TBW - The float and the toll: why stablecoin reserve economics are being repriced
A regulated payment-stablecoin issuer is a narrow bank that cannot pay for its liabilities. GENIUS and MiCA require one-for-one cash and short government paper, and bar the issuer from rebating that coupon to holders. The P&L is the spread between the front-end rate and the cut distributors take for placing the token. Brand and compliance are the cost of keeping that right.
A minority claim on a large float
Circle’s numbers are unusually clean. Reserve income was $2.637bn of $2.747bn in 2025, some 96 percent, and $668mn of $701mn in Q2 2026. It was earned on an average $76.5bn of USDC at a 3.48 percent reserve return rate. After distribution costs, Circle retained $1.083bn in 2025, a 39 percent share, and $273.88mn in Q2, or 41 percent. Note that this is retained reserve income before operating expense, not profit. The float is large. The issuer’s claim on it is already a minority interest.
Holders do not choose the coin. Distributors place it. Coinbase takes 100 percent of reserve income on balances held on its platform, and 50 percent of the residual elsewhere. At quarter-end, it held 27 percent of the $73.3bn in circulation. Total distribution cost for Circle came at $412mn in Q2 alone (margin applied to Q2 total revenue), and Coinbase-related distribution was the dominant part of roughly $1.55bn in total distribution and transaction costs across 2025. The contract rolled on 18 August 2026, terms unchanged, through 2029.
The share is also rising. Coinbase held about 5 percent of USDC in 2022 and 20 percent in 2024. GENIUS banned issuer-paid yield, not platform rewards. Coinbase can still fund a loyalty rate from its share of the float. Banks want that gap closed in CLARITY and in the OCC’s anti-evasion proposal. Close it, and platforms demand larger cheques. Leave it open, and they pull coins on-platform and keep the whole coupon. Either way, the issuer remains the residual claimant.
A distribution auction, not a product cycle
With the coupon capped by law and set by SOFR, the only free variable is the split. That is how to read the consortia. None is chiefly a product. Each is a bid to reset the distribution terms in favour of a different holder of the flow.
OpenUSD, backed by more than 140 firms including Visa, Mastercard, Stripe, BlackRock and Coinbase, returns reserve earnings to partners net of a management fee. Stripe wants it as a default. Coinbase is both Circle’s largest USDC distributor and an OpenUSD partner. In parallel, on 1 September, a group of 21 banks committed to launch a US dollar-denominated stablecoin in the first half of 2027. The OpenUSD announcement hit CRCL harder than this week’s bank news because it attacks partner economics now.
Tether still sets the market at about $184bn, against USDC's $73bn to $75bn. Consortium coins will press USDC first, not offshore USDT collateral. And rates move earnings before any consortium moves the split. The direction now favors the issuer. CME FedWatch puts a September hike near two-thirds, and some desks see a further move by December. A higher front end directly lifts reserve income. Circle's 10-K sizes a 100bp shift at $756mn in reserve income against $369mn in distribution costs, so the issuer keeps roughly $387mn of it. Reserve economics are not being repriced. The spread mechanism is unchanged. What is new is a distributor bloc bidding to place its own coin.
The Big Whale’s take
The distribution cut and the reserve yield move together. Earn more on the float, pay more to place the coin, because distribution is the scarce input, not the reserve. If the issuer defends its retained share, a higher rate is net positive in absolute terms even as the distribution cheque grows. That is why CRCL recovered the bank-news dip within days and trades near $100 again, having doubled off August's $60 low. The scare was competitive, not economic.
The more useful lens for allocators is what Circle has become. With 96 percent of revenue drawn from reserve income, the equity is an indirect and geared claim on front-end Treasury yield. It offers T-bill exposure with equity convexity, and a rising path makes that claim more valuable, not less. The durable risk is not rates. It is that as banks, card networks and acquirers each field a coin, the native issuer's share of any given flow narrows toward the cost of running reserves.
Watch the distribution share first, not SOFR. Then watch whether the platform-rewards channel is closed in CLARITY or by the OCC. Finally, watch whether OpenUSD or the still unnamed bank consortium moves balances rather than headlines.