UBS Group downgraded the rating of China's software industry, pointing out that generative AI is disrupting the traditional SaaS logic. AI is forcing software companies to shift from high-margin standardized subscriptions to low-margin customized services, resulting in 'revenue growth without profit growth.' The market valuation system is transitioning from a focus on growth through price-to-sales ratios to an emphasis on profitability and cash flow through price-to-earnings ratios, leading to a revaluation of the industry.
Over the past decade, the most appealing narrative for the capital markets regarding the software industry has been “SaaS-ification” — as long as subscription revenue grows, current losses are not only tolerated but even considered a necessary cost for capturing market share.
The market has paid an extremely high valuation premium, betting on a highly profitable future driven by economies of scale. However, the explosion of generative AI is dismantling the cornerstone of this logic: AI is not making software more standardized or more profitable; instead, it is forcing software companies to revert to customized service providers that “sell manpower.”
On February 10, according to information from the ZF Trading Platform, UBS Group noted in its latest research report that the rapid iteration of large language models (LLMs) is triggering a fundamental reassessment of the “standardized SaaS” model. The once-premium halo of SaaS is fading, no longer supporting long-term growth narratives but demanding immediate cash flow and profitability.
Historically, software companies enjoyed high valuations because scalable subscription revenue could bring operating leverage and high margins. However, as LLM-native Agents may commoditize standard SaaS workflows, standardization might no longer be an asset but rather a liability.
In the view of UBS analyst Sara Wang, revenue growth without profitability protection is no longer a sustainable investment thesis. The market is shifting from a sales-based (Sales) valuation framework to one based on price-to-earnings (P/E) or free cash flow (EV/FCF).
The research report stated that this shift in valuation paradigm has directly led to downgrades across the entire sector. As AI forces software companies to provide more customized services to meet customers’ ambiguous needs, their business models begin to resemble low-margin “IT services.”
Similarly, on February 4, Morgan Stanley also pointed out in a research report that this marks the beginning of a long-term narrative shift, ending the unreasonable upward trend of the sector by January 2026. Although the subscription-based SaaS model is not prevalent in China, traditional software (especially tool-based software) still faces disruption risks in the long term.
However, the report also mentioned that existing software vendors still have a time window to embrace new technologies and can leverage their vast installed customer base to fend off disruptors, but overall risk remains skewed to the downside.
The Curse of Standardization: The SaaS Premium is Disappearing
According to the research report, in the past, the valuation logic for leading Chinese software companies was based on a “convergence premium” — investors bet that they would eventually achieve high-margin standardized subscription models like Salesforce or Adobe. Therefore, despite having much lower profitability than their American counterparts, Chinese software stocks' price-to-sales ratios (EV/Sales) have long been benchmarked against U.S. stocks.
However, UBS Group believes that this logic has been completely disrupted by AI. Year-to-date, although there is no evidence to suggest that SaaS profitability has been overturned by AI, the share prices of leading U.S. software companies have fallen by 10%-40% amid the fading premium of the SaaS subscription model.
When large AI models (LLMs) are capable of replacing standardized workflows, software companies are forced to retreat to the old path of 'customization.' If the terminal value of a standard SaaS product faces the threat of being replaced by AI and requires more customized development for delivery, then the high valuation rationale for SaaS ceases to exist.
UBS Group stated that China's software industry valuation system is being compelled to decouple from SaaS and revert to traditional IT service valuations — implying that the price-to-earnings ratio (P/E) or free cash flow (EV/FCF) will replace the price-to-sales ratio as the new pricing anchor.
The AI Trap of 'Revenue Growth Without Profit Expansion'
In its research report, UBS Group cited data from the Ministry of Industry and Information Technology showing that since the 'DeepSeek Moment' in early 2025, revenue growth in China’s software industry has indeed accelerated, but profit margins have been declining.
The bank believes this reveals a harsh reality: while AI has driven an increase in corporate IT spending, this demand does not point towards standardized software products.
To bridge the gap between customers’ ambiguous needs and rapidly iterating large models, software companies have had to invest significant manpower in providing customized services. Under this model, revenue growth can no longer translate into margin expansion and may even drag down profitability due to the heavy burden of customized development.
This presents an awkward combination: enterprises are willing to spend on AI, but the money flows more towards delivery and transformation rather than the high-profit increments brought by standardized subscriptions. For valuation purposes, revenue growth no longer automatically equates to margin expansion.
In its research report, UBS Group broke down the bottlenecks of AI monetization for software companies into three parts:
1) Insufficient AI capabilities — the current product quality is not compelling enough for customers to continue paying;
2) Immature digital ecosystem – fragmented data, outdated hardware, and prolonged implementation cycles.
3) Credibility issues with AI expertise – traditional software vendors may be compared by clients to AI ventures and cloud providers in terms of AI capabilities.
However, UBS Group also noted an opportunity: the challenges themselves will leave room for vendors who can offer end-to-end solutions, understand vertical industries, and are able to cross-sell traditional digital products.
The cost is that as new versions of models emerge every 2-3 months and more large models claim to enter vertical scenarios, software companies must iterate and deliver faster; and 'more customization' often means harder to standardize and more difficult to expand profit margins.
Editor/Lambor
