In the AI era, what does the software industry rely on to win the competition?
In September this year, the Ministry of Industry and Information Technology issued the "AI + Software" Special Action Implementation Plan, deploying efforts to optimize the chain of intelligent programming tools and accelerate the intelligent upgrading of software products. China's enterprise-level AI agent market reached 21.2 billion yuan in 2025 and is expected to exceed 332 billion yuan by 2029, with a compound annual growth rate of 107%. The article argues that as AI lowers the threshold for programming, the value of industry knowledge and professional experience becomes increasingly prominent.
Body · AI translation
In September this year, the Ministry of Industry and Information Technology issued the "'AI + Software' Special Action Implementation Plan" (hereinafter referred to as the "Implementation Plan"), proposing measures such as optimizing the chain of intelligent programming tools and accelerating the intelligent upgrading of software products, in order to speed up the integrated application of artificial intelligence technology in the software and information technology services industry.
The rapid development of AI (artificial intelligence) technology is profoundly changing the production methods, product forms, and service models of software, and has also posed a new question for the software industry: when AI is increasingly capable of writing code and the threshold for programming keeps falling, what will the software industry rely on to win in competition?
In the past, turning an idea into a software product required going through multiple stages such as requirements analysis, program writing, and testing and revision, consuming a great deal of manpower and time. Today, by telling AI the requirements, it can generate code, find errors, and even coordinate and complete some development tasks. According to media reports, a software development company in Chongqing had three AI agents respectively take on the work of product manager, project manager, and technical manager, with engineers becoming "supervisors" responsible for scheduling and quality control. In the code-writing stage, the coverage rate of AI tools has risen sharply, and some companies have even achieved full coverage.
In terms of product forms, software in the past mostly executed instructions passively, but now some agent software can understand intent and complete tasks autonomously; the service model has shifted from delivering a program with a fixed set of functions to continuously providing intelligent applications. Data shows that China's enterprise-level AI agent market reached 21.2 billion yuan in 2025 and is expected to increase to 44.9 billion yuan in 2026, and is expected to exceed 332 billion yuan in 2029, with a compound annual growth rate as high as 107%. The "Implementation Plan" makes arrangements in areas such as advancing the transformation of software production, accelerating the intelligent upgrading of software products, and cultivating new business forms of agent software, precisely because it focuses on this series of new changes and new challenges.
While artificial intelligence technology lowers the threshold for software production, it also makes the value of industry knowledge and professional experience increasingly prominent. Artificial intelligence can help developers write programs faster, but the ability to understand industries and requirements cannot be developed overnight.
Take scheduling software in the manufacturing sector as an example: how different processes connect, under what circumstances equipment needs to be shut down, and how to balance delivery schedules with energy consumption all require an in-depth understanding of production scenarios. In the first eight months of this year, China's software business revenue reached 10.28 trillion yuan; China's manufacturing value added has ranked first in the world for 16 consecutive years. The vast software application market provides abundant demand for new software, while the complete manufacturing system has accumulated a large amount of production data, engineering methods, and industry knowledge. The "Implementation Plan" proposes supporting qualified industrial enterprises to join with software companies to encapsulate knowledge such as industrial experience and process parameters into reusable model libraries and industrial components, and to develop industry-specific intelligent industrial software. By using artificial intelligence technology to integrate professional experience into algorithms and software products, so that software truly understands production processes and adapts to complex working conditions, the advantages of China's real economy are expected to be further transformed into advantages in the software industry.
Seizing new opportunities also requires new technologies and real skills to reinforce each other.
Software development has sped up, but the requirements for quality and reliability will not be lowered. Whether a piece of AI-generated code placed in a complex project can run stably for a long time, whether it can handle abnormal situations, and whether it can ensure data security all need to withstand repeated testing.
It should be clearly recognized that "AI + Software" has not changed the underlying logic of software development. Physical principles and mathematical methods remain issues that developers need to continue to study in depth. The core of solving specific engineering problems requires long-term research accumulation, and extremely high precision requirements demand unconventional code design as well as validation in massive industrial scenarios. Artificial intelligence can empower software development, but it cannot replace the thinking of scientists and engineers.
It is necessary both to make good use of artificial intelligence and accelerate the pace of intelligent software upgrading, and to continue strengthening technical breakthroughs in core algorithms, basic architectures, and other areas. Promote deep cooperation between software companies and industry users, so that products are repeatedly refined in real scenarios; strengthen the building of open-source communities and promote technology sharing and collaborative innovation. In particular, attention should be paid to quality and safety, with code review and risk detection running through the entire development process.
Looking back at the historical development of the software industry, from personal computers to the mobile internet, every technological change has given rise to new products and new companies, and has also rewritten the competitive landscape of the industry. Now, artificial intelligence has opened another new door. By transforming a huge industrial scale into strong technical strength and turning rich application scenarios into software products, China's software industry is expected to win greater development space in the new round of technological revolution and industrial transformation.