilities that Kingdee built on top of the token.

After launch, Lingji quickly signed contracts with 26 customers and launched more than 40 smart devices, covering general fields such as finance, supply chain, procurement, and manufacturing. By the beginning of August, Lingji had expanded to more than 100 customers; the company plans to promote 10,000 customers to connect to Lingji in the second half of 2026, of which about 80% is expected to come from existing customers and about 20% from new customers.

Among them, Kingdee's ambition to set this goal comes from its huge customer base: Starry Sky has more than 40,000 active customers, and has served more than one million enterprise customers. For these customers, Lyngky is not a process of re-selecting a vendor, but rather an upgrade over existing ERP and SaaS systems.

At the same time, Kingdee is also focusing on deep AI cultivation in eight major industries: automobiles and spare parts, equipment manufacturing, electronic high technology, process manufacturing, life sciences, food and beverage, modern services, wholesale and retail. Among the newly signed customers, Hytera, Bose Quantum, Shuanghui Group, Youlu Robotics, Kunlun Core, and Facewall Intelligence have all entered the cooperation list.

In addition, efficiency improvements are already supported by specific data: expense review agents release 20% to 30% of reviewers, bank-enterprise reconciliation agents achieve automatic reconciliation for 7 x 24 hours, procurement and payment review agents reduce three-order matching from the hourly level to the minute level, and the checkout intelligence increases closing efficiency by 30%.

On the customer side, Yike Group has increased R&D efficiency by more than 65% based on the Lingji Build development equipment management system, and the delivery cycle has been shortened from weeks to days; Shuanghui Group's dealer sales order delivery time has been reduced by about 30%; and Wen's Group's development cycle has been shortened by 50%.

As a result, Kingdee's market position in the AI field has also been recognized by many parties. The company is the only Chinese manufacturer selected for Gartner's “Voice of Product-based Enterprise Cloud ERP Customer”, with the highest customer recommendation level in the world; it is also the only Chinese manufacturer selected for Gartner's “Magic Quadrant for PLM Software for Discrete Manufacturing”. Among IDC's multiple market segments, Kingdee ranked first. Domestic AI-enhanced enterprise-level ERP public cloud revenue and market share ranked first. Lingji was also selected for the Gartner “Innovation Insight: Enterprise Intelligence Capabilities” report.

The growth curve is getting steeper

While the moat of traditional SaaS mainly comes from product and customer relationships, the moat in the AI era may come further from the “corporate context.” As AI begins to understand an enterprise's business data, management rules, and long-term memory, and is deeply embedded in ERP, then replacement costs will naturally rise as usage time increases.

Kingdee Chairman Xu Shaochun quoted “Sun Tzu's Art of War” at the performance conference and said, “You cannot beat yourself; you can beat your enemy. To do its own business well, Kingdee uses intelligent devices to complete the end-to-end business processes and AI of the enterprise in a closed loop.”

This logic is being verified by data. Kingdee showed a set of comparative data in the performance PPT: the product broke 1 billion in revenue time, the software licensing era took 15 years, the cloud SaaS subscription took 6 to 7 years, and AI native products are expected to only take 1 to 2 years. Judging from current progress, AI's native revenue in the first half of the year was close to 300 million yuan. The second half of the year is a traditional peak season, and the target of 1 billion yuan for the whole year is not aggressive.

In terms of globalization, the company has set up local service networks in 6 markets of Malaysia, Thailand, Vietnam, Indonesia, Singapore and Qatar to achieve direct delivery of fiscal compliance in 6 countries. There are about 140 corporate customers with international business contracts. The new customers include Yongqing Group, Hong Kong Treasury, and CUCKOO International. At the same time, the company's total foreign investment during the period was about 775 million yuan. The investment layout includes AI companies such as Dark Side of the Moon, Stepping Stars, and Silicon-based Liquidity, etc., with net cash reserves of about 3.5 billion yuan. It has also repurchased approximately HK$350 million through repurchases and employee shareholding platforms, and 1% of the total share capital has been cancelled.

Of course, AI products are still in the rapid iteration stage, and customer demand has not completely stabilized. Management admits that at this stage, Lingji's gross profit calculated in points is not necessarily higher than that of traditional products because of token costs, but as model inference costs decrease and model routing and platform engineering capabilities increase, the margin space for AI products is expected to improve further.

According to the Zhitong Finance App, the company maintained its annual guidelines: Group revenue maintained double-digit growth, revenue from native AI products exceeded 1 billion yuan, subscription revenue maintained an increase of about 20%, adjusted net profit margin to mother increased from about 3.3% to about 7%, and net operating cash flow increased by more than 20% year on year. In the longer term, the company still maintains the goal of “using AI to recreate a golden butterfly” in 2030, with AI native and AI+SaaS revenue each accounting for 50%.

Today's Golden Butterfly has a subscription model that is already running, an AI native business that is accelerating on the other, and an organizational system that continuously improves efficiency. After the three are combined, the growth and profit ceiling of a mature SaaS company is being pushed upward at the same time.

The most important change in performance for the first half of 2026 was not a shift in profit from loss to profit, nor a 13.6% increase in revenue, or even an 189% increase in AI native revenue — what is really noteworthy is that for the first time, revenue from native AI products formed a scale of nearly 300 million yuan and began to be an important increase in Kingdee's revenue growth.

Over the past ten years, the customer, data, subscription models and enterprise service capabilities accumulated by the company's cloud transformation have become the foundation for the next round of AI commercialization. Kingdee previously spent ten years completing the migration from traditional software to cloud SaaS, but now, it is re-embedding AI into the business management system in a shorter period of time. Judging from the data for the first half of 2026, AI is not reinventing a curve outside of the old business, but rather reopening a larger enterprise AI management market on top of the SaaS business model that Kingdee has already run.