Circle CEO Jeremy Allaire sees AI agents as future participants in the economy, not just tools people use. In his view, that shift calls for financial infrastructure designed for software to transact. Circle’s Arc blockchain reflects that bet, combining agent-focused activity with features intended for financial institutions.
2
3
Why build a blockchain for AI agents?
Allaire’s argument starts with a change in who may use digital financial services. If software agents increasingly act on people’s behalf, they will need ways to make and receive payments as part of their work. Circle presents Arc as a network built for this kind of “agentic economic activity,” rather than a conventional blockchain adapted for agents later.
2
10
Arc’s mainnet went live on September 16, 2026. Its fees are paid in USDC, which Circle says removes the need for users to hold a volatile native asset just to pay network fees.
5
10 Circle’s Agent Stack adds policy-controlled wallets, nanopayments and a marketplace—tools intended to let developers and agents participate in digital commerce.
7
These products support Circle’s machine-economy thesis, but they do not establish that AI agents already dominate payments or will necessarily do so. That remains a prediction about how the internet and commerce may develop.
2
10
Arc’s institutional design: open access, known operators
Allaire has described Arc as a middle ground between open crypto networks and banking standards. According to his comments, anyone can build on the network, while its validators—the organizations that operate it—are known rather than anonymous.
1 Circle says Arc was designed for financial markets and real-time money movement as well as agent activity.
10
That distinction is central to the pitch: access to build on the network can be open even when the operators are selected and identifiable. Reports on Arc’s launch list 11 founding validators, including BlackRock, DTCC, Mastercard, Visa and Standard Chartered.
5
31 The sources support the validator roster, but do not establish every technical detail raised about consensus or a future move to proof of stake.
Why Circle chose to build Arc
The conversation also addressed why Circle would create a Layer 1 blockchain rather than rely solely on existing networks. A summary of the discussion says Circle concluded that existing networks did not meet its ambitions, prompting it to build Arc from scratch.
3 The company’s wider aim is to connect programmable money and financial infrastructure with new forms of economic activity, including transactions initiated by agents.
6
10
What the available evidence does—and doesn’t—show
The sources support the broad case: Circle is positioning Arc, USDC and its Agent Stack for a future in which software agents carry out more economic activity, while using known validators to address institutional expectations.
1
2
7
10
They do not substantiate every specific claim sometimes attached to that vision. The available source excerpts do not verify the cited $150 billion quarterly minting-and-redemption figure, a network reach of more than 20 blockchains through CCTP and Gateway, or detailed plans for agent-to-agent lending and dispute arbitration. Those ideas should be treated as unconfirmed here, not as established Arc capabilities.
The clearest takeaway is that Arc is Circle’s infrastructure bet on a possible machine economy. Its payment tools and validator model explain how Circle hopes to serve that future; whether AI agents become dominant users remains an open question.