Bernstein Research Report: Low Crowding in Software Opens a Stock-Picking Window, Earnings Recovery Determines Capital Inflows

Written by Rita

Crowded stocks do not necessarily win all the time. The cumulative return spread of the most crowded quintile versus the least crowded quintile was positive 51.0% in the first half of the year, but turned negative 30.2% in the second half. In its October 28, 2026 report, Bernstein used this sharp reversal as a lens to observe capital flows.

The changes within software are more layered. Bernstein believes that the sector's overall crowding remains well below long-term historical levels. Capital has begun returning to some companies with stable or reaccelerating results. Low crowding provides an opportunity to reassess companies; evidence of growth and AI monetization determines which companies can attract capital back.

Software Overall Remains in a Low-Crowding Zone

The crowding Bernstein refers to mainly means excessive concentration of active funds in the same stocks. Similar holdings and views can expose these stocks to common selling pressure when the market turns. The report uses this indicator to observe capital structure and look for divergence in expectations across companies.

Compared with July, crowding in the software sector rose in the third quarter. The report attributed part of this change to a rebound in application software results: stable or accelerating revenue led the market to reassess the urgency of the AI shock. Even with this recovery, software remains low relative to its own history over the past 25 years.

This assessment does not contradict the broader technology sector's elevated crowding. The report notes that semiconductors and hardware still concentrate higher enthusiasm, while software faces concerns about generative AI replacing applications, compressing seats, and competing for enterprise budgets. Looking at technology stocks as a whole can mask the divergence of capital across different business models.

Bernstein therefore sees an opportunity to reselect software companies. Low crowding has allowed some enterprise platforms to shed overheated holdings, while stable or improving business conditions are again attracting capital inflows. The report combines these two threads, focusing on the gap between operating performance and market confidence.

Capital Concentrates on Usage Growth

As of October 19, Snowflake's standardized relative crowding score among the companies covered in the report was 2.13. Palo Alto Networks was 1.86 and Datadog was 1.80. By contrast, Oracle was -0.37, Adobe was -0.70, and HubSpot was -1.55. These scores clearly show capital's preference differences across databases, infrastructure, and application software.

Snowflake's change was particularly notable. The report noted that increased usage on its core platform, together with revenue growth from AI products, drove business acceleration over the past two quarters. The usage-based pricing model makes increases in cloud workloads more easily translate into revenue, and investors have therefore increased their allocation enthusiasm.

But Bernstein did not raise its assessment in line with holding enthusiasm. For Snowflake, the report remains concerned about data warehouse market maturity, intensifying competition, and long-term competitiveness of AI tools and data federation solutions. Rising crowding can reflect improved growth, but it may also embed optimistic expectations about the future into the stock price ahead of time.

Datadog presents a similar divergence. Bernstein is bullish on usage growth from other AI-native customers while acknowledging pressures such as partial loss of large customers and slower non-AI cloud workloads. The report expects consumption growth from other AI-native customers to offset these pressures and drive continued revenue acceleration. Growth in AI workloads remains an important basis for institutional investors' confidence in this business model.

Application Recovery Must Be Proven by Results

The recovery of ServiceNow and Atlassian illustrates the specific conditions for capital returning to application software. Bernstein noted that revenue at these companies is reaccelerating, and their enterprise platform positions provide more solid support than a pure AI narrative. Market concerns about application software have eased, but the recovery has mainly been concentrated in companies that have delivered operating evidence.

The report is cautious about the thesis that enterprises will build their own complex applications as a substitute. Taking enterprise resource planning (ERP) systems as an example, large enterprises must bear process, data, and operational risks. The development of generative AI tools does not automatically eliminate these migration and replacement costs. Bernstein therefore sees durability in some core enterprise platforms.

Risks still need to be segmented by product and pricing method. The report noted that role automation in areas such as customer service may reduce demand for seat-based software. Whether vendors can offset losses by charging for intelligent agents remains to be verified. Single-point tools that lack core data and important business semantics are also more exposed to replacement pressure.

Bernstein sees growth in Oracle's cloud infrastructure, enterprise applications, and database businesses. Capital constraints and data center build-out execution continue to weigh on market confidence. For HubSpot, the report remains cautious due to weakening growth expectations, lower management confidence, and a lack of catalysts. Neither company is crowded, yet institutional views differ markedly.

The Microsoft case complements this framework. The report argues that concerns over capital expenditure returns and Copilot adoption have dampened positioning enthusiasm. Recent business performance, however, provides a basis for reassessment. Whether capital returns depends on reconciling growth, margins, and AI investment returns on the same operating ledger.

Crowding metrics should also be assessed alongside valuation.

In its risk model, Bernstein treats crowding as a risk arising from shared positions. When investors simultaneously reduce risk, stocks tied to the same narrative face concentrated selling pressure. Holdings originally spread across different companies may become highly correlated during exits.

In the 10% of months identified in the report when crowded trades performed worst, the most crowded quintile of stocks underperformed the market average by 3.0%. The least crowded quintile outperformed the average by 3.3%. Bernstein therefore believes that including stocks with different levels of crowding in a portfolio can help diversify common-trade risk. When popular holdings face concentrated selling, less crowded assets have historically provided protection.

CrowdStrike illustrates the importance of existing holdings. Bernstein notes that the company's institutional ownership is already high, limiting further accumulation, so actual crowding may exceed the model score. The report is positive on its business competitiveness but views valuation as elevated, making the growth acceleration expected by the market harder to achieve.

Following the report's stock-selection approach, opportunities in the software sector arise from narrowing dispersion. Some enterprise platforms alleviate replacement concerns through stable growth, some usage-based companies monetize AI demand, and other less crowded companies still need to address growth and execution issues first. The most concentrated direction for capital inflows remains companies whose operational improvements can ease market concerns.

Capital inflows into software have already emerged, but the recovery has not yet become broad-based. This Bernstein report offers a lens into positioning structure; ultimately, business fundamentals must answer: can growth be sustained, how much incremental revenue will AI generate, and how much improvement is already priced into current valuations.

Disclaimer

This article is a compilation and interpretation by TideFlow Research of a third-party brokerage research report (Bernstein, October 28, 2026), combined with publicly available market information. The ratings, price targets, earnings forecasts, and related judgments cited herein are the views of the brokerage's analysts and represent only their firm's stance. They do not represent the views of TideFlow Research and do not constitute investment advice.

Markets carry risks, and decisions should be made independently. This article should not be used as a basis for buying or selling any securities.

Source: Shenchao TechFlow