China software, data center and cloud sector Report Interpretation

Overview

This China software, data-center and cloud earnings review concludes that AI demand and monetization are supporting resilient growth and improving profitability for selected companies, but weak government, property, financial-IT and cybersecurity spending leaves the sector uneven.

Core views

The report reviews 19 covered software, data-center and cloud companies following 1H26/2Q26 results: four beat expectations, 10 were in line and five missed. It finds that AI infrastructure demand remains strong, with Kingsoft Cloud delivering better-than-expected 2Q26 revenue growth from its AI-cloud business. VNET and GDS experienced modest growth slowdowns from AI-chip constraints and rental-price dilution, but the institution believes their new orders support later growth acceleration. ERP provider Kingdee, office-software provider Kingsoft Office and Meitu’s photo-editing and productivity tools produced solid results, supported by a greater contribution from AI and recurring revenue. In contrast, property-related software, financial IT and cybersecurity continued to face weak demand. At the sector level, MIIT data showed China software and IT-services revenue up 9.2% YoY to Rmb8.98tn in 7M26, versus 9.5% in 1H26, while net profit rose 1.3% to Rmb1.04tn, versus 1.0% in 1H26. Net profit margin was broadly stable at 11.6%, compared with 11.7% in 1H26. July IT-services revenue growth of 7.6% exceeded the 7.4% sector average, but software-product growth slowed to 2.6% from 3-7% in April-June, and IT-security revenue growth of 6.2% for 7M26 remained below the sector average. The report attributes broad profitability improvement among covered names to stringent headcount control and AI-enabled efficiency in project delivery and internal management, rather than to uniformly strong end demand. The institution sees a widening divide across end markets. Industrial-sector revenue grew 6.5% YoY in 7M26 and may support software and IT-services demand, while government IT budgets remain pressured despite national general-public-budget revenue improving 5.8% YoY. SOE revenue declined 2.4% YoY. Property conditions remain a major headwind: national new-home sales volume and value fell 12.7% and 13.2% YoY in 7M26, leading the report to remain cautious on property-related software spending in 2H26E. Financial-software revenue remained soft, with Hundsun’s combined wealth-technology and asset-management-technology revenue down 11% YoY in 1H26, and cybersecurity demand also remained weak. AI monetization is the main positive differentiator. Kingsoft Cloud’s AI-cloud revenue represented 56% of public-cloud revenue in 2Q26, while the Xiaomi and Kingsoft ecosystem accounted for 26% of its total revenue. Kingdee’s ARR rose 18% YoY to Rmb4.41bn at end-2Q26 and its relevant net-dollar-retention metrics remained above 100%. Kingsoft Office’s WPS to-C, WPS 365 and WPS software businesses represented 61%, 15% and 22% of 2Q26 revenue, respectively; global WPS monthly active devices rose 1% QoQ to 676mn, and domestic cumulative annual paying users increased 15% YoY in 1H26. Meitu’s paying users rose 3% QoQ to 18.4mn at end-2Q26, and photo, video and design products contributed 80% of 1H26 revenue. These operating indicators underpin the institution’s preference for Kingdee, Kingsoft Office and Meitu. For data centers, the report favors GDS, VNET and Kingsoft Cloud because AI demand is improving order visibility and capacity utilization prospects. GDS had 542k sqm of utilized China capacity at 79.2% utilization by 2Q26; its existing Mainland China backlog could support roughly 40% utilized-area growth. VNET’s wholesale IDC revenue rose to 40% of total revenue in 2Q26 from 35% a year earlier, with 1GW of wholesale capacity in service and 73.9% utilized. The report nevertheless notes pricing dilution, competitive data-center service pricing and AI-chip availability as constraints. Despite an 11% 3Q26-to-date rebound in covered software stocks, the institution argues that investors are favoring high-quality SaaS businesses with durable advantages and credible AI monetization. Names with weak differentiation, weak profitability or slowing growth may continue to de-rate. Covered software companies traded at 3.5x FY27E P/S on average, while GDS and VNET traded at 10.3x FY27E EV/EBITDA. The report revised price objectives for seven companies and reduced FY26-28E revenue estimates by 2-3% on average for that group, mainly for weaker-demand areas, while retaining its selective preferred-Buy list.

Analysis framework

The institution combines the 1H26/2Q26 earnings outcomes of 19 covered companies with sector data, segment demand indicators, operating-margin trends, company operating metrics and relative valuation. It then separates AI-linked recurring-revenue and infrastructure beneficiaries from companies exposed to weak property, government, financial-IT or cybersecurity spending, and updates selected estimates, price objectives and valuation multiples accordingly.

Methodology notes

• Industry AnalysisSupply-demand framework

  Segment-by-segment demand assessment across cloud, ERP, office software, creative tools, financial IT, cybersecurity and property software.

  The report links each segment’s revenue outlook to downstream customer budgets, AI demand, industrial activity, property conditions and data-center capacity demand.

• Industry AnalysisVolume-price decomposition

  Data-center growth is assessed through capacity, utilization, move-in pace, backlog and rental-price trends.

  This distinguishes growth from additional occupied capacity and orders from pressure caused by pricing dilution or slower customer move-in.

• Valuation methodsP/E and PEG Valuation

  Forward P/E, P/S, EV/EBITDA, SOTP and DCF methods are used for individual covered-company price objectives.

  The report applies earnings multiples to profitable companies, sales multiples to growth businesses that are still loss-making, EV/EBITDA to data centers, and company-specific DCF or SOTP where appropriate.

Asset mapping & comparison

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

• Kingdee (268 HK)

  Preferred Buy based on resilient top-line growth, faster AI-native revenue expansion and improving margins.

  Strengths

  Subscription revenue, AI monetization, ARR growth and improving profitability.

  Weaknesses

  ERP demand remains exposed to competition and cloud-subscription growth execution.

  Comparison

  Kingdee has delivered faster revenue growth and better profitability than Yonyou since FY23.

  Risks

  Intensified SaaS competition, slower cloud or subscription growth, and valuation pressure.

• Kingsoft Office (688111 CH)

  Preferred Buy supported by WPS 365, WPS software and to-C recovery.

  Strengths

  Strong profitability and free cash flow, growing paying users, AI-related ARPU potential and enterprise SaaS penetration.

  Weaknesses

  AI-powered WPS adoption may take time as users change working habits.

  Comparison

  Valued using DCF with a 10.1% discount rate and 4% terminal growth rate.

  Risks

  Slower AI-WPS adoption, weaker Xinchuang demand and competition from other AI-powered software.

• Meitu (1357 HK)

  Preferred Buy for subscriber-led earnings growth and operating leverage.

  Strengths

  Rising paying users, productivity-tool expansion, increasing overseas contribution and improving operating margin.

  Weaknesses

  Gross-margin contraction was noted despite operating leverage.

  Comparison

  The report compares Meitu with global leading SaaS companies because of its higher overseas revenue exposure.

  Risks

  Slower paid-user conversion, competition, regulatory or geopolitical risk and valuation compression.

• GDS Holdings (GDS US / 9698 HK)

  Preferred Buy and AI-demand beneficiary in China data centers.

  Strengths

  Large backlog, improving China demand, 79.2% utilization and capacity pipeline.

  Weaknesses

  Revenue growth slowed slightly amid AI-chip constraints and rental-price dilution.

  Comparison

  Trading at 12.2x forward EV/EBITDA versus a 19.5x mean since 2019.

  Risks

  Data-center price competition, slower client move-in and AI-chip shortages.

• VNET Group (VNET US)

  Preferred Buy for wholesale capacity expansion and improving AI-related order visibility.

  Strengths

  Wholesale capacity pipeline, growing wholesale revenue mix and shareholder-related AI business synergies.

  Weaknesses

  Growth slowed slightly amid chip constraints and rental-price dilution.

  Comparison

  Trading at 8.5x forward EV/EBITDA versus a 13.8x mean since 2019.

  Risks

  Slower utilization ramp, data-center pricing pressure, shareholder stake reduction and AI-chip shortages.

• Kingsoft Cloud (KC US)

  Preferred Buy driven by rapidly growing AI-cloud demand and Xiaomi/Kingsoft ecosystem demand.

  Strengths

  AI cloud represented 56% of public-cloud revenue and operating leverage improved profitability.

  Weaknesses

  The company remains loss-making in the near term and has customer concentration exposure.

  Comparison

  Trading at 1.5x forward P/S versus a 2.6x mean since May 2020.

  Risks

  Pricing pressure, high customer concentration and AI-chip export restrictions.

Key data

• Covered earnings outcomes4 beats, 10 in line, 5 misses19 covered software, data-center and cloud companies reporting 1H26/2Q26 results.
• China software and IT-services revenueRmb8.98tn, +9.2% YoY7M26; compared with +9.5% in 1H26.
• China software and IT-services net profitRmb1.04tn, +1.3% YoY7M26; compared with +1.0% in 1H26.
• Sector net profit margin11.6%7M26, broadly stable versus 11.7% in 1H26.
• Covered-share performance+11%China software companies under coverage in 3Q26 to date.
• Estimate revisionsFY26-28E revenue cut by 2-3% on averageFor seven software companies under coverage.
• Kingsoft Cloud AI-cloud contribution56% of public-cloud revenue2Q26.
• Kingdee ARRRmb4.41bn, +18% YoYAt end-2Q26.
• GDS China utilized capacity542k sqm at 79.2% utilization2Q26; existing backlog could support roughly 40% utilized-area growth.
• Sector valuation3.5x FY27E P/SAverage for covered China software companies; GDS and VNET averaged 10.3x FY27E EV/EBITDA.

Impact & implications

The report’s central implication is that AI is creating both demand and productivity benefits for selected cloud and software businesses, reducing broad concerns about disruption. However, it expects performance and valuation to remain highly differentiated because customer budgets and growth visibility are still weak in property, financial IT and cybersecurity.

Risks

• Government IT-budget pressure may persist in 2H26E despite improved public-budget revenue.
• Continued property-market weakness could suppress property-related software spending.
• Financial-IT and cybersecurity demand may remain weak as customer budgets and competition constrain growth and margins.
• Data-center growth could be limited by AI-chip shortages, slower client move-in and service-price competition.
• AI-linked software monetization may disappoint if customer adoption, subscriptions or paying-user growth are slower than expected.

What to watch

• AI-related revenue contribution and monetization progress across cloud, ERP, office software and creative tools.
• New orders, utilization and move-in pace at GDS and VNET, as well as AI-chip availability.
• Kingsoft Cloud’s AI-cloud revenue mix and demand from the Xiaomi and Kingsoft ecosystem.
• Kingdee ARR and retention, Kingsoft Office’s WPS 365 and to-C growth, and Meitu’s paying-user and ARPPU trends.
• Government, SOE, industrial and property indicators as signals for downstream software spending.
• Further changes in cybersecurity, financial-IT and property-software demand and margins.