Bernstein: Software stocks remain at historically low levels of crowding, reversal signals have appeared in crowded trades, Microsoft (MSFT.US) becomes a "must-have" asset

As we enter the fourth quarter, just how “crowded” are software stocks? Bernstein provides the answer: Crowdedness in the software sector saw a slight rebound in Q3 compared to July, but it remains significantly below the historical average of the past 25 years. Based on this, Chief Analyst Modler’s team judges that software stocks overall remain “out of favor,” though within this lack of popularity there are structural opportunities to be found.

“Crowdedness” refers to whether actively managed funds’ holdings of a particular stock have become excessively concentrated. The higher the crowdedness, the more likely a stock is to suffer collective selling pressure and liquidity crunches when the market comes under stress.

Data shows that in months when the VIX volatility index is above average, the most crowded portfolio underperforms the least crowded portfolio by roughly 40 basis points per month. There was a rare exception in the first half of this year: crowded stocks outperformed unpopular stocks by 13.3% overall, while the TMT sector surged by 51%. However, in the second half, the trend reversed—within TMT, the most crowded names underperformed the least crowded portfolio by 30.2%. Bernstein cautions that crowdedness is critical for downside risk control, diversification, and future returns.

Looking at different sub-sectors, there is sharp internal divergence within software. Databases, consumer infrastructure, and cybersecurity are still the three most crowded directions—Snowflake (SNOW.US) has leapt from being one of the “least crowded” to the most crowded stock in the coverage universe, followed closely by Datadog (DDOG.US), Palo Alto Networks (PANW.US), Twilio (TWLO.US), and Okta (OKTA.US).

In contrast, traditional application software priced by “seat count”—such as Adobe (ADBE.US), HubSpot (HUBS.US), GitLab (GTLB.US), and Zoom (ZM.US)—continues to be neglected due to mediocre performance or strategic shifts. Within Application SaaS, stocks like ServiceNow (NOW.US), Atlassian (TEAM.US), and Salesforce (CRM.US) have seen crowdedness recover somewhat as their revenue growth has stabilized.

At the giant level, Microsoft (MSFT.US) has seen crowdedness decline but remains positive, marking it as a “must-own” according to the report. Its GAAP operating margin is steady at around 45%, payments for Azure and Office 365 Copilot are accelerating, capital expenditure growth is slowing, and its valuation is below its historical average.

Oracle (ORCL.US) slipped from being “most crowded” to third from the bottom due to capital constraints and concerns around data center execution. Bernstein believes its performance remains solid and its value could expand several-fold over a 5-10 year horizon, making its current unpopular status an attractive buying opportunity. SAP (SAP.US) is in the middle, regarded as a “compound interest stock.”

Bernstein’s core stance is that short-term narratives of an “AI-induced SaaS catastrophe” are exaggerated. Modler believes companies developing their own software with generative AI is overestimated both in terms of technology cycle and risk appetite, and that complex, large-scale enterprise ERPs are even less likely to be built internally. Real risks are concentrated in seat-based segments like helpdesk software and in point solutions lacking systematic data and semantic knowledge.

For the software sector as a whole, Bernstein maintains its conclusion that it is “worth investing” in, but stresses the importance of selectivity—find fundamentally solid names that can weather the AI cycle among the out-of-favor stocks, rather than chasing already crowded trades.