We built the utopia, then audited the ruins. The Lazard survey just told us something we already felt: code is dead. Long live data.
For those who missed it, Lazard’s latest private equity secondary market survey dropped a bombshell: 91% of institutional investors now see “proprietary data + network effects” as the only sustainable moat in software. Only 4% haven’t changed their investment approach. This isn’t a gradual shift—it’s a paradigm rupture.
Context: The survey, conducted among PE secondary market LPs and GPs, asks whether AI will disrupt traditional software. The answer is not “maybe” but “yes, and we’re repricing everything.” The 91% consensus is abnormally high—usually you see 50-70% agreement on any strategic question. This level of unanimity signals that the market has already priced in a fundamental revaluation of software assets. The old valuation framework—ARR multiples, NDR, gross margins—is being replaced by an AI exposure discount and a data moat premium.
Core: Let’s bring this home to crypto. Because the same logic applies to our stack. The smart contract is the new code—but the real value isn’t the bytecode; it’s the data and the network.
Take Uniswap. The constant product formula is public. Anyone can fork it. But the liquidity network—the billions in TVL, the order flow, the user behavior data—that’s the moat. In my own work auditing DeFi protocols, I saw this firsthand: projects with unique on-chain data (like a lending protocol’s credit history) had stickier users than those with just a clever algorithm. The code is a commodity; the data is the alpha.
Similarly, L2s like Arbitrum and Optimism aren’t winning because of their rollup architecture—they’re winning because of their network effects: developers, dapps, and users. That’s a data network. And when blob data saturates post-Dencun (my prediction: within two years), gas fees will double, but the real cost will be in data availability. The L2s that own the most user data—transaction history, MEV patterns—will survive. The rest will be commoditized.
Even Bitcoin’s Lightning Network, which I’ve called half-dead for seven years, suffers from a lack of network effects. Routing failure rates are high, channel management is a nightmare. It’s a protocol with no data moat—everyone can see the same channels. No wonder it’s niche.
Contrarian: But here’s the twist. The 91% consensus might be wrong. Because in crypto, “proprietary data” is an illusion. The blockchain is public. Anyone can query on-chain data. The real moat isn’t the data itself—it’s the ability to synthesize and act on that data in real time. And that’s where AI comes in.
AI agents can now analyze on-chain data, execute trades, and even govern protocols. The “data moat” becomes a “data processing moat.” And if AI can process public data faster than any human, then the only advantage left is the network effect—the community, the trust, the relationships. Code is not law; it is a negotiation. And the negotiator is the network.
Furthermore, the survey’s implication that “AI threatens software” assumes that AI is an external force. In crypto, AI is native. We’re already seeing AI agents on Autonolas, AI-powered oracles, and AI-optimized MEV. The line between “software” and “AI” is blurring. The real threat is not AI replacing software—it’s AI replacing intermediaries. And in a decentralized world, that’s a feature, not a bug.
Takeaway: The Lazard survey is a wake-up call for crypto builders. The era of “code as moat” is over. The next cycle will be about data sovereignty, network density, and AI-native applications. The winners will be those who own the relationships—not just the smart contracts.
Every bug is a lesson in decentralization. And this lesson is clear: we coded the dream, but the market wrote the code. The market is now telling us that data and networks are the new code. Build accordingly.
Decentralization is a verb, not a noun. It’s time to act.
Trust no one, verify everything, build always.