The software sector is currently at the intersection of two major trends: the "AI industrial transformation" and the "policy dividends from IT application innovation." The core investment logic can be broken down into four dimensions.
First, AI is accelerating growth, and rising cloud computing prices are supporting profitability. The year 2026 is seen as the first year of commercialization for AI Agents, with token consumption growing exponentially. Each token consumed directly monetizes software capabilities. The new business model of AI Agents, which involves "payment per process/result" (compared to traditional software's annual or per-license fees), is not only creating new markets but also reallocating existing budgets for traditional outsourcing and software procurement, potentially far exceeding the market space of traditional software. Simultaneously, cloud providers globally are entering a price increase cycle, and the revaluation of AI computing power and cloud services is further solidifying the pricing anchor for software services.
Second, IT application innovation policies are driving accelerated order releases. Document No. 79 from the State-owned Assets Supervision and Administration Commission mandates that by the end of 2027, all central government-owned and local state-owned enterprises must achieve 100% replacement of their information systems with domestic IT application innovation products. Government procurement is favoring "domestic products," not only defining preferred scopes through lists but also guiding concentrated demand release through price advantages and standardized assessments, promoting high-quality IT application innovation products from "usable" to "large-scale application." With the 2027 full replacement deadline approaching, 2026 is becoming a critical window for the concentrated landing of IT application innovation orders.
Third, breakthroughs in core technologies are opening up industry elasticity. Focusing on "bottleneck" areas like basic software, policies are driving breakthroughs in core technologies with extraordinary intensity, promoting deep integration of "cloud-native, AI-native, and IT application innovation-native" approaches, breaking the fragmented pattern of single-point adaptation, and building a self-reliant, collaboratively evolving industrial ecosystem. Meanwhile, private and small-to-medium enterprises are accelerating their entry, enhancing the vitality and elasticity of the industrial chain.
Fourth, valuations are entering a repair window, and the sector has a safety margin. The previous narrative of "large models consuming software" led to a significant valuation discount for the software sector, but this involved two major misjudgments. First, the surge in computing power expenditure did not squeeze corporate IT budgets; instead, AI Agent applications are expanding new increments (such as marketing budgets). Second, while AI coding reduces code costs, the value of non-code elements like data barriers increases, strengthening the moats of application providers. In the AI era, the pricing logic revolves around tokens, and the software track, with its high growth potential and low valuations, is entering a window for revaluation.
The HuaBao Software Development ETF (subscription code: 159036) passively tracks the CSI All Share Software Development Index (referred to as Software Development, code: 932094.CSI). It is 100% invested in the software development industry, comprising 117 constituent stocks, and aims to reflect the overall performance of securities from software development companies within the CSI All Share Index sample.
The software track is in an overall upward cycle with immense potential, but competition is fierce. In an environment where it's unclear which sub-sector or individual stock will outperform, a more comprehensive and diversified layout may capture the beta returns from the entire software industry's development.
Compared to similar software indices, the target index (Software Development Index) is purer in its sector focus and excels in diversification. It has three clear differential advantages: highest industry purity (100% focused on software development), more dispersed holdings (including 117 constituent stocks to mitigate single-stock risk), and a more balanced layout (top ten holdings account for only about 40%, with a single stock weight cap of 10%). In the software industry's upward cycle, this more comprehensive and balanced layout may capture industry beta returns driven by the resonance of multiple forces like AI, IT application innovation, and industrial software.
Specifically, in terms of industry distribution, the target index is 100% invested in software development according to the CSI three-level industry classification, while other indices like the Software Index, CSI Software, Chuangye Software, and Industrial Software Index have weights in software development of 97.3%, 83.4%, 48.1%, and 31.7%, respectively. Regarding the number of constituents, the target index has 117, compared to 50 or 30 for similar software indices. For the top ten holdings' aggregate weight, the target index is at 39.67%, less than 40%, while others exceed 48%, with the CSI Software Index nearing 60%. For single stock weight caps, the target index is limited to 10%, while others are capped at 15%.
The Software Development Index encompasses popular themes. The weight of constituent stocks in concepts like AI applications, cloud computing, IT application innovation industry, fintech, cybersecurity, and the HarmonyOS ecosystem are 43.93%, 42.15%, 41.51%, 29.66%, 16.62%, and 15.34%, respectively.
The target index has demonstrated strong performance, leading its peers. Since 2023, its annualized return is 2.16%, outperforming similar software indices like CSI Software (-1.97%), the Software Index (-0.46%), and Industrial Software (-7.35%).
The top ten holdings of the target index include leading companies in the software track such as iFlytek, Tonghuashun, Kingsoft Office, TOWEI Information, Hundsun Technologies, 360 Security Technology, Compass, HopeRun Software, Sangfor Technologies, and CAS Star, with a combined weight of 39.67%.
The narrative of "large models consuming software" refers to concerns that AI models might replace traditional software. In the SaaS industry, the subscription model involved fixed periodic payments for software access. In contrast, AI native companies often adopt usage-based or outcome-based pricing, such as charging per million tokens or based on task completion. This shift accelerated with the commercial success of companies like Anthropic, whose Claude Cowork, through open-source plugins, integrates into enterprise workflows, allowing users to accomplish tasks via natural language, raising fears of software being "consumed."
However, software companies are not passive; they are actively transforming and building core barriers. These include data barriers from accumulated industry and user data, complex process barriers in highly customized workflows, and industry know-how barriers in vertical sectors like energy and industry, where deep understanding and precise application of domain knowledge are crucial, requiring compliance with regulations and business practices.
From a performance perspective, the dense release of Q1 2026 reports is gradually dispelling doubts about "large models consuming software." A group of domestic software firms with vertical barriers have delivered better-than-expected results through AI-driven product upgrades and accelerated commercialization, marking the industry's transition from "AI impact" to "AI empowerment." For example, Kingsoft Office's Q1 report showed AI subscription revenue growing over threefold year-over-year, with AI feature monthly active users exceeding 120 million and rising paid conversion rates. The company stated that AI is not only a core driver of user growth but also a key variable in increasing average revenue per user.
Fundamentally, this is not a story of disappearance but of upgrade. The software sector faces a business model upgrade, not the demise of traditional software. This includes pricing model upgrades from per-license/annual fees to usage/outcome-based fees; value proposition upgrades from function-driven to value-driven; opening of incremental space where AI Agents' result-based收费 models create new markets and reallocate existing budgets in areas like human resources, marketing, and outsourcing; and a strengthened pricing anchor as cloud providers' price increases and the revaluation of AI computing power and cloud services further solidify software service pricing.
In summary, AI is not consuming software but is comprehensively reconstructing the software industry's business models and value distribution rules with tokens as the pricing unit and Agents as the new form. Software companies with data barriers, process depth, and industry know-how see their moats not eroded but amplified in the AI era.
Regarding the valuation of the Software Development Index, earlier this year, the "large models consuming software" narrative caused significant valuation discounts for the AI application sector compared to computing power and model layers. Some analysis suggests this pricing logic is a misjudgment. In the AI era, pricing revolves around tokens, forming a three-layer markup structure across computing power, model, and application layers, with the value reconstruction of scenario tokens in the application layer potentially not fully priced by the market. End enterprise users buy business outcomes, not tokens. The AI application layer, empowered by private data and industry know-how, allows generic model tokens to gain greater value, enabling higher token prices. High-growth, low-valuation, quality AI application targets with revenue from token consumption or outcome-based fees may attract market attention.
Since its high on January 14, 2026 (index point 1159.16), the Software Development Index had fallen 23.67% by April 28, temporarily underperforming AI hardware sectors (computing power, chips, etc.). The software sector has become a segment within the AI industry chain with relatively "lower water levels" and higher赔率 (limited downside, yet strong赚钱 elasticity once trends emerge), poised to gather momentum.
Valuation-wise, as of April 28, the Software Development Index's trailing twelve-month P/E ratio was 339.04 times, below its median of 413.36 times, at the 43.67% percentile since listing, indicating relatively high valuation attractiveness and safety margin.
IT application innovation policies benefit the software sector in several ways. Policy-wise, the 2027 full replacement deadline is pressuring 2026 to become a big year for order releases. Document No. 79 mandates 100% replacement for state-owned enterprises by end-2027. As the deadline approaches, replacement is expanding from government and party offices to eight key industries like finance, telecommunications, and energy, and from peripheral to core business systems. Refined government procurement demand standards further solidify the replacement pace.
Funding is secured through the expansion of ultra-long-term special government bonds and local debt resolution plans, providing financial保障 for IT application innovation procurement. From 2024 to 2026, ultra-long-term special bonds have expanded continuously (1 trillion yuan to 1.3 trillion yuan), with funds directed to "two majors" (national major strategies/security) and "two new" areas, with IT application innovation属于重点安全领域. The implementation of local debt resolution plans releases budget space, providing funding for IT application innovation procurement.
Market space is substantial. According to forecasts, China's IT application innovation industry scale is expected to exceed 1.8 trillion yuan in 2026, with a compound annual growth rate of 28.5% from 2023 to 2026. Some estimates suggest the market size could approach 6 trillion yuan by 2029.
Performance validation is turning policy expectations into real corporate orders and业绩. For instance, Seeyon Internet's 2025 annual report showed contract amounts from central and state-owned enterprise clients grew 13.6% year-over-year.
Some securities research points out that over 70% of the software sector's growth is contributed by IT application innovation, with 2026 being a peak year for performance realization. IT application innovation is the core growth engine for the sector, with AI providing弹性补充. Others view IT application innovation as the core beta of the software sector, dominating景气度 and valuation中枢, while AI is the core alpha, driving弹性 and growth space.
Under the major trend of AI reshaping software, vertical applications of "AI + software" are highlighting their value. Compared to the randomness and non-standardization of general large model responses, AI applications in vertical fields like finance, healthcare, and e-commerce, through预设场景, optimized prompts, and专属数据训练, can provide more precise, stable, and practical solutions. The deep integration of "AI + software" is creating real value across industries.
In AI + finance, tools like market software, investment research tools, and智能投顾 are rapidly integrating AI capabilities. In an environment where thematic investing prevails in A-shares and new concepts emerge constantly, financial AI tools that can quickly interpret information and梳理产业链 have clear application scenarios and growth potential.
In AI + healthcare, it covers areas like AI-guided diagnosis,影像辅助诊断, and drug molecule研发. AI can significantly shorten R&D cycles and improve diagnosis and treatment efficiency, representing a highly certain direction for融合. Although mature business models are still being explored, continuous attempts and technological advancements pave the way for long-term industry development.
In AI + other verticals, similar transformations are occurring in fields like e-commerce, education, government affairs, and industrial software. The process of software companies upgrading products and achieving value revaluation through AI integration is just beginning.
Some analysis indicates that software companies with advantages like industry know-how and high-quality data may build moats and fully benefit from AI-driven value growth in the AI Agent era. Others list AI applications and domestic computing power as dual mainline investment opportunities for 2026, focusing on细分赛道 like AI + finance, AI + industrial software, AI + healthcare, AI + office, AI + ERP, and AI + marketing.
Looking ahead, further positive catalysts include potential policy actions. At a policy briefing on April 28, 2026, it was mentioned that下一步 efforts will promote the extension of producer services towards specialization and high value chains, accelerating the innovative development of software and information technology services. Specifically regarding AI empowering the information service industry, an "Artificial Intelligence + Software" special action will be launched to accelerate the R&D and application of intelligent programming, and cultivate new business formats like model-as-a-service and agent-as-a-service. Additionally, open-source ecosystem construction will be strengthened to promote the intelligent upgrade of basic software and industrial software. The digital-intelligent transformation service system for manufacturing will be健全, and优质数智化转型服务商 will be cultivated分类分级.
Furthermore, the industrial internet innovation development project will be深入实施,算力布局 and edge computing construction will be有序推进, and the intelligent computing cloud service system will be完善. An industrial data筑基行动 will be implemented to建设一批工业领域高质量数据集.
Data shows that in 2025, China's software and information technology service industry revenue reached 15.48 trillion yuan, 6.2 times that of 2012, with an average annual复合增长率 of 15.1%.
Some securities firms believe the software industry in 2026 presents a格局 of共振 from "policy + technology + demand." Policy-wise, plans promote the纵深发展 of government digitalization, with over 90% of government services handled online, and the digital government market size is预计 to reach 213.4 billion yuan by 2028. Technologically, AI Agents and large models drive industrial升级, with China's generative AI software market预计 to reach $3.54 billion by 2028, and intent recognition accuracy for AI guides in government scenarios already reaching 95%. On the demand side, the digital economy's share of GDP has increased to 42.8% (2023), and the cloud computing market size reached 828.8 billion yuan (2024), indicating broad market space.
Regarding product basics and fees, for the HuaBao Software Development ETF, the subscription fee is 0.3% for amounts below 1 million yuan, and 1000 yuan per transaction for 1 million yuan and above. On-exchange trading fees are subject to what securities companies actually charge. When申购 or赎回 fund shares,申购赎回代理券商 may charge a commission of up to 0.3%. Specific fund fee rates are detailed in the fund's legal documents.
Risk提示: The HuaBao Software Development ETF passively tracks the CSI All Share Software Development Index. The index's base date is December 31, 2021, and its release date is March 29, 2023. The fund is issued and managed by HuaBao Fund. Distributors do not undertake the investment, redemption, and risk management responsibilities of the product. Investors should carefully read the Fund Contract, Prospectus, Fund Product Summary, and other fund legal documents to understand the fund's risk-return characteristics and choose products suitable for their own risk tolerance. The fund manager assesses this fund's risk等级 as R3-Medium Risk, suitable for Balanced (C3) and above investors. Suitability matching opinions are subject to those of the销售机构. Sales institutions (including the fund manager's直销机构 and other sales institutions) evaluate the fund's risk according to relevant laws and regulations. Investors should promptly pay attention to the suitability opinions issued by sales institutions and base their匹配结果 on those. Suitability opinions from different sales institutions may not necessarily be consistent, and the fund product risk等级评价结果 issued by fund sales institutions shall not be lower than the risk等级评价结果 made by the fund manager. There may be differences between the fund's risk-return characteristics as described in the fund contract and its risk等级 due to different考虑因素. Investors should understand the fund's risk-return situation,结合自身投资目的,期限,投资经验, and risk承受能力 to谨慎选择基金产品 and bear risks自行. The China Securities Regulatory Commission's registration of this fund does not indicate its substantive judgment or guarantee of the fund's investment value, market prospects, or returns. The fund's past performance and its net asset value level do not预示 its future performance. The performance of other funds managed by the fund manager does not构成保证 of this fund's performance. Funds involve risks; investment须谨慎.
Comments