The rapid advancement of artificial intelligence models over the past year has triggered a crisis of confidence in the global software-as-a-service (SaaS) sector. Investors, fearing that new agentic AI products will eliminate the need for traditional software, have driven down the valuations of major US software companies. Salesforce and Adobe are both down approximately 30 percent this year. This “SaaSpocalypse” sentiment has also extended to China, where the Hang Seng China A Software & Services Index has fallen 19 percent from its mid-January high.
However, a new analysis from HSBC Qianhai Securities suggests that this investor pessimism is fundamentally misaligned with the reality on the ground in China. According to Yiran Liu, head of A-share IT software research at HSBC, China’s legacy software companies are not being disrupted by AI model companies. They are being turbocharged by them.
The core of Liu’s argument is that AI model companies, while possessing powerful foundational technology, lack the deep industry know-how and vertical data required to meet specific enterprise needs. Legacy software companies, by contrast, have spent years accumulating that exact experience. The most likely outcome in the Chinese market, therefore, is not disruption but collaboration, where model companies and legacy software firms serve enterprises in tandem.
The Kingdee Example: Monetizing AI Early
The financial results of China’s leading SaaS providers appear to support this thesis. The South China Morning Post shows how Shenzhen-based Kingdee International Software Group, one of the country’s largest enterprise software companies, provides a compelling case study. Last month, Kingdee announced revenue guidance of 1 billion yuan (approximately US$146 million) for its AI products this year, a remarkable 180 percent jump.
More significantly, the company turned a profit last year after five consecutive years of losses. This turnaround was driven by the launch of a suite of AI-native products covering financial analysis, recruitment, and contract review. Kingdee’s AI financial tool alone has already amassed almost 400,000 registered users and 35 enterprise clients.
This accelerated growth reflects legacy software firms’ ability to monetize AI early in the adoption cycle. Because they already possess the industry knowledge and established customer relationships, they are uniquely positioned to provide immediate value to downstream users during this early stage of AI-driven industry transformation. They are not building the foundational models; they are building the applications that make those models useful to specific businesses.
The Limits of Pure-Play AI Startups
The dynamic between legacy software firms and pure-play AI startups in China is complex. There has been significant excitement about the growth potential of AI startups globally, and China is no exception. Two of the country’s leading players, Beijing-based Zhipu AI and Shanghai-based MiniMax, have seen their stocks soar in Hong Kong since their January listings, at times even surpassing the market capitalizations of established tech giants like Baidu and JD.com.
However, these startups face a structural challenge when it comes to enterprise adoption. As Liu points out, if a company like MiniMax wants to deploy its model in the manufacturing industry, it needs suppliers and established channels, which it currently lacks. To penetrate different industries effectively, these AI startups must rely on China’s legacy software firms.
Kingdee, for instance, significantly ramped up its penetration of China’s manufacturing industry last year, securing deals with major enterprises including Liuzhou Iron & Steel, the largest iron and steel conglomerate in South China. AI startups need access to those kinds of enterprise relationships to scale their revenue.
Interestingly, AI startups themselves are also consumers of traditional SaaS products. In a recent WeChat post, Kingdee revealed that it helped MiniMax upgrade its internal research and development expense management system prior to its January IPO. This upgrade resulted in a 30 percent improvement in MiniMax’s accounting efficiency, demonstrating that even the companies building the frontier models still rely on legacy software for their own operations.
FOMO and the Enterprise IT Budget
The enterprise AI landscape in China is also shaped by the absence of the US “big three” AI model developers, that is OpenAI, Anthropic, and Google. Because these services are not available in China, Chinese enterprises are experimenting with domestic models from companies such as DeepSeek and ByteDance to improve internal efficiency and productivity.
However, the capabilities of these domestic models, particularly in terms of enterprise applications, remain limited. According to Liu, AI model providers have yet to develop robust enterprise applications of their own that are ready for widespread adoption. The computational resources are constrained, and the application development ecosystem is not yet mature enough to produce highly polished products at this early stage.
This gap between the promise of AI and the current reality of enterprise applications has created an opportunity for legacy software companies. The frenzy over agentic AI products like OpenClaw at the start of the year instilled a sense of “FOMO” (fear of missing out) among Chinese enterprises. Companies are eager to adopt AI but lack the internal capability to do so effectively.
This dynamic presents a significant opportunity for software companies to push their clients to increase their IT budgets. Legacy SaaS providers can offer the AI integration enterprises demand, capturing early AI demand by developing applications on top of foundational models. While market sentiment may fluctuate with the latest AI breakthroughs, the fundamental reality in China’s software market is that companies with the deepest industry experience are currently best positioned to profit from the AI transition.