How AI Agents Will Transform the Financial System with Circle Co-Founder and CEO Jeremy Allaire

Summary

• Circle’s founding bet was not to replace the dollar, but to turn it into safer, full-reserve internet money. Jeremy Allaire expects major reserve currencies to persist for “30, 40, 50 years,” while stablecoins remove fractional-reserve leverage from the payment layer. USDC’s backing is concentrated in cash, overnight Treasury repo and short-duration T-bills averaging roughly 13 days; he says the GENIUS Act has now codified this “narrow money” model.
• USDC’s case rests on being a general-purpose settlement protocol, not a single payments product. The same token handles a $0.25 game object, an AI agent buying $0.20-$1 of intelligence and electronic trading firms settling multi-hundred-million-dollar transactions. Allaire’s analogy: email infrastructure does not care whether the payload is breakfast chatter or “a CIA dossier.”
• AI agents could supply the transaction volume and demand that crypto infrastructure has been building toward. Agents will perform white-collar work, contract with one another and purchase specialized output in $0.05-$0.10 increments, potentially creating billions or trillions of transactions. Existing USDC transfers can cost reliably below one cent; Circle says Arc can reach “a millionth of a penny.”
• Arc is designed as institutional financial infrastructure, with known validators and stronger settlement assurances than early chains’ alternative-universe model. Its model uses known financial-infrastructure or potentially large technology companies as validators, while USDC—not a volatile gas token—is the native unit. Circle is targeting deterministic finality within hundreds of milliseconds, built-in privacy and compliance, and assurance that “the bad guys aren’t running your transactions.”
• Tokenization is already moving through the securities stack rather than waiting for a speculative breakthrough. Allaire says Circle was the most active tokenized stock, while Circle operates USYC, which he calls the largest tokenized Treasury product, and EURC, which he calls the largest tokenized euro. Record keepers such as Computershare, DTCC, brokers, Nasdaq and the New York Stock Exchange are all pursuing tokenization, with SEC guidance issued roughly a month before the interview.
• The useful remnants of crypto’s research cycle—zero-knowledge proofs, privacy primitives and verifiable compute—may matter more as AI scales. Allaire calls this blockchain’s “broadband moment”: off-chain computation can be proved on-chain, corporate activity can remain private while compliant, and inference itself might become “productive proof of work.” He explicitly declines to assume Bitcoin must remain dominant: “I don’t know what we’re going to be using in 10 years.”
• Allaire sees double-digit GDP growth in parts of the world during the 2030s as plausible, but distribution is the decisive risk. AI diffusion could create discontinuous productivity gains while making GDP less meaningful if growth becomes “capital capturing more capital at the expense of humans.” His longer-term call is a turbulent renegotiation of the social contract, with a lag between disruption and new institutions combining human and agentic actors.

Deep dive

Not yet available upstream; scheduled sync will retry.