Summary of Key Points
The software industry in 2026, particularly SaaS, is undergoing a paradoxical transformation: on one hand, there are concerns that AI agents (which can automatically invoke software to complete tasks) will disrupt traditional business models; on the other hand, new opportunities arise due to the real revenue generated by AI. While AI has not eliminated software, it has changed the way its value is measured—instead of selling "features and licenses," companies now focus on selling the ability of AI to help businesses accomplish work. Financial reports from SAP, ServiceNow, Kingdee, and others indicate that their AI businesses are generating profits, and the capital market has shifted from panic to re-evaluating the value of software vendors with customer bases and data barriers.
Breakdown and Interpretation
1. AI is not a "software killer"; it's changing how software operates
In the past, software companies created "closed systems" (such as ERP or CRM) that users had to access, competing on the "entry point" and "usage duration." However, with the advent of AI agents, users no longer need to open multiple software applications individually; the agents can directly use various software to complete tasks (e.g., generating production plans or processing customer orders). The new approach is for software to "stand on the execution path of the agents," allowing them to call upon the software when needed. For example, Kingdee has made its ERP capabilities available as APIs for AI to use directly; SAP describes its products as "operating systems for the AI era," similar to apps in the Apple ecosystem, where AI leverages underlying capabilities through these interfaces. Therefore, software hasn't disappeared, but its role has shifted from being used by users to being invoked by AI.
2. AI businesses are no longer just a gimmick; they are generating real revenue
Second-quarter financial reports show tangible income from AI:
- SAP's cloud revenue increased by 22%, with AI-related orders growing by 27%;
- ServiceNow's annual AI contract value exceeded $1 billion for the first time, and the number of deployed agents increased ninefold in nine months;
- Kingdee's native AI business revenue grew by 189% year-over-year, with contracted values increasing by 159%.
These figures indicate that businesses no longer view AI as a mere additional feature but are willing to pay for its practical problem-solving capabilities. For instance, in manufacturing, AI can use order and capacity data to generate production plans and handle adjustments—these services are more valuable than simple software functions.
3. Business models are changing: from selling features to selling the results of work
Previously, businesses purchased software based on the number of licenses (e.g., 10 licenses for 10 employees) or specific functional modules (e.g., financial or sales modules). Now, they are more willing to pay for the amount of work completed by AI, such as based on the number of agent invocations, the volume of tasks completed, or even the business outcomes (e.g., how much labor AI saves or efficiency it improves). Kingdee's CFO noted that it took 15 years for revenue to reach $1 billion in the software licensing era, 6-7 years in the SaaS era, and possibly only 2 years in the AI era, as AI products are directly tied to business outcomes, leading to faster growth.
4. Opportunities and challenges for software companies
Opportunities: There are new opportunities to reshape technology stacks around AI agents, such as developing "operating platforms" that enable seamless integration of various software and "governance tools" to ensure compliant AI operations.
Challenges:
- The differentiating advantages of traditional software are weakening; large models can quickly replicate basic functions (e.g., accounting in financial software), which AI may also learn.
- Talent and processes need to adapt: employees must understand how to use AI, and business processes must accommodate AI in decision-making (e.g., production planning no longer relies solely on humans).
- Business model transformation is difficult: shifting from subscription-based models to outcome-based pricing requires redesigning pricing and service systems.
5. Capital market attitude: moving from panic to re-evaluation
At the beginning of the year, there was fear that AI would kill SaaS, leading to declining stock prices. Now, as companies show growth driven by AI in their financial reports, the market is re-evaluating:
- Vendors with established customer bases and industry data, like Workday, have seen their stock prices rise 18% due to potential acquisition rumors, as they can quickly integrate AI into their existing businesses for scaled profits.
- Pure AI startups, lacking customer bases, are at a disadvantage compared to traditional software vendors.
In short, the capital market now values platforms that can monetize AI capabilities, not just the concept of AI itself.
Final Conclusion
The future of the software industry depends on whether your technology can be integrated by AI agents and whether businesses are willing to pay for the work accomplished by AI. This will be the dividing line in the next round of competition.