Hook
The numbers are stark. A recent Lazard survey of private equity secondary market investors reports that 96% have already altered their approach to software investments. The reason? Artificial intelligence. This is not a future risk. It is a current capital pricing event. The math does not weep, it merely liquidates outdated valuation models.
Context
Lazard, a global investment bank, conducted a market survey on the impact of AI on the software industry within the private equity secondary market. The survey targeted institutional investors, fund managers, and limited partners actively trading stakes in software companies. The key findings: 96% of respondents have changed how they invest in software, 91% identify proprietary data advantages and network effects as the core moat, and a significant portion are moving capital to other sectors. The survey was conducted in mid-2025, capturing a moment of heightened AI anxiety.
This is not a tech report. It is a capital allocation signal. When 96% of sophisticated investors recalibrate, the market has already moved. The only question is whether the consensus is correct or whether it creates mispricing opportunities.
Core: The On-Chain Evidence Chain (in Capital Markets Terms)
The survey provides a rare window into the behavioral shift of capital allocators. Let me dissect the data with the rigor of a forensic auditor.
1. The 96% Threshold: A Behavioral Regime Change
When 96% of a surveyed population admits to changing behavior, it is not a trend. It is a regime change. The remaining 4% are either delusional or have a unique thesis that the market has not yet priced. In the private equity secondary market — a market known for its conservative, long-term orientation — this level of response is unprecedented. The last time we saw such a shift was during the 2020 COVID crash, but that was a liquidity shock, not a structural paradigm shift.
2. The 91% Consensus on Moat: Data + Network Effects
91% of investors now believe that the only sustainable moat for a software company is proprietary data combined with network effects. This is a direct rejection of the old SaaS valuation thesis: feature-depth, user habit, and integration ecosystem. AI has commoditized features. GitHub Copilot, for instance, has reduced the marginal cost of building a feature set by an order of magnitude. The user interface is shifting from GUI to conversational/agentic, lowering switching costs. Integration exclusivity is diluted by AI agents that can automatically adapt to any API.
What does this mean for valuation? The old DCF models based on ARR growth and gross margins are now insufficient. The new framework must embed an "AI substitution probability" into every cash flow projection. A typical mid-market B2B SaaS company with $50M revenue, 25% growth, and 80% gross margins — traditionally valued at 5-8x ARR ($250M-$400M) — may face a 15-35% discount if investors perceive a 30% chance of core functionality being replaced by an AI-native competitor within five years. That is a $40M to $140M valuation haircut.
3. Capital Migration: The Flow of Liquidity
Respondents are not just adjusting valuations; they are physically moving capital out of software. Some are rotating into AI infrastructure (compute, data services, AI security). Others are fleeing to non-tech sectors like infrastructure, energy, and healthcare to avoid AI uncertainty. This is a classic risk-off rotation, but it has a specific timing: the survey suggests that investors expect the AI disruption to materialize within 3-5 years, not 10.
4. The Hidden Contradiction: Consensus Is Priced In
Here is the contrarian angle. The 91% consensus on data and network effects as the moat means that factor is already priced. If everyone believes it, the excess return opportunity is gone. The real alpha lies in the factors not yet consensus: compliance moats, distribution advantages, and the sustainability of data exclusivity. Synthetic data is improving. Data regulations (EU AI Act, China’s Generative AI rules) may limit the exclusive use of proprietary data. The question is not whether a company has data today, but whether that data moat will hold in 2028.
5. The Pre-Mortem: Why This Consensus Could Be Wrong
We must apply a pre-mortem risk analysis. The most likely failure mode: the “data moat” is overestimated. If synthetic data generation becomes sophisticated enough to simulate real-world behavior, or if federated learning allows multiple companies to pool data without sharing it, the exclusivity of any single data set may collapse. Meanwhile, the capital herd effect may cause high-quality software assets with genuine data moats to be oversold, creating a contrarian entry point for buyers who understand the nuance.
Contrarian Angle: Correlation ≠ Causation
Investors are conflating AI’s threat with AI’s opportunity. The survey focuses on the defensive side: how to protect software assets from AI. But AI also expands the total addressable market for software. AI-native companies (AI agents, autonomous workflow platforms) are creating entirely new categories. The market is not yet pricing “AI-enhanced” software companies that successfully integrate AI to raise unit economics. Look at Salesforce: its AI-powered sales predictions and copilot features have already increased customer lifetime value. The next wave of re-rating will come from companies that can demonstrate “AI revenue” as a separate line item.
Takeaway: The Next Signal to Watch
Over the next 6-12 months, track two metrics: (1) the trading volume and discount rates of software stakes in secondary markets like Forge Global and NYPPEX — a widening discount signals deepening fear; (2) the earnings calls of public SaaS leaders (Salesforce, ServiceNow, Atlassian) for any mention of AI-related net revenue retention. If the discount widens beyond 30% for companies with real data moats, the contrarian play is to buy. If AI revenue becomes a material driver, the consensus shift will reverse. The math does not weep, but it also does not wait. I do not predict the future, I verify the past. And the past says that when 96% of investors run in one direction, the smart money is already looking the other way.