How AI and crypto are converging: eight use cases for allocators

Additional analysis by Karim AbdelMawla, Matt Mena, and Maximiliaan Michielsen

‍

AI has owned the market's attention this year and will keep it for a while. What gets lost in that mindshare is the assumption that crypto plays no role in the AI transformation. The case runs the other way: both are horizontal, general-purpose technologies, and each solves the other's biggest weakness. AI needs money, memory, and compute it can use effectively. Crypto needs something to do the hard part for users.

AI's contribution to global GDP is estimated to be $15.7 trillion by 2030 (PwC, September 2017), more than China and India produce today, and AI agents are projected to mediate $3 to $5 trillion of consumer commerce by then (McKinsey, October 2025). Crypto sits near $2.4 trillion and looks a lot like the internet in 2003: stablecoins have product-market fit, trading and decentralized finance (DeFi) have traction, mining has built industrial-scale power, but it is still too clunky for anyone who will never manage a seed phrase. As broadband and the smartphone did for the internet, agents will simplify the crypto experience by handling the finances, commerce, and security for users. The efficiency gains compound on both sides. Agents strip the time, errors, and cost out of using crypto: over $1 billion of ether (ETH) has been lost to plain user error that automation removes. Crypto, in turn, gives agents rails that settle in seconds for fractions of a cent, at any hour, in any size. If even a small slice of the AI economy settles, stores, or computes on crypto rails, the asset class benefits from this tailwind.