Circle Built a Chain to Get Paid for Money It Already Moves

USDC settled roughly $32 trillion in adjusted transfer volume in 2026 through August. Circle collected fees on almost none of it. Every one of those transfers paid someone else: a validator on Ethereum, a sequencer on Base, a leader on Solana.

Circle earns on the float. The movement has been free advertising for other people's chains.

Arc, Circle's own Layer 1, opens to the public on September 16, 2026. It is an attempt to own the meter as well as the money.

Reserve income pays the bills, and that is the problem

For the quarter ended June 30, 2026, reserve income delivered $667,7m of Circle's $701,3m in revenue and reserve income. That is 95,2% of the business. Transaction revenue came to $5,3m.

The cost side explains the urgency. Distribution and transaction costs reached $410,4m in the same quarter, of which $324,6m went to Coinbase. Circle pays for distribution because it does not own the surfaces where USDC sits.

Then there is rate risk. Circle modeled a 100 bp move in reserve yield as worth about $737m of reserve income over the following 12 months. A company whose revenue tracks the Fed needs a second line that does not.

Arc is that line. Jeremy Allaire has publicly called it a bigger opportunity than USDC itself.

Gas priced in dollars is a product decision, not a crypto one

Arc is an EVM Layer 1. Consensus runs on Malachite, a Tendermint-style BFT engine Circle acquired from Informal Systems. Execution runs on Reth. Blocks close in about half a second with deterministic finality.

The choice that matters is the fee asset. USDC is the gas token. The payment and the fee to make it come out of the same balance.

For a treasury desk that removes a category of work: no gas-token inventory on every chain, no hedging an asset you hold only to pay fees, no explaining to an auditor why the cost of a transfer doubled overnight.

Add opt-in privacy, protocol-level compliance hooks, CCTP for canonical USDC across chains, and an agent stack for machine payments. The buyer is obvious. Arc is sold to regulated firms, not to traders.

The validator list is the moat

Twelve named institutions secure the genesis network: Circle, BlackRock, DTCC, Galaxy, Global Payments, ICE, Mastercard, MoneyGram, SBI Group, Standard Chartered, Sumitomo Corporation and Visa. None are anonymous.

That is deliberate. Regulated firms cannot route client assets through counterparties they cannot name. More than 100 ecosystem and institutional builders have been running on private mainnet, and Circle plans to support tokenization of DTC-custodied assets on Arc from the second half of 2027.

Circle also states plainly that no regulator has reviewed or approved Arc. The validator cohort is a trust design, not a license.

Two assets, and only one of them earns today

This is where the model deserves a closer read.

USDC pays fees. ARC coordinates the network. The May 2026 whitepaper sets initial supply at 10b tokens, split 60% to ecosystem development, 25% to Circle, 15% to long-term reserves. Circle raised $222m in a presale at roughly $3b fully diluted valuation, led by a16z crypto with Apollo, BlackRock and ARK Invest among the buyers.

The intended loop: fees arrive in USDC, the protocol converts part of them into ARC, routes some to validator and staker rewards, and burns the rest. Issuance starts near 2–3% a year and declines on a governance-adjustable schedule.

Now name the layers, because they are different amounts of money.

Users pay fees in USDC. Under proof of authority, the model Arc launches with, those fees accrue to the block proposer. That is the institution running the node. The holder layer is zero.

Conversion, burn and staking rewards all arrive with proof of stake, in a phase two with no announced date. The burn-to-reward ratio is not specified. Neither is the inflation curve past the opening range. Both are left to governance.

So the token prices a mechanism that does not exist yet, on a network whose validators do not need it to participate. That is not unique to Arc. It is the standard shape of an institutional chain launch, and it should be read as what it is rather than as cash flow.

What Arc changes for USDC

Three things, in the order they show up.

Fee capture comes first. Activity denominated in USDC on Arc produces revenue that has nothing to do with Treasury yields. Small at the start, different in kind from everything Circle earns now.

Distribution economics come second. Circle pays exchanges to keep USDC prominent, and a higher on-platform mix is what pushed revenue less distribution costs to a record 41,4% margin in Q1 2026. Traffic on Circle's own network carries no such toll.

Liquidity shape comes third. CCTP burns and mints canonical USDC rather than issuing bridge IOUs, so Arc imports the form of the dollar other chains already accept.

One caution decides whether any of this is real. Volume that migrates from Ethereum or Base to Arc moves Circle's own numbers between pockets. The line worth tracking is third-party application activity in USDC on Arc, not headline float or day-one TVL.

The competition is not technical

Arc arrives late to a category it helped name. Plasma launched in September 2025 with USDT gas and protocol-sponsored zero-fee transfers. Tempo, from Stripe and Paradigm, went live in March 2026 with no native token at all and gas payable in any stablecoin through an enshrined AMM. Stable uses USDT itself as gas. Tron still settles more stablecoin transactions than any of them.

Arc does not win on consensus or block time. It wins, if it wins, because Circle issues the asset and sells the rail to the institutions that already clear its trades.

The counterargument is fragmentation. If every issuer runs its own network, institutions end up with more settlement venues to reconcile, which is the opposite of what they asked for.

What founders should take from this

Four design lessons transfer directly, and none of them require a $3b valuation.

Separate the unit of account from the coordination asset. Users want a stable price per transaction. The network wants a scarce asset to secure and govern. Forcing one token to do both turns your gas market into a tax on product design.

Do not ship a token ahead of the mechanism that pays it. If rewards and burns begin in phase two, that belongs in the same sentence as the supply table, not three pages later.

A burn is only as strong as its ratio. "A portion is burned," with the portion left to a future vote, is a promise rather than a mechanism.

Check whether your volume is pull or migration. Moving your own liquidity onto your own chain proves distribution. It does not prove demand.

The scorecard after September 16

Watch four numbers instead of the launch-day chart. Transaction and service revenue in Circle's next disclosures. The share of Arc activity generated by third-party applications rather than by Circle's own flows. Whether validators outside the presale list hold meaningful stake once proof of stake arrives. And whether the burn-to-reward ratio gets published rather than deferred again.

Arc does not need to replace reserve income to matter. It needs to prove Circle can charge for activity that USDC already enables somewhere else.

Sources. Circle pressroom (public testnet launch, October 28, 2025; founding validator cohort and integrations, August 5, 2026), Arc whitepaper coverage (May 2026), Circle Q1 and Q2 2026 results and disclosures, Coin Metrics transfer-volume measurement (August 2026), CoinDesk, CryptoSlate, Figment, Everstake.