AI Agents and the Next Wave of Crypto Demand | The Breakdown

What It Covers

Haseeb Qureshi, managing partner at Dragonfly Capital, discusses token valuation frameworks, the role of disclosure over revenue in crypto market structure, and how AI agents transacting on-chain over the next two to five years represent a broad demand wave likely to benefit multiple blockchains simultaneously.

Key Questions Answered

• •Token disclosure over revenue: The core market structure problem in crypto is not insufficient cash flows but inadequate disclosure. Fully on-chain, open-source protocols self-disclose via tools like Dune Analytics, eliminating information asymmetry. Investors should evaluate whether a protocol's metrics are fully publicly verifiable before treating revenue multiples as the primary valuation lens.
• •AI agent adoption timeline: Expect high-risk-appetite users to deploy AI agents as on-chain economic actors within two years, with mainstream adoption arriving around five years out. Agents will optimize block space usage across Ethereum, Solana, and Base based on fees and liquidity rather than chain loyalty, spreading demand broadly rather than concentrating it.
• •Productivity gains distribute widely: Historical precedent from electricity and automobiles shows large productivity increases share gains broadly rather than concentrating them in one winner. Investors should resist the assumption that AI-driven on-chain activity accrues only to a single chain or corporate infrastructure layer, and instead position across liquid, fast-settling networks.
• •Lower software costs benefit startups, not incumbents: Cheaper AI coding tools reduce capital leakage in startups, improving efficiency rather than eliminating competitive advantage. Legacy companies like PayPal will not aggressively deploy these tools. The disruption will come from founders running multiple AI agents simultaneously, making early-stage crypto and software startups more capital-efficient and faster-moving than before.
• •AI adoption is far earlier than perceived: Only 14% of the global population has used any AI product, and only 1% of those users have paid for a subscription. Most paying users are on the $20 monthly tier, not advanced plans. Builders and investors operating at the frontier are years ahead of aggregate economic data, which currently shows no measurable AI impact on GDP or labor markets.