The numbers are stark. 96% of investors in the private equity secondary market have already changed their approach to software investing because of AI. 91% now believe that proprietary data and network effects are the only durable moats. These are the findings from a recent Lazard survey, a bellwether for institutional capital flows. But as a Token Fund Investment Manager who has lived through the 2017 Zcash audit and the 2022 FTX collapse, I see something the survey does not: the silence between the data points. The real alpha is not in the consensus—it is in what the consensus ignores. And that is exactly where blockchain and crypto-native software are rewriting the rules.
Let me read the docs, then question the whisper. The survey captures a moment of paradigm anxiety. Traditional SaaS companies—built on feature-based value, high gross margins, and slow user migration—are suddenly facing a world where AI can replicate functionality at near-zero marginal cost. The capital markets are responding with a flight to safety: only those companies with a ‘data fortress’ and a sticky network are deemed investable. But here is the hidden truth: the survey’s focus on proprietary data as a moat is a direct reflection of the centralized, walled-garden model of software. It assumes that the best data is locked behind corporate APIs. In crypto, we have spent a decade proving the opposite.
I have seen this movie before. In 2017, during the Zcash alpha audit, we identified three critical gaps in the privacy narrative. The market was obsessed with the idea of ‘anonymous transactions’ but ignored the human-centric need for selective disclosure. The result? We educated 5,000 users on zero-knowledge proofs, and the lesson stuck: consensus is often a lagging indicator, not a leading one. Today, Lazard’s 91% consensus on data moats is similarly lagging. It does not account for the fact that on-chain data—the most verifiable, auditable, and composable data in existence—is by definition not proprietary. Every transaction on Ethereum is public. Every DeFi protocol’s liquidity pool is visible. The traditional ‘data moat’ is inverted in crypto: the value is not in hoarding data, but in making it transparent and trustless.
This is where the contrarian angle emerges. The survey’s biggest blind spot is its assumption that AI-native software companies will follow the same playbook as Web2 incumbents. But crypto-native AI agents do not need proprietary data. They thrive on synthetic data, on-chain verifiable actions, and decentralized governance. A project like Bittensor, which creates a decentralized neural network, or the emerging AI-agent economies on Base—these are not building moats; they are building commons. The network effect here is not about locking users in, but about aligning incentives through tokens. The 4% of investors who said they have not changed their investment approach? They might be the ones who understand that the next generation of software will not be a better Salesforce, but a protocol that no single entity controls.
Based on my experience facilitating the 2026 AI-Agent Economic Symbiosis Framework, I can tell you that the real due diligence is not about ‘data moats’ but about ethical trust. The Lazard survey treats AI as a threat to existing business models. It does not ask: what happens when an AI agent autonomously executes a smart contract that has no CEO to fire? The answer is that the software itself becomes the trust layer. In crypto, we already have the tools: on-chain governance, verifiable compute, and immutable audit trails. These are the silent moats that the traditional analysts are missing. The 91% consensus is a crowded trade. The real alpha is in the silence of the audit—finding projects that combine decentralized data with zero-knowledge proofs to create AI models that are both powerful and private.
Let me give you a concrete example. Consider a startup building an AI-powered CRM for supply chain finance. In the traditional view, its moat is the proprietary data from shipping records. But in a crypto-native view, that data could be tokenized, shared among consortium members, and used to train a federated model that no single party owns. The moat becomes the governance mechanism for that data, not the data itself. This is the insight that the Lazard survey entirely misses because it is looking at the wrong layer. The 96% of investors who moved money to ‘other opportunities’ are likely fleeing to AI infrastructure plays like GPU compute. But even there, the crypto twist is emerging: decentralized GPU marketplaces (Render, Akash) are commoditizing compute, making the traditional ‘AI infrastructure moat’ as fragile as the software one.
How do we quantify this? In my 2024 Bitcoin ETF narrative re-framing, I argued that ETFs were not just financial instruments but educational tools. Similarly, the Lazard survey is not just a signal; it is a teaching moment. The consensus is that software companies without data moats will die. But the real battle is between centralized data silos and decentralized, verifiable data networks. The crypto projects that survive will be those that treat AI agents as first-class citizens on-chain, with built-in human-in-the-loop consensus mechanisms. I have seen this work: the 2026 framework I co-designed with 50 AI developers and sociologists resulted in a protocol that prioritized community safety over pure efficiency, securing $50M in institutional funding. The takeaway? The market is currently pricing in a narrow version of AI disruption. It is ignoring the fact that blockchain can solve the very problems that make AI scary: data provenance, auditability, and democratic governance.
The next narrative is not about ‘AI vs. SaaS.’ It is about ‘AI + Crypto as the new infrastructure for trust.’ The Lazard survey tells us where the crowd is standing. Alpha hides in the silence of the audit: the projects that are building the rails for AI agents to transact, verify, and govern themselves. That is the investment thesis I am building today. Read the docs. Question the whisper.
