CITIC Securities: Expect the current painful state of the U.S. software sector to potentially continue for some time

CITIC Securities Research Report states that the US stock software sector is currently facing intense selling pressure. Looking back at history, the best and only response to repeated doubts about technological waves has been corporate performance, and this AI wave is no exception. Compared to AI startups, traditional software companies have clear advantages in customer resources, delivery services, domain knowledge & data accumulation, and are actively responding to potential AI impacts & opportunities through M&A, product & business model adjustments. Who will ultimately benefit & suffer may remain a topic for ongoing discussion over a long period. However, considering the current overall risk-off environment in the US stock market and the slow short-term improvement in software companies’ performance, we expect the painful state of the US software sector to persist for some time. From a mid- to long-term perspective, combined with the underlying logic of AI, enterprise valuation levels, etc., now is the best window to allocate into asset-liability-robust, platform-based application software, cybersecurity companies, and foundational software (data management, ITOM, etc.) providers that continue to benefit from computing infrastructure.

Full Text Below

Forecast | When Will the US Software Sell-Off End?

The US stock software sector is currently under intense selling pressure. Looking back at history, the best and only way to counter repeated doubts about technological waves has been corporate performance, and this AI wave is no exception. Compared to AI startups, traditional software companies have clear advantages in customer resources, delivery services, domain knowledge & data accumulation, and are actively responding to potential AI impacts & opportunities through M&A, product & business model adjustments. Who will ultimately benefit & suffer may remain a topic for ongoing discussion over a long period. However, considering the current overall risk-off environment in the US stock market and the slow short-term improvement in software companies’ performance, we expect the painful state of the US software sector to persist for some time. From a mid- to long-term perspective, combined with the underlying logic of AI, enterprise valuation levels, etc., now is the best window to allocate into asset-liability-robust, platform-based application software, cybersecurity companies, and foundational software (data management, ITOM, etc.) providers that continue to benefit from computing infrastructure.

Event Background:

Since the beginning of the year, with the release of series AI Agent products like Anthropic Claude cowork, market concerns about damage to the US stock software sector have further intensified. Coupled with recent extreme pro-cyclical trading styles and the steady but unremarkable quarterly reports of companies like Microsoft and ServiceNow, market pessimism towards the US stock software sector has rapidly evolved into irrational, stampede-like selling. As of February 4, the IGV ETF has fallen 19.2% year-to-date (while the Nasdaq Composite Index is up 0.1%), and the short-term severity of the sector’s decline is comparable to 2001 and 2016. How long will this painful, disorderly decline last? Is AI affecting the fundamental logic of traditional software? What signals could reverse the current sector dilemma? Our initial thoughts are as follows:

AI Devouring Software: Still Mainly at the Narrative Stage in the Short Term

Recently, startups like Anthropic have launched series AI Agent products, fueling market expectations of a future where everyone can code, insource software, and AI engulfs software. From an objective and rational perspective, whether considering the capabilities of LLMs themselves or their current application scenarios, we believe the market’s current judgment is overly optimistic and unrealistic: 1) At the model level, many academic and industry voices suggest that current LLMs still operate at a probabilistic, statistical game level, have not touched human cognition core, and have clear theoretical limitations in hallucination, multimodal alignment, reasoning, and other five areas; 2) At the application level, analyzing a sample of S&P 500 companies, although mentions of AI in financial reports are increasing each quarter, specific application scenarios mainly focus on coding, customer service, etc., which are either logically simple, rigorous, or tolerant of errors. In more demanding, complex logical chain scenarios, AI still struggles. Recent research from Salesforce also shows that current AI Agents, despite excellent benchmark performance, still have accuracy issues in specific enterprise applications, especially as task steps and complexity increase, accuracy drops rapidly.

Traditional Software: Multiple Factors Limit Short-term Performance Improvement

In the face of AI narrative suppression, the best response is improving performance. However, short-term improvements are unlikely due to: 1) Macro environment—persistent high inflation, high volatility, and large CAPEX investments in AI by major companies, along with ongoing de-leveraging from pandemic bubbles, are reducing hiring of white-collar employees in the US and affecting subscription software sales, while the shift of IT spending towards AI drags down overall enterprise software budgets; 2) Corporate level—facing the GenAI wave, software companies are rapidly adjusting and restructuring their R&D, marketing, and service systems, with aggressive M&A being the most effective short-term approach. Historically, the market has held a negative view of M&A in the software sector, mainly worried about slowing endogenous growth and integration risks, and recent M&A activity has intensified product overlaps and competition concerns. As the US macro environment improves and AI products gradually land, we expect quarterly performance of US software companies to improve, but at a modest pace.

Market Judgment: Extreme Pro-cyclicality, Risk Off

Since the beginning of the year, major indices like IWM and SOX have shown that in an environment of overall risk suppression, the market has pushed pro-cyclical trading to the extreme, abandoning all uncertainties including AI’s impact on software & internet and AI computing power sustainability. In the short to medium term, considering this year’s US midterm elections, the new Federal Reserve Chair, and rapid changes in AI industry narratives, we expect risk-off, high volatility, and short-term performance chasing to remain core market features. The recovery of the US software sector will mainly depend on performance improvements, with attractive valuations reflecting current risk levels but not sufficient as a basis for market rebound.

Risk Factors:

• Unexpected rebound of US inflation
• Escalation of geopolitical conflicts
• Continued tightening of policies and regulations in the tech sector
• Policy and regulatory risks related to private data
• Underperformance of the global macroeconomic recovery
• Macroeconomic volatility leading to weaker-than-expected IT spending in Europe and the US
• Potential ethical, moral, and privacy invasion risks associated with AI

Investment Advice: Build a Portfolio of Asset-Liability-Robust Platform Companies

In every technological wave, the software sector faces serious doubts, and this AI wave is no exception. Compared to AI startups, software companies have clear advantages in customer resources, delivery, domain knowledge & data, and are actively responding through M&A, product, and business model adjustments. Who will ultimately benefit & suffer may remain a long-term discussion. Given the current risk-off environment and slow performance improvements, we expect the painful state of the US software sector to continue for some time, with significant performance improvements likely only after 2026 Q1, during which weaker competitors may exit. From a mid- to long-term perspective, now is the best window to allocate into asset-liability-robust, platform-based application software, cybersecurity firms, and foundational software (data management, ITOM, etc.) providers that benefit from infrastructure investments.

(Source: First Financial)