HSBC: AI is an Opportunity for Software Industry, Prioritizing Applications and Export Targets
AI summary card
HSBC: AI is an Opportunity for Software Industry, Prioritizing Applications and Export Targets
The report argues that AI will not replace software companies but will drive growth by improving efficiency and new business models, initiating coverage on IntSig and ArcSoft with Buy ratings.
- AI is a software industry growth driver rather than a threat
- Preferred Smart Text Recognition and Embedded Software
- Chinese Software Exports Show Potential
- Multi-Agent Systems Becoming a Monetization Trend
- Consumer Hardware like AI Glasses Has Potential
Report interpretation
Overview
HSBC Qianhai Securities released a thematic research report on the Chinese software industry, with core views that AI technology upgrades are opportunities rather than threats to the software industry. The report points out that although competition at the large model layer is fierce and trending towards commoditization, application-layer software companies can accelerate product iteration and innovate business models (such as RaaS) through AI due to their industry Know-how and workflow understanding capabilities. It initiates coverage on IntSig and ArcSoft with Buy ratings, while remaining positive on ERP and office software vendors such as Kingdee, Yonyou, and Kingsoft Office, as well as infrastructure providers such as Sangfor and Baosight Software.
Core views
Reconstruction of AI and Software Relationship: Institutions believe AI finds it difficult to understand complex business flows, while software companies have years of accumulation; AI improves coding efficiency, helping enterprises quickly monetize new products; Enterprise clients have high requirements for security compliance, benefiting record-holding software vendors. Business models shifting to 'Results-as-a-Service' (RaaS) favor revenue growth. Markets over-worry about 'AI eating software', but actual AI will unlock operational efficiency improvements, margin expansion, and TAM growth opportunities. Application Layer and Overseas Logic: Preference for software with industry Know-how (such as ERP), companies creating new products (such as smart text recognition), and software vendors embedding into AI hardware. Chinese software exports are a new increment, overseas users have high willingness to pay, preferring vertical applications (such as IntSig) and solution providers following hardware exports (such as ArcSoft, ThunderSoft). Chinese model API costs are only 20-30% of US models, possessing cost advantages for exporting. Infrastructure and Hardware Opportunities: Popularization of Multi-Agent Systems (MAS) will drive explosion in Token usage, benefiting AI software infrastructure (such as Sangfor) and IDC operators (such as Baosight Software). AI consumer hardware (such as AI glasses) shipments expected to reach CAGR of 64% from 2025-2030, embedded software value increases, algorithm vendors like ArcSoft will benefit.
Analysis framework
The report adopts upstream and downstream transmission analysis of the industry chain, distinguishing between the large model layer (fierce competition, commoditized) and the application layer (benefiting from cost reduction, demand growth). By comparing Chinese and US model costs and Token usage growth data, it demonstrates logic for Chinese model export and infrastructure benefits. Simultaneously combining hardware shipment forecasts (such as AI glasses) to derive embedded software opportunities, and analyzing potential positive impact of business model evolution from License to RaaS/WaaS on software vendor revenue structures.
Methodology notes
Large model supply trending towards commoditization, application layer demand benefits
The report analyzes that the performance gap between foundational models narrows and prices drop (supply side commoditization), which will reduce usage costs for downstream applications, thereby stimulating explosive demand at the application layer.
Penetration rate growth forecast for AI glasses, AI agents
By forecasting compounded annual growth rates of shipments for new products like AI glasses in future years (e.g., 64% CAGR), judge the growth space for related software suppliers.
Value allocation from model layer to application layer, infrastructure layer
The report divides the software industry chain into Model, Application, and Infrastructure three layers, analyzing how AI technology progress distributes value across different layers, concluding that the application layer and infrastructure layer benefit more.
Business model evolution analysis (RaaS/WaaS)
Analyze the shift in software charging models from seat-based to workflow or result-based (Workflow/Result-as-a-Service), demonstrating the positive significance of this evolution for long-term revenue growth of software vendors.
Software index trades at 3.6x TTM PS
Quote the Price-to-Sales (PS) multiples of the software index and its historical average as anchors to judge if industry valuation is reasonable.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- IntSig (688615 CH)Smart Text Recognition, Benefiting from AI feature enhancement and export expansion
- Strengths
- Overseas MAU growth space is large, AI features improve payment willingness
- Weaknesses
- Overseas paid user penetration is currently low
- ArcSoft (688088 CH)Image Algorithms, Benefiting from AI glasses and mobile algorithm value increase
- Strengths
- Algorithm value on AI glasses can be 10x that of mobile phones
- Weaknesses
- Short-term revenue contribution is small
- Sangfor (300454 CH)AI Infrastructure, Benefiting from HCI/SDS demand brought by multi-agent systems
- Strengths
- Hyper-converged infrastructure optimizes AI workloads
- Baosight Software (600845 CH)Data Centers, Benefiting from AI computing power demand and steel digitization
- Strengths
- Tier-1 city IDC resources scarce, high dividend yield
- Kingdee International (268 HK)ERP Software, AI Suite drives subscription revenue growth
- Strengths
- AI contract value exceeds expectations, subscription ratio increases
- Yonyou Network (600588 CH)ERP Software, AI orders strong, promoting subscription transformation
- Strengths
- AI orders account for 18% of revenue, ARR growth accelerates
- Kingsoft Office (688111 CH)Office Software, WPS AI user growth strong
- Strengths
- WPS AI MAU and Token usage grow rapidly
- ThunderSoft Tech (300496 CH)Edge-side AI Deployment, Benefiting from AIoT and automotive intelligence
- Strengths
- Edge large model deployment expertise, high overseas revenue ratio
Key data
- 2029E Global AI Software TAM228 billion USD2024-29E CAGR 37%
- China Software AI Contracts % of Revenue (2025)7%1H25 was 6%, presenting upward trend
- Chinese Model API Cost vs US20-30%Possess significant cost advantage
- AI Glasses Shipments CAGR (2025-30E)64%Reach 90 million units by 2030E
- Kingsoft Office WPS AI MAU (End of 2025)80 millionYoY Growth 307%
- Kingdee 2025 AI Contract Value356 million CNYAccounts for approx 4% of revenue
- Yonyou 2025 AI Orders1.67 billion CNYAccounts for 18% of revenue
Impact & implications
The report believes markets overly worry about 'AI eating software', but actual AI will unlock operational efficiency improvements, margin expansion, and TAM growth opportunities. Chinese software vendors, relying on cost advantages and manufacturing ecosystems,有望 obtain new growth curves in overseas markets and hardware embedded software fields. Popularization of multi-agent systems will significantly drive Token usage and computing power demand, benefiting domestic infrastructure suppliers.
Risks
- Expensive Token costs may hinder agent demand
- Immature multi-agent systems affect user experience
- Cybersecurity and data leak risks
What to watch
- AI Order Growth Status
- Overseas monetisation progress
- AI Glasses Shipments
- Token Usage Growth