Circle’s Chandhok outlines Arc’s growth strategy for AI agents

Circle just launched Arc, a Layer-1 blockchain built from the ground up for stablecoins and AI agents. And in a move that might surprise the usual zero-sum tribalists in crypto, the company’s Chief Product and Technology Officer says the goal isn’t to poach users from Ethereum.

Nikhil Chandhok put it plainly during the Arc mainnet launch on September 16: “If all we did was we took activity from Ethereum and moved it to ARC we didn’t really grow the market.”

A blockchain designed for machines, not just humans

Arc’s core thesis is that the next wave of onchain economic activity won’t come from humans swapping tokens on DEXes. It’ll come from AI agents conducting transactions at machine speed, verifying their own work, and building reputations onchain.

To support that vision, Circle built what it calls the Agent Stack, which rolled out in May 2026 after a public testnet went live in October 2025. The stack gives AI agents access to primitives like provenance proofs, reputation systems, and nanopayments.

On the technical side, Arc uses USDC as its native gas token rather than minting a separate fee currency. That’s a deliberate design choice: it means every transaction on the network is denominated in a dollar-pegged stablecoin, removing the volatility friction that makes gas fees on networks like Ethereum unpredictable. The chain also features sub-second deterministic finality through what Circle calls its Malachite consensus engine, meaning transactions settle in under a second with no probabilistic ambiguity about whether they’re confirmed.

The money behind the machine

Arc didn’t launch on goodwill alone. The ARC network token has a fixed initial supply of 10 billion, and Circle completed a $222 million private presale of 740 million tokens at $0.30 each. That prices the project at roughly $3 billion post-sale.

The validator set reads like a Davos attendee list. BlackRock, Visa, and Mastercard are among the institutional validators supporting the network at launch.

Why not just build on Ethereum?

Chandhok’s framing is worth unpacking. The argument is that Ethereum already serves its existing user base well enough. What it doesn’t serve, according to Chandhok, is the emerging category of AI agents that need to transact autonomously, build reputations, and settle payments in stable value at speeds that match software, not human reaction times.

The emphasis on not developing applications directly is also telling. Chandhok positioned Circle as an infrastructure provider, building the rails and letting others create the trains. The Agent Stack provides the primitives. Third-party developers and enterprises are expected to build the actual AI agent applications on top.

What makes Arc’s position more defensible is the USDC integration. Circle controls the world’s second-largest stablecoin by market capitalization, and making USDC the native gas asset creates a built-in demand loop. Every AI agent transaction on Arc generates USDC gas fees, which reinforces USDC’s utility, which potentially drives more USDC adoption, which makes Arc more attractive as a platform.

The $222 million presale at a $3 billion valuation also sets a high bar for what Arc needs to deliver. Early token buyers at $0.30 are going to want to see real network activity, not just validator logos.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.