From the chaos of 2017, we forged a compass. But the compass I built then—pointing to decentralization, code sovereignty, and trustless execution—is now trembling under a new magnetic field: artificial intelligence. Last week, Lazard’s survey on private equity secondaries landed in my inbox, and I felt the same chill I did when I first read the DAO hack post-mortem. 96% of investors have already changed how they evaluate software companies. 91% now believe proprietary data and network effects are the only true moats. And here’s the kicker: capital is flowing away from software entirely, fleeing the uncertainty of AI disruption.
I am Andrew Martinez, a cryptography PhD who spent 14 years in the Web3 trenches. I’ve audited DeFi protocols, built community Trust Scores, and watched rollups fight over liquidity. But this survey isn’t about blockchain—it’s about the software industry that underpins half of our crypto infrastructure. What happens when the same AI-driven valuation shift hits our own stack? The data signals a coming reckoning for crypto projects that rely on software moats that are now being priced as liabilities.
Context: The Lazard Signal
Lazard’s Q2 2025 survey targeted institutional investors in the private equity secondary market—the same capital pools that eventually flow into token sales, node sales, and protocol treasuries. The headline: 96% of respondents have already altered their investment approach to software due to AI. 4% remain unchanged. That’s not a whisper; it’s a stampede. And the money is moving to “other opportunities,” meaning they see AI as a net destroyer of value for traditional software companies.
For crypto, this is a canary. Our entire ecosystem—from Ethereum clients to DeFi front-ends to L2 sequencers—runs on software. If the same capital that funds our infrastructure is retreating from software equities, where does that leave the tokenized versions of those assets? The survey’s 91% consensus on “data moats” is particularly haunting: it suggests that without a defensible dataset, a crypto project’s software layer is just a commodity.
Core: The Three-Pronged Attack on Crypto’s Software Soul
Let me drill into the technical and value implications. Based on my audit experience, I see three distinct ways AI will reshape crypto’s software foundation, and the Lazard data provides the market lens.
1. Smart Contract Development Becomes Commoditized
The survey’s implicit assumption—that AI can replicate software functionality at near-zero marginal cost—hits Solidity and Rust developers directly. AI code generation tools (GitHub Copilot, Codeium) already produce audit-worthy contracts. In 2026, I tested a GPT-4o agent that wrote a Uniswap v3 clone in 12 minutes. The moat of “we have the best dev team” is evaporating. The Lazard 91% consensus tells us investors will soon demand a different moat: proprietary data or unique network effects. For crypto, that means the protocol’s value will shift from code to the data it generates and the community it retains.
2. DeFi’s “Feature-Based” Valuation Is Dead
Traditional DeFi protocols are valued on TVL, fee generation, and code security. But the Lazard survey reveals that investors now see “functionality” as a trap. If AI can replicate a lending protocol’s core logic overnight, the only defensible advantage is the user’s trust and the data accumulated—loan histories, reputation scores, governance patterns. Trust is not a metric; it is a memory we share. AI can’t fake a decade of on-chain reputation. Projects that fail to build this memory will see their valuation halved, just as the survey predicts for software companies without data moats.
3. L2 Rollup Economics Under Siege
Post-Dencun, blob data will be saturated within two years, and rollup gas fees will double. But the Lazard survey adds a new layer: the sequencer software itself is a commodity. If AI can optimize sequencer logic or even replace it with a model, the economic moat of a rollup collapses. The real value shifts to the data layer—the blob content, the user transactions, the MEV patterns. Those who control the data, not the software, will survive. I’ve been tracking this since 2024, and the Lazard data confirms that the market is already pricing this shift.
Contrarian: The Blind Spot of “Data Moats”
Here’s the twist. The 91% consensus is itself a danger signal. When everyone agrees that data is the moat, that moat is already priced in. The real alpha lies in what the survey ignores. For crypto, the overlooked moat is human agency. In a world where AI can generate synthetic data, the uniqueness of on-chain, human-verified actions becomes priceless. From the chaos of 2017, we forged a compass—and that compass was resilience through community. AI cannot replicate the emotional and social capital of a DAO that has weathered three bear markets.
Furthermore, the survey’s capital flight from software might be a misread. In crypto, software is not just a product; it’s a settlement layer. An AI-copied smart contract still runs on Ethereum, and the network effect of the base layer is not replicable. The Lazard investors are looking at software as a standalone business. Crypto software is embedded in a trust network. This is the contrarian opportunity: buy the data-rich, community-heavy protocols that the AI panic is selling off.
Takeaway: The New Compass
In 2026, the convergence of AI and crypto is not a hype cycle; it is a filter. The Lazard survey is a mirror for our own industry. Projects that treat software as a commodity and data as a fortress will survive. Those that rely on code alone will be priced for extinction. The question is not whether AI will disrupt crypto—it already has. The question is whether we have the courage to rebuild our moats on the one thing AI cannot forge: a shared memory of trust. Trust is not a metric; it is a memory we share. And in the age of AI, that memory is the only asset that matters.