Looking at the trajectory of past technological revolutions, profits consistently migrate from hardware toward applications—where hardware reaches its limit, software springs forth. With the current AI computing boom fueling domestic hardware advancement, software development—serving as the vehicle for implementation and monetization—is now positioned to capture the commercial value unlocked by technological upgrades. Since early 2026, AI application commercialization has entered its payoff phase: MiniMax and Zhipu AI have both listed in Hong Kong and been included in the Stock Connect, surging over 30% within just two days of inclusion. Meanwhile, DeepSeek announced API price increases, and Goldman Sachs projects China's AI models could generate a combined annualized recurring revenue (ARR) of $13 billion by the end of 2026. Add to this the approaching 2027 deadline mandated by SASAC Document No. 79 for 100% domestic substitution, coupled with a wave of supportive software industry policies, and the software development sector appears ready to ride a powerful tailwind.

Where Hardware Ends, Software Begins

Historical patterns from technology revolutions show that after upstream hardware breakthroughs, profit centers gradually shift toward applications and software. AI follows the same "hardware foundation, software payoff" progression—computing power and chips form the base, but software and applications are where value ultimately gets realized. In early 2026, as global narratives around "large models devouring software" rattled sentiment, AI application stocks traded at significant valuation discounts compared to computing power and model layers, making software development one of the "lowest water level" segments within the AI supply chain, offering comparatively attractive valuation metrics and safety margins. From an industry positioning standpoint, software sits between AI's midstream technology layer and downstream application layer—bridging large model capabilities on one end and real-world industry deployment on the other—making it the critical link where AI's technological dividends convert into tangible revenue.

Overseas Volatility Creates Valuation Discount Opportunity

Overseas software stocks experienced sharp fluctuations early this year: after Anthropic unveiled its Claude Cowork series of AI agents, US software equities broadly sold off—Adobe fell approximately 27% year-to-date, ServiceNow and Salesforce dropped around 35%, and Workday declined roughly 50%—as markets worried about disruption to traditional SaaS seat-based models. However, numerous industry insiders argue that the software sector has "identified the right direction but mistimed the execution"—enterprise revenue won't vanish overnight, and moats including proprietary data, regulatory compliance lock-in, and network effects remain intact, suggesting that excessive market pessimism may have created oversold opportunities. Compared to the US, Chinese assets have been relatively less affected by this software valuation correction, and domestic software was already trading at lower multiples to begin with. Industry research institutions point out that while SaaS may "die," more software will be "reborn"—an AI agent-driven renaissance is underway, and software's value as AI's carrier layer is actually being underestimated.

AI Application Commercial Value Gradually Materializing

Early August brought a wave of catalysts for the "twin giants" of AI large models: MiniMax-W was officially included in the Hong Kong Stock Connect on August 6, with shares climbing over 30% in just two trading days of eligibility. DeepSeek announced across-the-board API price increases, with peak-hour V4 Pro output pricing reaching 27 yuan per million tokens—signaling a shift in large model pricing strategy from "subsidized customer acquisition" to "value-based pricing." A recent Goldman Sachs research report indicates that with growing demand and accelerating commercialization, China's AI models are expected to reach a combined ARR of $13 billion by end-2026, up from the previously projected $10 billion. China Merchants Media data shows MiniMax's ARR surged from approximately $100 million in December 2025 to roughly $430 million by April 2026. As the industry transitions from "technology competition" to "commercial realization," monetization capability and ecosystem maturity have become the new competitive battleground. Continued breakthroughs in domestic large models are opening up scalable monetization opportunities for downstream AI applications, providing substantive earnings support for the software development sector.

Domestic Substitution Accelerating as 2027 Full Replacement Deadline Approaches

SASAC Document No. 79 explicitly requires all central and local state-owned enterprises to complete domestic substitution of their information systems by the end of 2027, achieving 100% replacement. The 2026-2027 period represents the critical phase for concentrated delivery and volume ramp-up. Huatai Securities notes that with this milestone approaching, both party-government and industry-wide domestic substitution are expected to enter a comprehensive replacement peak. Cinda Securities estimates the total market opportunity for domestic substitution across government agencies, public institutions, and state-owned enterprises at approximately 632.5 billion yuan. GF Securities, reviewing two previous domestic substitution cycles, believes the industry-wide PC and server market alone could exceed one trillion yuan—significantly larger than the government-only figure. On the demand side, 2026 marks the third phase of domestic substitution entering full industry penetration, with energy, military, construction, pharmaceuticals, and other "2+8+N" sectors simultaneously launching localization overhauls. In Q1 2026, contract liabilities in the domestic substitution sector rose 31.2% year-over-year, with orders and advance payments continuing to climb, strengthening certainty around earnings delivery.

Financial Sector Domestic Substitution Provides Earnings Direction Signal

Financial technology leaders have moved their domestic substitution projects from "pilot demonstrations" into scaled, normalized delivery. Hundsun Technologies has disclosed completion of 100% full-stack domestic substitution adaptation across all core business systems, with over 2,000 domestic substitution projects implemented cumulatively, alongside published full-stack solution blueprints and engineering implementation standards. Financial institutions continue advancing localization across core systems, databases, and middleware—domestic substitution is progressing from peripheral applications to core trading and settlement systems, with the performance and stability of domestic software and hardware now validated in critical scenarios. As one of the first "2+8" key industries to achieve scaled deployment, the delivery pace of financial sector domestic substitution orders provides the entire software development industry with an earnings anchor and directional benchmark.

Software Policy Support Continues to Intensify

Software industry policy support keeps strengthening. Suzhou has released its Action Plan for Promoting High-Quality Development of the Software Industry (2026-2027), targeting six major areas including foundational software, industrial software, and application software, aiming for software industry output to exceed 400 billion yuan by 2027. Nanjing is advancing "AI + Software" industry development, positioning itself as "China's first city for intelligent software industry." From "AI+" action plans to government procurement standards for foundational software and hardware, policies continue driving intelligent upgrades and self-reliance in the software industry. 2026 marks a resonance window where policy dividends and industrial innovation converge for the software sector. Software-as-a-Service is evolving toward "Generation-as-a-Service/Agent-as-a-Service" (GaaS), with AI agents reshaping software delivery and pricing rules—shifting from "selling tools" to "selling outcomes"—opening up significantly larger value creation potential for the industry.

Low Valuations Highlight Safety Margin

In early 2026, as "large models devouring software" narratives roiled sentiment, AI application segments traded at notable valuation discounts versus computing power and model layers, establishing software development as a "lower water level" segment within the AI supply chain with comparatively attractive valuations and stronger safety margins. Combined with AI empowerment and domestic substitution drivers, the software development sector appears positioned to seize momentum. Data through August 18 shows the software development index trading at a PE (TTM) of 173.91 times—lower than over 75% of its historical trading range since inception—highlighting compelling valuation efficiency and safety margins.

Positioning for the Opportunity

While the software industry overall is in an upward cycle, predicting which specific sub-sector or individual stock will outperform remains challenging. Rather than betting on a single niche, investors may consider the Software Development ETF Huabao (159036), whose underlying index covers the entire industry with 107 constituent stocks, providing broader and more balanced exposure designed to capture the industry's beta returns. Data shows the ETF's constituents span AI+finance, AI+healthcare, AI+office, AI+education, AI+cybersecurity, and AI+government applications, with AI application concept stocks representing 46.56% of total weight as of July 15. Top ten constituent weightings are available in the product's official documentation.

Fee Disclosure: The Software Development ETF Huabao charges no sales service fee. When subscribing or redeeming fund shares, the agency broker may charge commissions up to 0.3%. On-exchange trading fees are subject to actual charges by the securities company.

Risk Warning: The Software Development ETF Huabao passively tracks the CSI All Share Software Development Index, with a base date of December 31, 2021, and a publication date of March 29, 2023. The fund is issued and managed by Huabao Fund; distribution institutions do not bear responsibility for the product's investment performance or redemption obligations. Investors should carefully read the Fund Contract, Prospectus, Fund Product Information Summary, and other legal documents to understand the fund's risk-return characteristics and select products appropriate to their own risk tolerance. The fund manager assesses this fund's risk level as R3-Medium Risk, suitable for balanced (C3) and above investors; suitability matching opinions are subject to the sales institution's determination. Sales institutions (including the fund manager's direct sales channels and other sales channels) conduct risk assessments of the fund in accordance with relevant laws and regulations. Investors should promptly review the suitability opinions issued by sales institutions and defer to their matching results. Suitability opinions across different sales institutions may not be consistent, and the risk level rating issued by fund sales institutions must not be lower than the risk rating assessed by the fund manager. Differences may exist between the fund contract's risk-return characteristics and the fund's risk level due to differing considerations. Investors should understand the fund's risk-return profile and carefully select fund products based on their own investment objectives, time horizon, experience, and risk tolerance, assuming investment risks themselves. Registration of this fund with the CSRC does not constitute a substantive judgment or guarantee regarding the fund's investment value, market prospects, or returns. Past performance and net asset value levels do not predict future performance, and performance of other funds managed by the fund manager does not guarantee the performance of this fund. All funds carry risk; invest with caution.