Only 4% of institutional investors have not changed their investment approach to software. 91% cite proprietary data and network effects as the primary moat. This is not a gradual shift. It is a systemic repricing event raging across the private equity secondary market, and its shockwaves are already hitting crypto software projects. The math holds, but the humans did not verify it. Assumptions are just risks wearing disguises.
Lazard's survey, released in August of an unspecified year between 2023 and 2025, captures a moment of consensus so strong it borders on dogma. The report, targeting the private equity secondary market for software assets, reveals that nearly all institutional investors now view AI as a deflationary force on software value. Their response is not to flee, but to reallocate capital toward assets with structural moats. For the crypto ecosystem, this is a direct signal: the same forces that are commoditizing traditional SaaS are now eroding the premium on smart contract code.
Context: The Survey as a Systemic Signal
The Lazard survey is not a casual poll. It is a temperature check on the most sophisticated capital allocators in the private markets. The key data points: 91% of respondents identify proprietary data plus network effects as the primary defensible moat. Only 4% have made no changes to their investment methodology. The remaining 96% have either shifted capital away from software or are actively waiting for clarity. This is a textbook example of a paradigm shift in valuation framework.
For crypto, the implications are stark. The industry has long touted “code is law” and “open-source composability” as competitive advantages. But Lazard’s data suggests that institutional investors no longer value code as a moat. They value data and network density. A DeFi protocol that relies solely on a fork of Uniswap’s core logic has zero defensibility. A protocol that accumulates proprietary order flow, user behavior data, and cross-chain liquidity networks has a moat that investors will pay for. Value is consensus; truth is optional.
Core: A Systematic Teardown of Data Moats in Crypto
Let me dissect this from first principles. I spent the 2020 DeFi summer analyzing Compound’s cToken interest rate models, discovering a flash loan edge case in liquidation thresholds that exploited oracle latency. That experience taught me one thing: market efficiency is an illusion during rapid capital influx. The same dynamic applies here. The 91% consensus is not a reflection of deep technical understanding, but a collective response to a perceived threat.
Why “Proprietary Data + Network Effects” is Not a Universal Shield
The claim that proprietary data forms a moat relies on the assumption that large language models cannot replicate the distribution of that data. For crypto, this is partially true. A protocol that holds years of on-chain transaction data, off-chain order book data, and user behavior patterns (e.g., lending, borrowing, liquidation history) cannot be easily replicated by a model trained on public blockchain data alone. The reason: public blockchain data is pseudonymous, but the behavioral signatures are unique. I verified this in my 2021 analysis of Bored Ape Yacht Club’s metadata storage on IPFS, which relied on a single AWS node. The illusion of decentralization masked a centralized data dependency. The same flaw exists in many crypto projects that claim data moats but actually store sensitive data on centralized infrastructure.
The Network Effect Fallacy
Network effects are real, but they require a critical mass of active users generating data. In crypto, many protocols have inflated user counts through sybil attacks or incentives. The Lazard survey’s 91% consensus may be pricing in the existence of a network effect without verifying its authenticity. I have seen this before: in 2017, I spent two weeks mathematically proving that Tezos’ on-chain governance voting did not guarantee consensus stability under Byzantine conditions. The math was ignored by retail hype, but the flaw was real. The same is happening now. Investors are accepting “network effects” as a moat without verifying the degree of genuine user lock-in.
The Valuation Vacuum
The old framework valued software on ARR growth multiples. The new framework will apply an “AI exposure discount” and a “moat quality premium.” But the new framework is not yet standardized. This creates a valuation vacuum. In crypto, this vacuum is already visible: token prices for generic DeFi protocols are compressing, while tokens for protocols with verifiable data assets (e.g., Chainlink’s oracle network, The Graph’s indexing data) are holding value. The exit liquidity is someone else’s regret.
Contrarian: What the Bulls Got Right
Despite my skepticism, the 91% consensus has a kernel of truth. Some crypto protocols do possess genuine data moats that AI cannot easily replicate. For example, liquidity network effects in DEXs like Uniswap or Curve are not just code; they are the result of years of user behavior, arbitrage strategies, and MEV extraction patterns. These are data sets that are continuously generated and are unique to each protocol. A model trained on public blockchain data cannot replicate the fine-grained latency dynamics of a specific liquidity pool.
Furthermore, AI can enhance these protocols. I have been working on a formal verification framework for AI-agent smart contract interactions, and I see potential for AI to improve security auditing, risk modeling, and automated market making. The bulls are right that AI is not purely a threat; it can be a tool to strengthen moats. But the key is that the moat must be built on data, not on code. Correlation is the comfort of the unprepared.
Takeaway: The Accountability Call
The Lazard survey is a wake-up call for crypto project founders and investors. The era of raising money on a whitepaper and a forked GitHub repo is over. The new standard requires verifiable data assets: on-chain transaction history, user engagement metrics, and network density measures. Investors must demand transparency on data provenance, storage architecture, and the degree of real (not sybil) network effects. The valuation vacuum will persist until the market develops a standardized framework for quantifying data moats. Until then, the only safe assumption is that every generic protocol is overvalued. Provenance is a story we agree to believe in.
I have been analyzing systemic fragilities in crypto since 2017. The 91% consensus is a powerful signal, but it is also a trap. Those who blindly accept data moats without verification will be the exit liquidity for those who do the math.
