Stablecoin Weekly: The Float Moves to the Venue

Three things landed in the same seven-day window and they rhyme. The Senate Banking Committee advanced the CLARITY Act 15-9 on May 14, the first comprehensive Senate crypto market-structure bill to clear committee, with the Section 404 stablecoin-yield compromise intact and two Democrats crossing over. Two days earlier, on May 13, Coinbase and Circle announced a deal that makes Coinbase the official treasury deployer of USDC on Hyperliquid and routes roughly 90% of the reserve income on Hyperliquid's ~$5.1B of USDC to the venue itself — and that came two days after Circle's first-quarter print showed circulation and on-chain volume exploding while net income fell. And on May 15, Jerome Powell's term as Fed chair ended; Kevin Warsh, an on-record skeptic of private stablecoins, was confirmed 54-45 and takes the seat. The legislative framework that legitimizes the operating model took its biggest step forward, the economics of the model got contested at the reserve float, and monetary leadership turned skeptical. The fight moved from legitimacy to economics.

• CLARITY cleared committee. Senate Banking advanced H.R. 3633 15-9 on May 14 — bipartisan, with Democrats Ruben Gallego and Angela Alsobrooks joining all Republicans. The Tillis-Alsobrooks Section 404 compromise held essentially unchanged: yield "economically or functionally equivalent" to a bank deposit is barred on idle balances, activity-based rewards are permitted, and the SEC, CFTC, and Treasury get 12 months to write joint rules. Senator Warren filed 44 amendments and none passed. The bill now needs 60 votes on the Senate floor and reconciliation with the Senate Agriculture version; Alsobrooks said her committee vote does not guarantee her floor vote.
• The reserve float moved to the venue. Coinbase becomes the official USDC treasury deployer on Hyperliquid and Circle becomes the primary liquidity provider under Hyperliquid's Aligned Quote Asset framework; Hyperliquid captures roughly 90% of the reserve income on its ~$5.1B of USDC (about $135M to $160M a year), Hyperliquid's native USDH sunsets into USDC, and Circle will stake 500,000 HYPE toward validator status. Compass Point estimates the arrangement removes $60M to $80M of combined annual EBITDA from Circle and Coinbase, and warns other venues — Polymarket, Jupiter — may demand the same.
• Circle's Q1 showed the squeeze. Revenue of $694M (up 20% year over year, a slight miss versus ~$715M consensus), reserve income $653M, USDC circulation $77B (up 28%), on-chain volume $21.5T (up 263%), distribution and transaction costs $407M (up 17%), and net income $55M — down 15%. The volume engine is roaring; the margin per dollar of float is being competed down by the rails Circle depends on. Circle also disclosed a $222M Arc token presale at a $3B valuation with BlackRock, Apollo, a16z crypto, ICE, and Standard Chartered Ventures participating.
• A skeptic takes the Fed. Powell became chair pro tempore on May 15 until Warsh is sworn in (May 22). Warsh has said he is "skeptical that a host of private cryptocurrencies are sufficiently strong and reliable proxies for the U.S. dollar" and prefers a wholesale digital dollar. The Eurosystem split went public the same week: Banque de France's Denis Beau broke with ECB President Lagarde to back a private tokenized euro, and the Bank of England dropped its proposed individual holding limits for sterling stablecoins.

Aggregate supply printed another marginal all-time high while the internal mix rotated hard. The headline was not the level — it barely moved — but the reversal underneath it: USDC handed back the prior week's net issuance even as Circle reported 28% annual circulation growth, while Sky's USDS, Paxos's USDG, and a recovering USDe did the work.

• Total stablecoin market cap is roughly $323.3B as of May 16, a fresh all-time high, up modestly on the week — about $0.6B on the level against last week's ~$322.7B, with weekly net inflows reported nearer $1.5B depending on the window measured.
• USDT sits at about $189.7B, roughly flat, with market share ticking up to about 58.7%.
• USDC is at about $77.1B as of May 16, down roughly $950M (about -1.2%) on the week — a reversal of the prior week's dominant issuance, partially recovered by May 21. The contrast with Circle's circulation-growth narrative is a timing-and-venue story, not a contradiction: balances are migrating across chains and venues faster than the net number moves.
• USDe (Ethena) recovered to about $4.44B, up roughly +6.8%, reversing the multi-month contraction that bottomed near $3.8B last week — the cleaner-than-expected outcome the prior issue flagged as an open question.
• USDS (Sky) grew on the order of +11% on the week, with the continued DAI-to-USDS migration driving it; readings vary widely by source (roughly $8.8B on DefiLlama versus higher elsewhere), so treat the direction as solid and the level as contested.
• USDG (Paxos) rose about +9.6%, a second consecutive near-+10% week. USD1 (World Liberty) is up about +2% near $4.5B. RLUSD rebounded toward $1.75B. PYUSD is in the $3.6B to $4.0B range.
• Tokenized U.S. Treasuries sit at about $15.3B, with Ethereum holding roughly half (~$8.0B). The broader tokenized real-world-asset market reads in the low $30Bs once commodities, private credit, and tokenized equities are counted.

• USYC at about $2.98B, category leader, displayed 7-day APY near 3.1%.
• BUIDL at about $2.54B, up roughly +8% on the week — the fastest-growing large product, against the backdrop of BlackRock's reserve-vehicle filings.
• USDY (Ondo) at about $2.14B, now reconciled across dashboards after weeks of methodology divergence; about 3.55%.
• iBENJI (Franklin Templeton's institutional share class) at about $1.48B; the retail BENJI asset at about $827M, still Stellar-dominant with EVM chains and Aptos gaining.
• JTRSY near $995M, WTGXX about $951M, USTB about $832M (now a clean print after the Invesco-handoff distortion), CUMIU about $547M.
• OUSG (Ondo) more than doubled to about $627M from ~$285M — the asset that cleared the first cross-border, cross-bank tokenized-Treasury redemption on May 6 is now seeing the inflows that follow a settlement-rail win.

The base layer did what a base layer does: nothing dramatic. Constituent yields held in the 3.3% to 3.5% band across the largest products, with the front of the curve stable-to-slightly-soft — effective fed funds near 3.62% and the 3-month bill around 3.66%. There are no incentives anywhere in this layer; the entire spread across products is fee and maturity methodology. The interesting motion was in flows, not rates: aggregate AUM kept climbing, OUSG more than doubled, and BUIDL led the large products on growth as reserve-vehicle filings ratified the category.

The pooled-lending premium over the savings-rate floor has now effectively closed. Sky's Savings Rate sUSDS held at about 3.75%, flat, with no new Spark spell after the May 7 one (the next is due around May 21). Aave's Ethereum USDC supply rate fell to about 3.9% from roughly 4.4%, and USDT sits near 2.7% — both now level with or below sUSDS. The de-risking that began with the post-rsETH USDT cap cut continued: Risk Stewards 24921 on May 14 cut supply caps on PYUSD, syrupUSDT, and Base syrupUSDC, with the only stablecoin cap increase going to Treasury-backed PT-USDG. Morpho's curated USDC vaults compressed into a roughly 3.1% to 4.9% base band (Steakhouse USDC near 4.2%, its largest vault at about $466M), and Morpho's total deposits surged to about $11.8B, making it the second-largest lending network as the Aave-to-Morpho rotation continued; Coinbase added Solana-backed lending through Morpho on May 13.

Maple's syrupUSDC is printing roughly 4.8% and syrupUSDT about 4.3%, both below the platform's stated 6% to 10% target band and compressing with the broader credit market — real loan interest, thinner carry. The more notable move was Ethena. USDe supply recovered and sUSDe yield ticked back up to roughly 4.0% to 4.3% from last week's ~3.5%, and the reason matters: Ethena's collateral is now weighted roughly half to on-chain lending and half to liquid stablecoins, with the perpetual-funding share cut to about 11%. sUSDe is becoming less a pure funding-carry instrument and more a blended lending-and-reserve yield — steadier, but still structurally variable rather than fixed. Pendle's fixed-yield demand rolled forward into June and August tenors rather than collapsing at the May 7 maturities, and more than half of Pendle's TVL still rests on the Ethena complex. Where double-digit yields appear in this layer — Falcon USDf at 8% to 12%, Resolv's RLP loss-absorption tranche at 20% to 40%, the reward-boosted 7%-plus Aave prints — they are incentive- or leverage-driven and should be read as such, not as base yield.

Circle's first quarter and the Hyperliquid deal are the same story told twice. Circle reported revenue of $694M, up 20% year over year but a slight miss; reserve income of $653M, dragged by a 66-basis-point decline in the reserve return rate; USDC circulation of $77B, up 28%; and on-chain transaction volume of $21.5T, up 263%. Those are extraordinary distribution and usage numbers. And yet net income was $55M, down 15%, because distribution and transaction costs rose to $407M — the line that captures what Circle pays Coinbase, growing on a greater Coinbase share of USDC circulation. The market took the print as mixed: an EPS beat ($0.21 versus $0.17) and the Arc raise offset by the revenue miss and visible margin compression, leaving the stock up only modestly near $115.

Then Hyperliquid showed where that margin pressure goes next. Under the May 13 arrangement, Coinbase becomes the official treasury deployer for most of the USDC on Hyperliquid and Circle handles minting, redemptions, and cross-chain infrastructure, with USDC installed as Hyperliquid's Aligned Quote Asset. The economics are the headline: Hyperliquid receives roughly 90% of the reserve income generated by those USDC deposits — about $135M to $160M a year on the current ~$5.1B balance — and Hyperliquid's native USDH will be wound down into USDC. Compass Point's Ed Engel and Mike Donovan estimate the deal strips $60M to $80M of combined annual EBITDA from Circle and Coinbase, against roughly $180M of gross profit those two earn on Hyperliquid's USDC today, and flag the obvious follow-on risk: large venues such as Polymarket and Jupiter now have a template to demand the same split.

This is the structural point for institutional underwriting. The thing that makes a stablecoin issuer valuable is the reserve float — the interest earned on the assets backing the tokens. This week demonstrated that the float increasingly accrues not to the issuer but to whoever controls distribution: the exchange that holds the balances (Coinbase's roughly 50% of USDC economics), the trading venue that mandates the quote asset (Hyperliquid's 90% on its balances), and the asset manager that runs the reserve. On that last point, two of the largest managers filed dedicated stablecoin-reserve vehicles within four days — BlackRock's BRSRV and on-chain BSTBL share class, and JPMorgan's JLTXX through Kinexys, an Ethereum tokenized money market fund explicitly built as a GENIUS-compliant reserve asset. Reserve management is now its own SEC-registered product category, with the largest balance sheets competing for issuer mandates. Circle's own answer to disintermediation is to move up the stack: its May 12 Agent Stack puts USDC at the center of machine-to-machine payments — agent wallets, an agent marketplace, and nanopayments down to one-millionth of a dollar — competing directly with Coinbase's x402 and AWS for the agentic settlement layer rather than ceding it. The issuer that wants to keep its float has to own more of the rail.

The 15-9 vote is the most consequential stablecoin-policy event of the year to date, and the detail matters as much as the headline. The committee is 13 Republicans and 11 Democrats; all 13 Republicans plus Democrats Gallego and Alsobrooks carried it. The Section 404 yield framework that the prior issue described survived the markup essentially unchanged — idle-balance yield that mimics a bank deposit is barred, activity-based rewards tied to payments, transfers, trading, staking, liquidity, and loyalty are permitted, and the SEC, CFTC, and Treasury are directed to write joint rules within a year. The banking lobby's preferred "substantially similar" tightening, which would have swept more reward types into the ban, was not adopted. Senator Warren's 44 amendments — on tokenized-asset disclosures, DeFi sanctions, bank crypto activity, and retirement-account limits — all failed, mostly along party lines; a Rounds AI-sandbox amendment passed 15-9 and a McCormick portfolio-margin amendment passed 18-6.

The fight is not over, it has moved venues. The bill needs 60 votes on the Senate floor, which means roughly seven Democrats, and it has to be reconciled with the Senate Agriculture Committee's market-structure text; the unresolved ethics provisions around Trump-family entities are widely seen as gating those 60 votes, and the operative deadline is now framed as the August recess rather than Memorial Day. Alsobrooks was explicit that her committee vote was "a vote to keep working in good faith" and not a floor commitment. The six-trade-association bank coalition, which formally rejected the compromise on May 9, issued a measured May 14 statement crediting Senators Tillis and Alsobrooks while continuing to push for tighter idle-balance language — a softer posture than the pre-markup letter, but the same underlying demand, and the same underlying claim that stablecoin rewards will pull deposits out of community lending.

The legislature spent the week legitimizing private-stablecoin reward economics. The central banks spent it hedging. Powell's term ended and the Board named him chair pro tempore on May 15 "until Kevin M. Warsh is sworn in," consistent with past practice; Warsh was confirmed 54-45, nearly on party lines, and takes office May 22. Warsh is not a stablecoin enthusiast. He has said he is skeptical that private cryptocurrencies are reliable dollar proxies, doubts that bank-like regulation would keep them stable "absent government bailouts," and has signaled a preference for a wholesale digital dollar over a proliferation of private coins. The tension is direct: the framework advancing through Congress permits exactly the activity-linked stablecoin economics the incoming Fed chair is most wary of.

Europe fractured in public on the same question. ECB President Lagarde used a May 8 Madrid speech to reject private euro stablecoins outright; on May 12, Banque de France deputy governor Denis Beau broke with her, calling for a private tokenized-euro effort to counter dollar-stablecoin dominance, a position closer to the Bundesbank's. The Qivalis bank consortium expanded to 37 banks across 15 countries by May 20, pressing toward an H2 2026 euro-stablecoin launch under Dutch supervision. In the UK, the Bank of England signaled it will drop its proposed individual holding limits for sterling stablecoins in favor of aggregate issuance caps — a notable softening after industry pushback. And the SEC's Paul Atkins, keynoting the agency's May 12 tokenization roundtable, analogized on-chain ledgers to "Spotify" and legacy settlement infrastructure to "vinyl records," and urged Congress to finish CLARITY. The pattern across all of it: the legislative and securities-regulator track is moving toward the private-stablecoin operating model, and the central-banking track is split between accommodating it and resisting it on monetary-sovereignty grounds.

The year's largest DeFi incident closed operationally and reopened legally in the same week. By May 15, Kelp DAO and Aave had restored rsETH deposits, withdrawals, redemptions, and bridging, with rsETH fully backed again, after rebuilding the bridge that failed: a BailSec-audited reconfiguration to four independent attestors, block confirmations raised from 42 to 64, and all layer-2-to-layer-2 routes deprecated outright — the structural fix to the single-verifier flaw that caused the exploit. The DeFi United recovery coalition closed at over $300M.

Two tails are still open, and they are the ones worth tracking. The ~$71M of frozen ETH at the center of the recovery has not been released: a Manhattan federal judge declined Aave's emergency motion for full release around May 13, with a competing restraining claim from terrorism-victim creditors who argue the ETH is North Korean property still attached; supplemental briefs are due May 22 and the next review is around June 5. And the Aave DAO's own 25,000-ETH contribution (ARFC 24740) never advanced to a vote — the recovery was completed through partner pledges, Mantle's credit facility, and recovered collateral without it. The collateral-framework overhaul the incident was supposed to produce — the Asset Safety Tier and Risk Firewall proposals — remains at the temperature-check stage. The base layer survived; the governance and legal machinery the episode was meant to harden has not yet moved. Underneath, the rotation it triggered is now the operating baseline: Aave's deepest stablecoin markets are smaller, Morpho is larger, and the spread to sUSDS is gone.

The USDe/USDm leveraged loop on MegaETH that crossed $1B of Aave deposits on May 7 kept growing through the window with no stress event. Risk Stewards 24901 on May 11 raised the USDm borrow cap from 450M to 600M with top loopers running health factors of 1.01 to 1.06, and further cap increases followed just after the window. But the token the loop is built to support went the other way: MEGA traded near $0.089, down about 15% on the week and close to its all-time low, well below the ~$0.130 it reached on the early-May buyback. The first programmatic buyback ran May 7 to May 8; no second one was confirmed in the window, with the cadence likely monthly. The risk vector has shifted. A week ago the question was whether the loop's dependency chain could hold above $1B; this week it held, and the open question is token-economic — the loop keeps filling its caps to full utilization while the MEGA buyback leg that is supposed to capture that activity weakens.

The GENIUS implementation cycle ground forward in the background: the FDIC's Approval Requirements rule for subsidiary stablecoin issuance closed for comment on May 18 (the substantive bank-versus-issuer letters were not yet docketed at writing), and the AICPA filed urging regulators to adopt its stablecoin reporting criteria. On enforcement, Tether's trailing 30-day freeze tally held around $514M, and a more interesting precedent question opened on May 15: terrorism-victim judgment creditors filed in the Southern District of New York to redirect the ~$344M of previously frozen Iran-linked USDT toward their awards — an early test of whether frozen stablecoin balances are attachable property. OFAC's May 20 Sinaloa-cartel designations, just outside the window, again included Ethereum addresses, continuing the issuer-freeze-plus-sanctions-plus-forensics tool stack.

The tokenization land grab kept pace with the reserve story. Ondo's Global Markets tokenized-equities platform surpassed $1B in TVL on May 11 with more than 260 tokenized U.S. stocks and ETFs and over 70% share among tokenized-equity issuers. New entrants kept arriving at the edges: Genius Group detailed a GENIUS-targeted "JUSD" through Bermuda's Jewel Bank for H2 2026, the stablecoin neobank Fasset raised $51M from SBI Group and Investcorp to scale 50-plus emerging-market corridors at a ~$32B annualized run rate, and Grupo Salinas tapped Anchorage to run stablecoin settlement for its Mexican exchange — a second Anchorage bank-solutions win in two weeks, evidence its issuance pipeline is converting. And the bridge-exploit theme that drove the rsETH incident did not stop: the Verus-Ethereum bridge lost about $11.6M on May 18 to the same forged-message class of attack, with 2026 DeFi losses now cited above $750M.

The stablecoin question this week stopped being "is this legitimate" and became "who captures the economics, and who governs it." CLARITY clearing committee narrows the legitimacy question — a comprehensive framework with a workable yield compromise is now one floor vote and one reconciliation away, and the market is building as if it will arrive. But the same week answered a sharper question that institutions have been slower to price: in the operating model that just went live, the reserve float migrates to whoever controls distribution. Circle can grow circulation 28% and volume 263% and still watch net income fall, because the exchange takes its half and the trading venue now takes 90% of its own balances. That is not a Circle problem; it is the structure of the business. The issuer is becoming the lowest-margin layer in its own stack, sandwiched between distribution venues above it and reserve-vehicle managers (BlackRock, JPMorgan) below it.

For ArkenYield, this changes what underwriting a stablecoin exposure actually means. Reserve composition and attestations tell you whether the dollar is good. They do not tell you who earns the money, and this week the answer to that question moved. A stablecoin's economic profile now depends on its distribution contracts — the Coinbase share, the Hyperliquid split, the next venue's demand — as much as on what backs it. We underwrite stablecoin exposure by counterparty identity, venue design, and operating economics; this is the week the operating economics detached from the issuer and reattached to the venue, and the week a stablecoin-skeptic Fed chair and a fracturing Eurosystem made clear the governance question is not settled even as the legislative one nearly is.

• CLARITY on the Senate floor: the 60-vote whip count, reconciliation with the Senate Agriculture market-structure text, the unresolved ethics provisions, and whether leadership targets a vote before the August recess.
• Hyperliquid-deal fallout: whether Polymarket, Jupiter, or other venues extract similar reserve-income splits; Circle and Coinbase guidance on the margin hit; and the USDH-to-USDC migration mechanics.
• Warsh's first signals after May 22 on stablecoin supervision, master-account and payment-system access, and the Fed balance sheet.
• The FDIC May 18 comment letters once docketed — the sharpest bank-versus-issuer fight, on subsidiary stablecoin issuance.
• The Aave ~$71M frozen-ETH proceeding: the May 22 supplemental briefs and June 5 review (terrorism-victim creditors versus the DeFi United pool), whether ARFC 24740 ever reaches a vote, and whether the collateral-framework overhaul escalates past temperature check.
• MegaETH: the next MEGA buyback (expected early June) and whether a weakening token pressures the USDm reward economics that sustain the loop.
• Ethena: whether the USDe recovery toward $4.44B and the sUSDe move back toward 4.3% hold, and the next Anchorage attestation.
• The USDS true circulating level, where dashboards disagree by several billion depending on whether sUSDS is included.
• Europe: whether the Beau-Lagarde split widens, Qivalis's 37-bank consortium toward an H2 2026 launch, and the Bank of England's detailed sterling-stablecoin rules.
• BlackRock's BRSRV/BSTBL and JPMorgan's JLTXX approval timelines, and whether issuers actually commit reserves to them.
• Western Union USDPT corridor volumes and whether any Tempo design partner converts from announced to live.