Kingdee International Software Group Company Limited (00268): JPMorgan assumes Kingdee coverage at Neutral as Cloud profitability improves but AI value capture remains unproven

Overview

The report examines whether Kingdee's maturing Cloud transition and emerging AI products can create a second growth curve. JPMorgan sees improving earnings quality and valuable system-of-record assets, but considers the risk-reward balanced until AI adoption translates into broader customer spending, consolidated recurring revenue, gross profit and free cash flow.

Core views

JPMorgan assumes coverage of Kingdee at Neutral and sets a Dec-27 price target of HK$8.30, based on 20x 2028E P/E. The target multiple is above the roughly 15x 2028E P/E applied to mature application peers because Kingdee is moving toward sustained profitability, has a growing subscription base and holds a strategic enterprise-system position. However, the report believes that much of the underlying Cloud improvement is already reflected in expectations: its 2026-28E revenue and profit forecasts are broadly in line with consensus, while the shares trade at 26x JPMorgan's 2027E P/E. Clearer AI-driven revenue acceleration, sustained positive cash flow and stronger large-enterprise execution would be needed to support a higher valuation. The first transition—from licenses to Cloud—is increasingly producing a more visible earnings foundation. Cloud revenue has reached 86% of group revenue. In 1H26, group revenue rose 13.6% year on year, subscription revenue increased 20.4%, ARR grew 18.3%, gross profit rose 17.2%, and gross margin reached 67.7%. Revenue per employee rose 20.8%, while selling and R&D expense ratios declined. JPMorgan interprets these changes as evidence that a scaled recurring-revenue base is beginning to grow faster than corporate costs. It forecasts revenue of Rmb7,941mn in 2026E, Rmb8,904mn in 2027E and Rmb10,043mn in 2028E; adjusted EBIT margin is projected to rise from 5.2% in 2026E to 9.5% in 2027E and 13.7% in 2028E. The report nonetheless stresses that implementation-intensive large-enterprise work can still consume delivery labor and working capital, making operating cash flow, free cash flow and cash conversion more meaningful tests than accounting earnings alone. AI may expand the enterprise-software value pool by changing monetization from employee seats and software modules toward usage, actions, workflows and completed work. As agents undertake accounting, procurement, supply-chain, HR and other tasks, software spending could be linked to a broader labor and service-cost pool. Conversely, higher employee productivity could compress the number of human users and pressure seat-based revenue. The report's central economic question is therefore whether usage- and workflow-based revenue grows faster than any seat compression. AI could also improve supplier economics through coding agents, natural-language configuration, automated testing and reusable workflow agents, potentially reducing the customized delivery burden that has historically constrained Chinese SaaS scalability. Early AI metrics demonstrate product momentum but not yet material group-level value capture. AI-native revenue rose 189% year on year to Rmb296mn in 1H26, including Rmb146mn of AI-native subscription revenue, up 283%; this compares with 18.3% growth in consolidated ARR. JPMorgan regards the gap as important: growth from a small AI revenue base is not sufficient proof of a new earnings pool. It expects gradual monetization rather than a near-term step-up in consolidated growth, and lays out a required progression of evidence: AI-enabled revenue, incremental annual contract value and net revenue retention, consolidated ARR acceleration, incremental gross profit, and ultimately free cash flow. The report places greater weight on recurring workflow usage, paid attach rates, incremental ACV and NRR than on agent launches or reported AI capabilities. Kingdee's system-of-record position is the core strategic support for the investment case. Its installed base contains live enterprise state, permissions, business rules, integrations and transaction-execution capability—inputs that are difficult to reconstruct, especially in large organizations with subsidiaries, approval hierarchies, customized processes and regulatory requirements. JPMorgan argues that these assets remain essential as AI shifts from answering questions to carrying out actions; an external agent may still need Kingdee's underlying data and execution infrastructure to complete a workflow. This gives Kingdee a credible role in agent-driven enterprise software even if users spend less time in the traditional ERP interface. That same structure creates a material competitive challenge. AI-native enterprise applications and general-purpose agents can build a new interface around finance, procurement, HR and operational workflows while connecting to existing ERP systems through APIs. In such a scenario, Kingdee may preserve system-of-record value but capture a smaller share of the incremental AI wallet. Lingee, Kingdee's enterprise agentic operating system, is therefore strategically important because it could enable the company to retain the workflow and customer-relationship layer. Its economic significance depends on repeated paid use and its contribution to customer spending, rather than merely on product functionality. Competition from Yonyou, AI-native vendors, model providers and customer-developed systems makes the timing of this workflow transition critical. The report expects the maturing Cloud model to fund AI and product investment internally, but says returns on incremental investment should become the key capital-allocation measure. JPMorgan forecasts adjusted net income of Rmb523mn in 2026E, Rmb856mn in 2027E and Rmb1,306mn in 2028E, with free cash flow to firm of Rmb1,065mn, Rmb1,533mn and Rmb2,159mn respectively. It sees AI as value-accretive only if incremental customer spending and internal productivity gains exceed inference, development, delivery and customer-acquisition costs. Strong AI growth accompanied by improving consolidated margins and cash conversion would validate favorable economics; rapid AI growth with limited ARR acceleration or weaker cash conversion would support caution.

Analysis framework

JPMorgan starts with Kingdee's Cloud-transition financial progress, then assesses whether AI expands the addressable workflow and monetization pool. It evaluates Kingdee's enterprise data and execution assets against the risk that AI-native products control the customer workflow layer, and tests the thesis through ARR, ACV, NRR, margins, revenue per employee, operating cash flow and free cash flow. Valuation is anchored on a 2028E P/E multiple and compared with mature application peers.

Methodology notes

• Valuation methodsP/E and PEG Valuation

  Forward P/E valuation

  JPMorgan derives its HK$8.30 Dec-27 target price from 20x 2028E P/E, applying a premium to mature application peers at roughly 15x because Kingdee is still normalizing profitability.

• Industry AnalysisUpstream-Midstream-Downstream Transmission

  AI value-pool and workflow value-capture analysis

  The report examines how AI agents may shift value from ERP seats and modules toward workflow usage, while separating Kingdee's system-of-record role from the potentially more contested application and interface layer.

• Corporate Fundamentals and FinanceFree cash flow analysis

  Cash conversion and incremental free-cash-flow analysis

  JPMorgan treats operating cash flow, free cash flow, margins and revenue per employee as evidence of whether Cloud scale and AI investment are producing sustainable economic returns.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Kingdee International (0268.HK)

  Primary covered company; Cloud profitability is improving while AI could create a second workflow-monetization curve.

  Strengths

  A large enterprise installed base, recurring Cloud subscriptions, and scarce enterprise data, permissions, business rules, integrations and transaction-execution capabilities.

  Weaknesses

  AI monetization remains early, while large-enterprise delivery can remain implementation-intensive and constrain cash conversion.

  Comparison

  JPMorgan applies 20x 2028E P/E versus around 15x for mature application peers, reflecting Kingdee's earlier-stage profitability normalization.

  Risks

  Weaker enterprise IT spending, Yonyou and AI-native competition, seat compression, higher inference costs, and failure of AI growth to convert into consolidated ARR, margins and free cash flow.

• Yonyou

  Competitor in enterprise software and AI-enabled workflows.

  Comparison

  Identified as one of the competitors that could weaken Kingdee's ownership of the customer workflow.

  Risks

  Its competing AI and ERP capabilities may capture part of the incremental workflow spending pool.

Key data

• Rating and target priceNeutral; HK$8.30 Dec-27 target pricePrevious rating was Overweight and prior Dec-27 target price was HK$12.00.
• Share priceHK$7.47Price as of 23 Sep 2026.
• Cloud revenue mix86% of group revenueCloud provides Kingdee's more visible recurring earnings foundation.
• 1H26 subscription revenue and ARR growth20.4% YoY and 18.3% YoYThe report views these as a healthy base but seeks evidence of faster consolidated growth.
• 1H26 gross margin and revenue per employee67.7% and +20.8% YoYIndicators of emerging Cloud operating leverage.
• AI-native revenueRmb296mn, +189% YoYIncludes Rmb146mn of AI-native subscription revenue, up 283% year on year.
• 2028E financial forecastRevenue Rmb10,043mn; adjusted EBIT margin 13.7%; adjusted net income Rmb1,306mn; FCFF Rmb2,159mnJPMorgan expects profitability and cash generation to improve through 2028E.
• Valuation20x 2028E P/E target multipleCompared with mature application peers at around 15x 2028E P/E; Kingdee traded at 26x 2027E JPMorgan P/E.

Impact & implications

JPMorgan sees Kingdee as transitioning from Cloud investment toward earnings normalization, with AI providing upside optionality rather than a confirmed near-term earnings driver. The report considers a rerating dependent on proof that Lingee and AI workflows expand customer wallets, accelerate consolidated ARR and convert into gross profit and free cash flow while Kingdee protects meaningful workflow ownership.

Risks

• Weaker enterprise IT spending could lengthen sales cycles and slow subscription, ARR and revenue growth.
• Yonyou, AI-native vendors, model providers and customer-developed systems could capture the workflow layer while Kingdee remains the underlying system of record.
• AI adoption may grow without producing incremental customer spending, consolidated ARR acceleration or revenue growth.
• Large-enterprise deployments may remain labor- and working-capital-intensive, delaying gains in productivity, margins and cash conversion.
• Seat compression, inference costs, implementation costs and customer-acquisition costs could limit incremental margins and free cash flow.

What to watch

• Recurring Lingee workflow usage, paid attach rates and evidence that AI adoption is moving beyond pilots.
• ACV, NRR and incremental spending per customer as measures of wallet expansion rather than contract substitution.
• Consolidated subscription revenue, ARR and total revenue growth for signs that AI is affecting the group-level trajectory.
• Revenue per employee, gross margin, expense leverage, operating cash flow and free cash flow as tests of scalable Cloud and AI economics.
• The extent to which Kingdee retains workflow ownership against AI-native applications and general-purpose agents.