Circle processed $32 trillion in USDC transfers, yet 95% of its revenue relies entirely on interest rates

Adjusted USDC transfer volume reached $32 trillion in 2026 through Coin Metrics' August measurement, with each dollar of supply turning over 741 times at an annualized rate. Those figures signal reach and settlement intensity. Circle's second-quarter revenue, however, remained dominated by yield on the assets backing USDC.

For the three months ended June 30, reserve income supplied $667.7 million of Circle's $701.3 million in total revenue and reserve income, or 95.2%. Transaction revenue was $5.3 million. Volume is a usage signal; balances and yields still determine most of Circle's revenue base.

Arc, Circle's blockchain infrastructure scheduled for a Sept. 16 public mainnet launch, is the company's clearest attempt to build a direct fee surface around some of that activity. Its test is whether traffic can become retained recurring revenue.

How USDC transfer volume is generated

Coin Metrics described the $32 trillion as adjusted USDC transfer volume in 2026 through its August analysis, not consumer payments, unique economic settlement or a full-year total. Its annualized velocity estimate measures how often supply moves relative to its size.

The composition of that movement matters. Coin Metrics' bottom-up analysis examined raw transfer volume through tagged lending contracts, decentralized exchange pools and known exchange wallets. On Base, 69% of USDC volume involved DEX liquidity provision and 23% involved flash loans. On Ethereum, flash loans accounted for 65%.

Those categories represent real crypto-market demand. Liquidity rebalancing, collateral movement and arbitrage make markets function, while generating enormous gross transfers without a matching increase in net capital moved, purchases made or fees collected by Circle. A liquidity position that is repeatedly rebalanced may move the same dollars many times, while a flash loan is borrowed and repaid within one transaction.

Coin Metrics also treated its tagged shares as lower-bound estimates. About 8% of Base volume and 33% of Ethereum volume remained outside the identified categories, and that residual could include payments, bridging, treasury activity and other settlement. It cannot safely be relabeled as commercial payments.

The result is a better reading of USDC's velocity: the stablecoin is deeply embedded in crypto's financial plumbing, but headline transfer volume is not a revenue ledger.

Circle's own Q2 activity metric reinforces the gap. The company said USDC onchain transaction volume rose 151% year over year to $14.8 trillion, while period-end circulation increased 19% to $73.3 billion. Total revenue and reserve income rose 6.6% to $701.3 million.

The Coin Metrics and Circle volume measures are not interchangeable. Coin Metrics reports adjusted transfer volume; Circle's filing defines its figure as native and canonically bridged USDC processed across supported chains other than Solana. Both show scale, but neither implies that Circle charges a toll on every transfer.

The SEC filing shows where Circle's growth translated into revenue. The company attributed about $147.4 million of year-over-year reserve-income improvement to a 25.2% increase in average daily USDC circulation. A 66-basis-point decline in average yields offset about $113.9 million of that gain, leaving reserve income up roughly $33.5 million.

That bridge explains why circulation matters more to current revenue than velocity. More USDC outstanding expands the reserve base that earns interest. The same USDC moving hundreds of times does not automatically create hundreds of revenue events for Circle.

The filing also separates reserve income from retained economics. Circle recorded $410.4 million of distribution and transaction costs in the quarter, including $324.6 million of Coinbase-related distribution costs, and $412.5 million when other costs were included. The Coin Metrics transfer total does not directly drive those costs. Their scale still shows why gross reserve income cannot be read as operating margin.

Rates remain the larger near-term sensitivity. Holding circulation and reserve allocation constant, Circle modeled a 100-basis-point move from June's average yield as changing reserve income by about $737 million and distribution and transaction costs by about $360 million over the following 12 months. That was a hypothetical sensitivity, not guidance, but its scale shows how far Circle remains from a revenue mix driven primarily by transaction fees.

Arc's monetization test

Arc is Circle's clearest attempt to bring more economic activity onto infrastructure it helps operate. Circle's latest Aug. 5 statements placed Arc in private mainnet with more than 100 builders and scheduled public mainnet for Sept. 16. As of those statements, the public launch had not occurred.

Arc's design gives the experiment a direct connection to USDC usage. Its gas and fee system denominates transaction fees in USDC, creating a dollar-denominated charge each time the network processes activity. That is a more visible fee surface than ordinary USDC transfers across third-party chains.

Recurring Circle revenue from Arc remains unproven. Arc's documentation does not establish how much gas-fee revenue Circle will retain, and Circle's Aug. 5 list of builders and prospective integrations does not prove transaction volume, commercial demand or margin after the scheduled public launch. The important post-launch signal will be whether activity appears in Circle's transaction and service revenue rather than only in network statistics.

The ARC Token presale is a separate economic event. Circle agreed to sell 807.5 million tokens for about $242.2 million, but the Q2 filing recorded the proceeds as deferred revenue, not recognized quarterly revenue. The prospective token is tied to a possible later transition from proof of authority to proof of stake or delegated proof of stake, whose timing and terms remain conditional.

Treating the presale as proof that Arc has already diversified Circle's operating revenue would therefore confuse financing and future performance obligations with recurring network economics.

Arc does not need to displace reserve income immediately to matter. It needs to show that Circle can capture a repeatable share of activity that USDC already enables elsewhere.

That creates a concrete scorecard after Sept. 16: whether public mainnet launches as scheduled, what kinds of transactions dominate, whether applications generate sustained activity outside mechanical liquidity loops, and whether Circle begins reporting material growth in transaction or subscription and service revenue. Retained economics will matter as much as gross fees.

USDC's $32 trillion year-to-date transfer total demonstrates reach and liquidity. Coin Metrics' decomposition shows why that number should not be mistaken for payments revenue. Until Arc or another Circle product turns more of that movement into durable fees, Circle's business will continue to be governed primarily by how many USDC are outstanding, what their reserves yield and how much of that income remains after distribution costs.