Bernstein: Software stock crowding remains at historic lows; crowded trades have shown reversal signals; Microsoft Corporation (MSFT.US) becomes a "must-own"

Entering the fourth quarter, just how "crowded" are software stocks? Bernstein has the answer: in the third quarter, software sector crowding edged up slightly from July but remained significantly below the 25-year historical average. Chief analyst Moedler's team concluded on this basis that software stocks as a whole remain "out of favor," but within that neglect lie structural opportunities. So-called "crowding" measures whether actively managed funds' holdings in a given stock have become overly concentrated. The more crowded a stock, the more vulnerable it is to collective selling and liquidity stampedes when the market comes under pressure. Data shows that in months when the VIX volatility index was above its average, the most crowded portfolio underperformed the least crowded portfolio by about 40 basis points per month. The first half of this year saw a rare exception: crowded stocks as a whole outperformed neglected stocks by 13.3%, with the TMT sector reaching as high as 51%; but the trend reversed in the second half, with the most crowded names in TMT instead underperforming the least crowded portfolio by 30.2%. Bernstein warns that crowding is critical to downside risk control, diversification, and subsequent returns. By subsector, software is sharply divided internally. Databases, consumer-oriented infrastructure, and cybersecurity remain the three "most crowded" areasSnowflake (SNOW.US) jumped from once being the "least crowded" to the "most crowded" stock in coverage, followed by Datadog (DDOG.US), Palo Alto Networks (PANW.US), Twilio (TWLO.US), and Okta (OKTA.US). By contrast, traditional application software priced by "seats," such as Adobe (ADBE.US), HubSpot (HUBS.US), GitLab (GTLB.US), and Zoom (ZM.US), continues to be neglected due to lackluster results or strategic adjustments. Among application SaaS, ServiceNow (NOW.US), Atlassian (TEAM.US), and Salesforce (CRM.US) have seen crowding recover somewhat as revenue growth stabilizes. At the mega-cap level, Microsoft Corporation (MSFT.US), though its crowding has pulled back, remains positive and is the report's explicitly favored "must-own" nameits GAAP operating margin is stable at around 45%, Azure and Office 365 Copilot paid adoption is accelerating, yet capex growth is slowing, and its valuation is below its historical midpoint. Oracle Corporation (ORCL.US), meanwhile, has slid from "most crowded" to third-least crowded on capital constraints and data center execution concerns; Bernstein believes its results remain solid and its value could multiply over a 510 year horizon, making its neglect an attractive buying opportunity. SAP (SAP.US) sits in the middle and is viewed as a "compounder." Bernstein's core stance is that the short-term "AI disrupts SaaS" (SaaS apocalypse) narrative has been overblown. Moedler argues that the idea of enterprises building their own software with generative AI is overstated both in terms of the technology maturity cycle and willingness to bear risk, and that complex large-enterprise ERP is even less likely to be built in-house. The real risk is concentrated in seat-priced niches such as help desks, and in point solutions lacking system-of-record data and semantic knowledge. For the software sector as a whole, Bernstein maintains its conclusion that it is "worth investing in," but stresses that selectivity is essentialseeking out names with solid fundamentals that can weather the AI cycle among the neglected, rather than chasing crowded trades that are already overweighted.