While everyone is still debating whether AI will replace software engineers, the data from Lazard’s latest private equity secondary market survey delivers a quieter but far more decisive verdict: 91% of institutional investors now believe that proprietary data plus network effects are the only defensible moats for software companies. Only 4% have not changed their investment approach. This is not a gradual shift—it’s an avalanche. The market has stopped arguing about if AI will disrupt software and is now positioning for how and who survives.
Context — The $1T Signal in PE Secondaries
The private equity secondary market is the canary in the coal mine for institutional capital flows. With over $100 billion in annual transaction volume, it reflects how sophisticated LPs and GPs price risk across asset classes. Lazard, as a leading advisory firm, has been tracking this pulse. Their survey, conducted in June 2025 (based on the report’s timeline), captures a consensus that has formed with unnerving speed. When 91% of investors agree on a single moat thesis, it’s not a debate—it’s a new orthodoxy. The implication: old software valuation frameworks (ARR multiples, growth rates, net dollar retention) are being discarded. A new paradigm is emerging, and the market is scrambling to price AI exposure as a discount or premium factor.
Core — The Valuation Vacuum and the New Moat Economics
Let me be blunt: the old playbook is dead. When only 4% of investors haven’t changed their methods, the entire reference class for software pricing has shifted. Follow the liquidity, ignore the hype. The capital that once flowed into 'growth at all costs' SaaS is now rotating into data infrastructure, AI-native platforms, and companies with structural network effects. The 91% consensus on 'proprietary data + network effects' is not a qualitative opinion—it’s a quantitative re-rating signal.
Chaos is data in disguise. What the survey reveals is that the market is now assigning a 'moat quality premium' to software assets. Companies with truly unique data sets (e.g., vertical industry transaction logs, compliance-sensitive behavioral data) and strong network effects (e.g., two-sided marketplaces, collaborative platforms) will command higher multiples. Meanwhile, generic SaaS—think simple CRUD interfaces, basic reporting, or customer support automation—will face a systematic discount. The algorithm has no conscience. It doesn’t care about your last quarter’s revenue beat; it cares about whether your data is replicable by a foundation model.
But here’s the catch: this new valuation framework is still a blank slate. There is no standardized 'AI exposure score' yet. Every GP is making up their own spreadsheet. This creates a valuation vacuum—a window where the old rules don’t apply and the new ones aren’t codified. Volatility is the price of admission. For those who can identify which companies have genuine data moats versus marketing claims, this is the alpha opportunity of the cycle.
Contrarian — The Consensus That’s Too Perfect
A 91% consensus should make every skeptic’s spidey sense tingle. In my 29 years of watching markets, extreme agreement is often the precursor to a blind spot. Here are two that the survey glosses over:
First, the time dimension of moats. The proprietary data advantage is only as durable as the frontier of AI capabilities. With synthetic data, federated learning, and context windows expanding from 4K to 1M+ tokens, what is 'unreplicable' today may be inferable tomorrow. Investors are pricing the present moat, not the future erosion. Second, the regulatory risk. Data as a moat runs headfirst into privacy laws and data portability mandates. The EU AI Act, China’s Data Security Law, and even California’s CCPA could force companies to open their data vaults, collapsing the moat overnight. The algorithm has no conscience, but regulators do—and they are watching.
Moreover, the survey’s assumption that 'traditional software + AI features' is insufficient may underrate the value of reliability in enterprise B2B. Hallucination issues in LLMs give deterministic software a defensive window that the market is ignoring. The 91% might be over-correcting for the hype cycle.
Takeaway — Positioning for the Next 24 Months
The Lazard survey is not a snapshot; it’s a roadmap. The next 12–18 months will see a wave of M&A as capital-rich buyers acquire data-rich but capital-poor software companies. The secondary market’s wait-and-see stance will drive discounts, creating a buyer’s market for those with dry powder. For crypto-native investors, the parallel is clear: the same data+network moat thesis applies to DeFi protocols, L1/L2 ecosystems, and data DAOs. But the transparent nature of on-chain data changes the game—moats are harder to hide. Follow the liquidity, ignore the hype. The value is in the data you can verify, not the narrative you can sell.
The quiet avalanche has already started. The question is not whether you’re caught in it, but whether you’re skiing at the front or buried at the back.