Circle Internet Group, Inc. (NYSE: CRCL) — A Toll Bridge That Pays Most of Its Tolls to the Landlord

An independent fundamental research note | Report date: 2026-06-10 Sector: Financial / Crypto-Financial Infrastructure (stablecoin issuer) | CIK: 0001876042 Price at analysis: $78.93 | 52-week range: $31.00–$298.99 | Economic market cap ~$19.6B

⚡ Claude’s Take

This block is the author’s own subjective opinion and general information only — not investment advice. The analysis that follows takes no position and carries no price target; it discusses valuation only as embedded expectations and scenarios.

Verdict: HOLD / AVOID-at-this-price — a genuine franchise wrapped around structurally taxed, rate-dependent economics. Accumulate only on real weakness (sub-$55, ~20x normalized Adj. EBITDA / ~$13–14B EV). Not a short here. Conviction: medium.

Tag: “The most important dollar on the internet — that keeps 39 cents of it.”

Circle is the rare crypto-era IPO that is built on something real: USDC is the most-transacted digital dollar in the world, the GENIUS Act handed the category a federal charter, and serious enterprises (Stripe, Meta, DoorDash, Visa-measured commerce) are adopting it rather than building their own. That is the bull case, and it is not fake. But the financial architecture underneath is brutal in three compounding ways the $299 IPO mania ignored and the current price still under-discounts: (1) the product is a zero-yield, zero-switching-cost commodity dollar — there is no pricing power; (2) ~96% of revenue is simply the Fed funds rate applied to the reserve float, a number Circle does not control and which already fell from 5.0% to 3.5% in a year; and (3) Circle pays away ~60% of that reserve income to distributors — $1.4B to Coinbase alone in 2025 — under an agreement that gets worse for Circle as USDC succeeds on Coinbase’s rails, and that comes up for renegotiation in 2026 with Coinbase holding the trademark and the customer. The market is pricing Circle as an “internet financial platform” (Arc, CPN, agentic payments). Today those are testnet, beta, and unmonetized — real option value, but option value. Strip the narrative and you own a regulated narrow bank that nets ~39 cents on every reserve dollar, levered long-duration to a cutting Fed.

What the market is getting right: the franchise is durable enough that USDC won’t be displaced, the balance sheet is debt-free with ~$2.3B net cash, and a 74% drawdown from the peak has already purged the worst of the froth. What it is getting wrong: extrapolating 40% float growth as if it offsets rate cuts cleanly (it roughly cancels, leaving earnings flat for years), capitalizing Arc before a dollar of Arc revenue exists, and under-weighting the Coinbase renegotiation as a live, asymmetric threat to half the P&L. Framing: this is a quality-franchise-at-the-wrong-price / “great toll bridge, but the landlord takes the tolls” situation, not a falling knife and not a compounder-on-sale. At ~30x an Adjusted EBITDA that itself overstates true earnings (recurring SBC is ~$200M+/yr), $78.93 sits at the rich end of a $55–75 fair zone. Flip bullish if the 2026 Coinbase renegotiation lands materially in Circle’s favor and Arc/CPN convert to real high-margin revenue. Flip bearish if the Fed cuts 100bp+ while the Coinbase share of circulation keeps rising — that combination quietly halves the earnings power the multiple assumes.

1. Executive Summary

Circle Internet Group issues USDC, the world’s second-largest and most-transacted dollar stablecoin (~$77B in circulation at Q1 2026, ~28% of the fiat-backed stablecoin market). Its economic model is simple and severe: holders deposit dollars and receive USDC tokens bearing no interest; Circle invests the reserves in short-dated Treasuries and government money-market funds and keeps the yield. In FY2025, 96% of revenue ($2.64B of $2.75B) was this “reserve income” — net interest on the float. There is essentially one revenue line, and it equals float × short rate, both exogenous to management.

The business is genuinely profitable on a cash basis — FY2025 Adjusted EBITDA of $582M at a 54% margin (on revenue-less-distribution-costs) — but reports a GAAP net loss of −$69.5M because the June 2025 IPO triggered ~$516M of one-time RSU stock-based compensation. The more durable problem sits above the EBITDA line: Circle pays ~60% of revenue (~63% of reserve income) to distribution partners, overwhelmingly Coinbase, which collected $1.4B in FY2025 and whose share grows as USDC migrates onto its platform. The distribution take has risen from 50.9% of reserve income in FY2023 to 63.1% in FY2025. Circle keeps roughly 39 cents of every reserve dollar.

Three structural facts define the investment debate. First, rate sensitivity: a 100bp decline in short rates cuts reserve income by ~$756M, only ~half-cushioned by lower distribution costs, for a ~$387M net pre-tax hit — roughly two-thirds of Adjusted EBITDA. The reserve return rate already compressed from 5.0% (FY2024) to 3.5% (Q1 2026). Second, the Coinbase dependency: a single counterparty takes half the residual economics, holds a perpetual USDC trademark license (assignable to Coinbase on certain Circle defaults), and faces Circle across a 2026 renegotiation. Third, the platform optionality: Arc (a new Layer-1 blockchain), Circle Payments Network, USYC, StableFX, and an “agentic payments” stack represent the bull’s path to a durable, diversified, higher-margin business — but as of FY2025 they generate ~4% of revenue and Arc is pre-mainnet.

Regulation is the swing factor and it cuts both ways. The GENIUS Act (signed July 2025) legitimizes the category, removes securities-law risk, and validates Circle’s compliance-first posture — but it also bans paying yield to holders (entrenching the zero-yield, distribution-bidding model) and explicitly invites banks and bigtech to enter. The stock has fallen ~74% from its $299 October-2025 peak to $78.93, leaving an economic market cap of ~$19.6B (~$17.3B EV against ~$2.3B net cash), or ~30x FY2025 Adjusted EBITDA. This memo takes no position; it lays out what the price embeds and what would falsify each side.

2. Business Overview

What Circle does. Circle Internet Group is the issuer of USDC (a US-dollar stablecoin) and EURC (a euro stablecoin), and operates a widening set of blockchain-financial infrastructure around them. A stablecoin is a blockchain-native token redeemable 1-for-1 for fiat currency; Circle mints USDC when an institutional customer deposits dollars via Circle Mint and burns it on redemption. The deposited dollars become reserve assets that legally back the tokens in circulation. (FACT, FY2025 10-K, Business.)

How it makes money — one engine, three letters: NII. The overwhelming majority of revenue is reserve income: the interest Circle earns on the reserve assets. Holders of USDC receive no yield. Circle therefore behaves like a narrow bank — it takes non-interest-bearing “deposits” (stablecoins), invests them in cash, ≤3-month Treasury bills, and overnight Treasury repo (held ~88% in the BlackRock-managed, SEC 2a-7 “Circle Reserve Fund,” custodied at BNY), and keeps the spread. (FACT, 10-K, MD&A and reserve footnote.)

Revenue composition (FACT, 10-K Statement of Operations and Note 11, in $000s):

“Other revenue” — subscription & services ($84.8M FY2025, mostly blockchain-network partnerships and USYC fees) plus transaction revenue ($24.3M) — grew 624% in FY2025, but from a trivial base, and remains ~4–6% of the total. (FACT, Note 11.)

Products and roadmap. The franchise today is USDC. Around it sits a portfolio at varying stages of monetization (FACT, 10-K + Q1 2026 call):

• USDC / EURC — the entire revenue base (reserve float). EURC is the largest euro stablecoin but tiny (€358M at Q1 2026).
• USYC — a tokenized money-market fund (acquired via Hashnote, Jan 2025); ~$3B AUM and now the largest tokenized MMF; small fee revenue in “other.”
• Circle Mint / CCTP / Gateway — institutional mint-redeem and cross-chain transfer rails; CCTP reached ~$50B quarterly volume, ~60% of cross-chain traffic; minimal direct revenue.
• Circle Payments Network (CPN) — B2B cross-border payments; ~$8–10B annualized total payment volume, 136+ financial institutions enrolled; deliberately not yet monetized (building volume first).
• Arc — Circle’s own Layer-1 blockchain (“economic OS”); public testnet only at FY2025, mainnet “coming soon”; an Arc token presale raised $222M at a $3B fully-diluted network value (a16z lead) in May 2026. Pre-revenue.
• StableFX, Agent Wallets / Nanopayments / Agent Marketplace, cirBTC — beta/launch-stage; pre-revenue.

Recurring vs. non-recurring. Reserve income is “recurring” only in the sense a bank’s non-interest deposit base is: sticky in aggregate, but each token is redeemable on demand with zero switching cost and no minimum-balance commitment. USDC minted and redeemed ~$150B in a single quarter against a ~$77B balance — extraordinary velocity that underlines the float is a high-churn pool, not a contracted annuity. (FACT, Q1 2026 call.)

Verdict. Circle is, economically, a single-product regulated narrow bank whose product is a zero-yield dollar token, supplemented by a portfolio of promising-but-immature platform businesses. The quality of the business model hinges almost entirely on two things it does not control — the level of short-term interest rates and the share of economics it must surrender to distributors — and on whether the platform roadmap converts narrative into high-margin revenue. As of today, the engine is NII; everything else is option value.

3. Industry Dynamics

The profit pool is the Fed funds rate on the float. This is the defining structural fact of stablecoin issuance. Industry-wide economics = (aggregate stablecoin float) × (short-term rate) − (distribution + operating costs). There is no consumer pricing, no take-rate on transactions (issuers generally don’t charge holders), and — post-GENIUS — no ability to pay holders for loyalty. The entire industry is a spread business on other people’s money. (INTERPRETATION, grounded in 10-K economics.)

Market structure: a duopoly on a fast-growing TAM. Two issuers — Tether (USDT) and Circle (USDC) — account for ~90%+ of fiat-backed stablecoin circulation; “all other stablecoins combined round to roughly zero,” in management’s framing. The total stablecoin market grew ~32% year-over-year into 2026 and bulls project a multi-trillion-dollar pool as payments, collateral, and (prospectively) agentic/machine commerce migrate on-chain. USDC circulation grew 72% in FY2025 to $75.3B. (FACT, 10-K + calls.) On the surface, a consolidated duopoly on a secularly growing market looks like an attractive structure.

Why it is structurally weaker than it looks. Apply a Greenwald/Marathon lens:

• The good sold is a commodity. One issuer’s fully-collateralized $1 token is fungible with another’s. There is no product axis on which to differentiate or charge.
• Switching costs are near-zero for the holder. A USDC holder redeems for $1 and mints a rival token, or swaps on a decentralized exchange in seconds. Circle’s own 10-K concedes customers have “no obligation to continue,” “no minimum volume commitments,” and switching costs that “may not be significant enough.”
• Revenue is 100% exogenous on the price axis. Circle has zero pricing power over its revenue line — it is set by the Federal Reserve. Q1 2026 proved it: circulation rose 28% YoY but revenue rose only 20% as the reserve rate fell 66bps.
• Competition is fought on distribution, and distribution is rented. Because the product can’t differentiate and the yield can’t go to holders, issuers compete by paying distributors (exchanges, wallets, fintechs) to carry and promote their token. That bids the float spread away to whoever owns the eyeballs — precisely Circle’s Coinbase problem (discussed below).
• Supernormal margins invite capital (Marathon). A 54% Adjusted-EBITDA-margin business being handed regulatory legitimacy is exactly the setup that attracts entrants. The GENIUS Act formalizes the on-ramp for banks and bigtech.

Regulation — the single biggest industry variable. The GENIUS Act (signed July 18, 2025; most provisions effective ~January 2027) created the first US federal framework for “permitted payment stablecoin issuers.” It: (i) requires full backing in high-quality liquid assets, redemption rights, AML compliance, audits, and holder priority in insolvency; (ii) explicitly excludes payment stablecoins from “security” status — removing a major legal overhang; and (iii) prohibits issuers from paying holders any yield or interest. Net read: legitimizing and validating for Circle’s compliance-first model, but it entrenches the zero-yield/distribution-bidding dynamic and expressly invites bank entry — the 10-K warns GENIUS “may further encourage new entrants, including banks with established customer bases.” In Europe, MiCA governs EURC (small today). (FACT, 10-K + calls.)

Verdict: a structurally mediocre industry dressed as a good one. It has scale economics and real regulatory barriers to operating at global scale, but the core product is undifferentiated, holder switching costs are near-zero, the revenue line is a rate spread no issuer controls, and competition channels into paying away that spread to distributors. It most resembles the money-market-fund / transaction-banking complex: scale matters, but margins get competed toward the cost of distribution. The regulatory tailwind is real but double-edged — it legitimizes the incumbents while lowering the drawbridge for far larger balance sheets.

4. Competitive Position

Name the moat precisely. Circle’s defensible asset is “USDC the standard” — a genuine, hard-to-replicate network of liquidity, interoperability (native support on 30+ blockchains), institutional integrations, and regulatory legitimacy. A new entrant cannot conjure $77B of circulation, 185-country banking access, and ubiquitous exchange/wallet/DeFi integration overnight. This network is why serious platforms (Stripe “all in,” Meta creator payouts, DoorDash driver payouts, Visa-measured commerce, Polymarket settlement) adopt USDC rather than mint their own. It is also why USDC leads on the metric that signals utility — transaction volume (~63% of commercial stablecoin transactions per Visa’s analytics; ~80% of on-chain volume per third parties including Solana). This is a real moat mechanism: demand-side network effects plus a regulatory license. (FACT, calls + 10-K.)

Now pressure-test it against financial outcomes — the test that matters. A moat is only a moat if its removal would degrade economics. Here the moat is narrow, shared, and rate-dependent:

1. The network effect accrues to the standard, but Circle captures only ~39 cents on the dollar of the resulting float income. The other ~61% is the toll Circle pays distributors to keep USDC distributed. A network effect you must rent back from your distributors at ~half your gross margin is a leaky moat.
2. Switching costs for holders are ~zero — so there is no Greenwald “customer captivity.” The captivity that exists is on the distributor side (it is costly for Coinbase to rip out USDC), and Circle pays dearly for it.
3. The regulatory license is commoditizing. Pre-GENIUS, being the “regulated stablecoin” was a differentiator. Post-GENIUS, the license becomes a cost of entry others — including banks — can also pay.

The Coinbase dependency — the crux. This is where the competitive position is most compromised, and it deserves to be stated bluntly. Under the August 2023 Collaboration Agreement (FACT, 10-K + S-1):

• Coinbase receives 100% of the reserve income on USDC held on Coinbase’s platform, plus 50% of the residual “ecosystem” reserve income after Circle’s issuer retention and third-party distribution payments — even on USDC Coinbase had nothing to do with distributing.
• Coinbase paid distribution amounted to $1.4B in FY2025 (up from $924.5M) — ~53% of Circle’s gross reserve income to one counterparty.
• The split structurally worsens for Circle as USDC migrates onto Coinbase: only ~18% of USDC sits “on Circle’s platform”; management explicitly flagged in Q1 2026 that net reserve margin shifted against Circle “as Coinbase represented a larger share of circulation during the quarter.”
• Coinbase holds a perpetual trademark license to USDC/EURC with a “flipped license” provision that — if Circle is ever legally barred from paying Coinbase — forces Circle to assign the USDC trademark to Coinbase. The 10-K concedes this “could have a significant and detrimental effect on our ability to operate as the issuer of USDC.”
• The agreement’s initial term opens a 2026 renegotiation window. Coinbase holds the leverage: it owns the largest distribution channel, a minority equity stake in Circle, the trademark, and its own stablecoin/onchain ambitions (Base L2).

In plain terms: Circle owns the regulated-issuer license and the brand; Coinbase owns the customer and the distribution, and has structured the deal to capture ~half of a rate spread that is itself shrinking — with a renegotiation pending and competitive optionality of its own.

The competitor set (named, with specifics). (FACT, 10-K + Bernstein conf.)

• Tether (USDT) — ~2x USDC’s circulation, offshore/less-regulated, and far more profitable because it pays little distribution and holds riskier, higher-yielding reserves (Bitcoin, gold, secured loans — Allaire: “almost like a macro hedge fund”). Tether competes from greater scale and margin; its US-compliant USAT is tiny so far.
• PayPal (PYUSD), Ripple (RLUSD) — regulated, distribution-backed, small but credible.
• Stripe (via Bridge) — strategically the most dangerous; “all in on USDC” today but now owns stablecoin issuance/orchestration and the distribution to compete or renegotiate.
• Banks / deposit tokens (JPM and others) — post-GENIUS, banks may issue payment stablecoins atop vastly larger balance sheets and customer bases. BlackRock has contractually agreed not to launch a competing payment stablecoin (a tell the threat is real).
• Tokenized MMFs (BlackRock BUIDL, Circle’s own USYC) — the threat to the holder side: once you can hold a yield-bearing tokenized dollar, the rationale for holding zero-yield USDC narrows to pure transactional float.
• Coinbase itself — distributor, equity-holder, trademark-licensee, and latent competitor.

Can a Walmart or a bank consortium just issue their own? Post-GENIUS, legally yes. Circle’s defense is the network-effects / “winner-take-most” argument — that issuing a coin is easy but building scaled liquidity and developer/distribution network effects is hard, so big players adopt USDC rather than “go it alone” (Meta is cited as proof the feared bigtech-coin wave didn’t materialize). This is the bull’s strongest argument and the bear’s biggest open question: it remains unproven that a large captive ecosystem (a Walmart, a bank consortium) couldn’t peel off exactly the high-value, sticky float Circle most needs.

Verdict. Circle has a narrow, shared, rate-dependent moat — not a wide one. USDC the standard is genuinely defensible and explains adoption-over-build. But the product is a zero-switching-cost commodity, the revenue is a Fed-set spread, and ~half of that spread flows to a single counterparty with renegotiation leverage and competitive optionality. This is a competitively-exposed financial intermediary with a concentration problem — not a toll-road monopoly, and emphatically not a software platform, until Arc/CPN prove otherwise.

5. Growth History and Forward Opportunities

Historical growth. Circulation and revenue have grown fast off the 2023 USDC depeg trough (USDC briefly broke the buck during the March 2023 Silicon Valley Bank failure, where Circle held ~$3.3B of reserves; it recovered, but circulation fell sharply and only re-accelerated in 2024–25). (FACT, 10-K history.)

Two things stand out. First, FY2025’s 64% revenue growth was driven by a 94% rise in average circulation partly offset by a 90bp yield decline — i.e., volume outran price, but price (rates) is now the headwind. Second, Q1 2026 already shows the squeeze: circulation +28% YoY but roughly flat sequentially, revenue +20%, because the reserve rate fell. Management itself notes the stablecoin market was “relatively flat in the quarter” — circulation growth is cyclical and digital-asset-market-linked, not a smooth secular ramp (circulation held up despite a ~45% crypto-market drawdown from the October 2025 peak, which bulls cite as evidence of growing non-crypto utility).

Quality of historical growth: mixed. It is organic (USDC adoption, not acquisition) and reflects genuine utility expansion (payments, treasury, EM “digital dollar” demand). But it is low-quality in the sense that matters for earnings: revenue per unit of circulation is falling with rates, and incremental circulation increasingly lands on Coinbase (worsening the split). Growing the float is not the same as growing the economics.

Forward opportunities (the bull’s real case). (FACT, calls + 10-K.)

• Float compounding — management guides ~40% USDC CAGR “over a multi-year through-cycle.” If rates stabilize, float growth is the dominant earnings lever.
• Other revenue diversification — guided to $150–170M in FY2026 (from $110M), from USYC fees, network partnerships, and CPN as it monetizes. Still <6% of revenue.
• CPN (payments) — ~$8–10B annualized TPV growing ~75% in months; the path from “free volume” to monetized B2B cross-border payments is the most credible near-term diversification.
• Arc (Layer-1) — the largest and most speculative vector: a stablecoin-native blockchain with $222M of presale capital and an institutional partner roster (a16z, BlackRock, ICE, Apollo, Standard Chartered). If Arc becomes the settlement layer for tokenized assets and agentic payments, it could add a network-economics revenue stream and deepen USDC’s moat. Mainnet is pending; revenue is unquantified (management will guide “next call”).
• Agentic / machine payments — Circle claims 99.8% of x402-protocol agent payments settle in USDC; a genuine call option on AI-agent commerce.

Verdict. High circulation growth, declining-quality economics on the core engine, and a credible-but-unproven set of diversification options. The forward thesis requires float growth to outrun rate compression and at least one of CPN/Arc to convert into real high-margin revenue. The first is a coin flip tied to the Fed; the second is option value. This is not yet a self-evidently high-quality growth story — it is a fast-growing float on deteriorating unit economics, with attractive but immature optionality stacked on top.

6. Financial Quality

Margin structure. The headline GAAP picture is misleading in both directions, so the layers matter:

• Gross of distribution, Circle earns the full reserve income. Net of distribution (“RLDC” — revenue less distribution costs), Circle keeps ~39–41% (FY2025 RLDC margin 39.4%; Q1 2026 41.4%). This is the real top line. (FACT.)
• Distribution & transaction costs: $1,664M FY2025 (60.6% of revenue), up from $1,017M (60.7%) FY2024 and $728M (50.2%) FY2023. As a share of reserve income, the take rose from 50.9% → 61.2% → 63.1%. Coinbase is the bulk; Binance and “other” partners are growing faster. (FACT, 10-K.)
• Adjusted EBITDA: $582M FY2025 (54% of RLDC), $285M FY2024. The 54% margin is struck on net revenue (RLDC), not gross revenue — an important nuance when comparing to other financials.

The FY2025 GAAP loss is IPO mechanics, not operating deterioration. Net loss was −$69.5M vs +$157M in FY2024. The bridge (FACT, 10-K):

• +~$516M one-time SBC — RSUs whose liquidity-event vesting condition was met at the NYSE listing. This single item swings the result.
• −$87M non-cash loss on convertible-note fair value (rose with the stock; converts retired Jan 2026).
• −$23M charitable contribution of Class A shares to the Circle Foundation.
• +$33M tax benefit (vs a $65M expense prior year), from SBC deductions and R&D credits. Underneath, the business was solidly profitable (Adj. EBITDA +104%). Q1 2026 returned to +$55.2M GAAP net income.

But do not take “Adjusted” at face value. Management concedes SBC “has been, and will continue to be, a recurring expense.” Post-IPO SBC is running ~$200M+ annualized (Q1 2026 SBC $51.8M). A normalized earnings figure must keep recurring SBC as a real cost: stripping the one-time IPO catch-up but charging ~$200M/yr of ongoing SBC, normalized FY2025 net income is on the order of ~$300–380M — meaningfully below Adjusted EBITDA and the basis on which the multiple should be judged. (Interpretation.)

The dominant quality-of-earnings issue: rate sensitivity. The 10-K (Item 7A) quantifies it on the Dec-31-2025 float (FACT):

A 100bp cut removes ~$387M of net pre-tax income — ~two-thirds of FY2025 Adjusted EBITDA. The distribution-cost structure provides a partial automatic hedge (Circle pays away ~half of any reserve-income swing), but only half. The reserve return rate already fell 5.0% → 4.1% → 3.5%. The model is a leveraged long-duration bet that float grows faster than yields fall.

Balance sheet — clean and a genuine strength. Corporate cash $1.53B + $0.82B segregated corporate-held stablecoin cash = ~$2.3B liquidity; no traditional debt (the legacy SeedInvest convertible converted to equity by January 2026). Total equity $3.33B (6x FY2024, on IPO + follow-on + preferred conversion + SBC paid-in-capital); goodwill $266M + intangibles $411M → tangible equity ~$2.7B. The $75–77B of segregated reserve assets and the matching deposit liability gross up the balance sheet but are pass-through customer funds, not Circle’s capital. (FACT, 10-K/10-Q.)

Cash flow — light capex, but read the geography. FY2025 operating cash flow was $542M; the massive $31B of reserve-deposit movement sits in financing (not operating), so reported OCF reflects the corporate economics. Capex is genuinely light (~$70M/yr, mostly capitalized software) — an asset-light model. Two cautions: ~$270M of RSU tax-withholding cash also sits in financing (understating the true cash cost of comp in OCF), and Q1 2026 OCF fell to $21M on a $101M accruals paydown. (FACT/INTERPRETATION.)

Returns on capital. Conventional ROE/ROIC are not yet meaningful — FY2025 shows a GAAP loss, and the equity base was just reset by the IPO. On normalized earnings (~$300–380M) against ~$2.7B tangible equity, normalized ROTE is ~12–14% — respectable but not exceptional for a business of this volatility, and flattered by the asset-light structure. (Interpretation.)

Verdict: do economics improve with scale? Only partly, and not where it counts. Operating leverage on fixed costs is real (opex grows slower than RLDC). But the dominant variable cost — distribution — rises with scale and success, and the revenue line is rate-set. So scale improves the opex ratio while the distribution tax and rate exposure cap the through-cycle economics. This is a high-margin-today, structurally-pressured financial intermediary, not a self-evidently scaling software P&L.

7. Capital Allocation

The record is short (one year public) and dominated by the IPO/follow-on. (FACT, S-1, 10-K, DEF 14A.)

• June 2025 IPO @ $31.00: company sold 19.9M primary for ~$583M net; insiders sold 19.2M secondary (~$595M, none to Circle).
• August 2025 follow-on @ $130.00 (4.2x the IPO price, ~10 weeks later): company sold 3.5M primary for ~$445M net; insiders sold 8.0M secondary (~$1.04B).
• Total primary to Circle ~$1.03B; insiders cashed out ~$1.6B across the two offerings. ~$111M of IPO proceeds funded the tax withholding on insiders’ RSU vesting.

The timing was excellent for the company (raising $445M at $130 weeks after a $31 IPO is superb execution). But the offerings were as much an insider exit as a capital raise — pre-IPO holders monetized more than Circle itself raised.

M&A — small, coherent tuck-ins. Hashnote (Jan 2025, ~$100M, of which $96.2M was goodwill — effectively an acqui-hire for the USYC tokenized-MMF franchise and team); Malachite consensus software ($15M); and the August 2023 buyout of the remaining 50% of Centre Consortium from Coinbase ($209.9M) to take sole governance of USDC. The $96M of goodwill on a $100M Hashnote deal carries impairment risk if USYC stalls, but the dollar amounts are immaterial to the enterprise. (FACT, 10-K.)

Arc token presale ($222M @ $3B FDV). This is a network capital raise (Arc token economics), not Circle equity dilution; proceeds and economics will flow through “other revenue” and Arc incentive costs over time, with management deferring guidance. It is simultaneously a credibility signal (a16z, BlackRock, ICE, Apollo, Standard Chartered as stakeholders) and an unquantified new variable in the P&L. (FACT, Q1 2026 call.)

No buybacks, no dividends (never paid; none anticipated). Appropriate for a hyper-growth recent IPO — but there is no capital-return discipline to assess yet, and the incentive structure points toward growth/M&A deployment rather than return. (FACT, S-1.)

Incentive alignment — rewards scale, not per-share value. (FACT, DEF 14A.)

• FY2025 NEO comp: CEO Allaire $16.3M total ($900K salary; a notable $4.1M “all other” = personal/home security; one-time $500K IPO bonus); CFO Fox-Geen $7.8M; CCO Razzaghi $20.8M (incl. a $16M one-time promotion grant). 2026 CEO equity grant raised to $15M.
• The short-term plan is 70% Adjusted EBITDA / 30% non-financial growth milestones (USDC circulation, CPN volume, USYC integration, trust charter). Equity vests on time + a one-time liquidity condition. There is no TSR, no ROIC/ROE, and no per-share metric anywhere in the plan.
• A management team paid on Adjusted EBITDA and circulation is structurally incentivized to chase float growth even at deteriorating reserve-income economics — precisely the dynamic the Coinbase split punishes.

Founder control — moderate and time-limited (milder than the headline). Allaire holds ~23.9% of voting power (via 5-vote Class B, capped at 30% aggregate); all directors & officers 31.4%; not a controlled company (8 of 9 directors independent). Class B sunsets June 5, 2030 or when Allaire ceases to be CEO-or-Chair. The 30% cap and hard sunset are genuinely more shareholder-friendly than typical uncapped founder dual-class structures — but a classified board, 66⅔% supermajority provisions, and Delaware §203 entrench against near-term change. (FACT, DEF 14A.)

Litigation overhang. The FT Partners suit claims a 7% advisory fee on ~$1.9B of IPO/follow-on proceeds (~$133M order-of-magnitude exposure), is contested, and is not reserved; settlement could be in cash or equity and “may be substantial.” (FACT, 10-K.)

Verdict: mixed-to-cautious. Well-timed raises and coherent tuck-ins, but a one-year record dominated by insider monetization, incentives geared to scale rather than per-share value, no capital-return track record, and a live ~$130M litigation tail. Not red-flag capital allocation — but not the disciplined, owner-operator profile the founder-led framing might suggest.

8. Changes and Headwinds — Last Two Years

Major changes (FACT, 10-K, calls, proxy):

• IPO (June 2025) and follow-on (August 2025) — transition to public company; ~$1.03B raised; ~$516M one-time SBC; balance-sheet recapitalization.
• GENIUS Act signed (July 2025) — federal stablecoin regime; removed securities-law risk; banned holder yield; opened the door to bank/bigtech entrants. The defining regulatory event.
• OCC conditional approval (Dec 2025) for Circle National Trust, a national trust bank to custody/oversee reserves with first-priority holder security interest — strengthens the safety/bankruptcy-remoteness narrative.
• Hashnote acquisition (Jan 2025) → USYC, now the largest tokenized MMF (~$3B).
• Arc — testnet launch (2025) → token presale $222M @ $3B FDV (May 2026); mainnet pending.
• Product proliferation — CPN (launched 2025, scaling to ~$10B TPV), StableFX (beta), xReserve, Agent Stack (Agent Wallets/Nanopayments/Marketplace), planned cirBTC.
• Enterprise adoption — Stripe, Meta, DoorDash, Visa-measured commerce, Polymarket, Kyriba, Ramp, Intuit partnership, Korean exchange expansion.
• Reserve concentration — New BlackRock MOU (Mar 2025): ≥90% of US fiat reserves with BlackRock products; BlackRock won’t launch a competing payment stablecoin.

Headwinds (FACT/INTERPRETATION):

• Falling rates — reserve return rate 5.0% → 3.5%; the central earnings headwind.
• Crypto drawdown — ~45% from the October 2025 peak; circulation held roughly flat sequentially in Q1 2026 (resilient, but growth stalled).
• Coinbase 2026 renegotiation — looming, asymmetric, and tied to half the economics.
• Post-lockup insider supply — broad, continuous selling Dec 2025–June 2026 (founders, officers, IDG/Oak VCs), entirely planned/diversification-driven, with zero open-market purchases. A real share overhang and, at best, a non-signal.
• Stock collapse — −74% from peak; a reputational/sentiment headwind and a reset of the cost-of-equity for future raises.

Verdict. The two-year change set is genuinely thesis-relevant in both directions: the regulatory and adoption wins strengthen the franchise, while falling rates, the Coinbase renegotiation, and the post-lockup supply weaken the near-term setup. On balance the franchise strengthened and the economics/sentiment weakened — which is precisely why the stock round-tripped.

9. Risk Analysis (Risk Matrix)

Catastrophic-loss assessment. A total loss is unlikely given a debt-free balance sheet, ~$2.3B net cash, fully-collateralized pass-through reserves, and a real franchise — but a severe impairment of equity value is entirely plausible from the combination of (a) sustained rate cuts and (b) an adverse Coinbase renegotiation, which together could halve normalized earnings power against a full multiple. The reserve/depeg tail is low-probability but high-severity (2023 precedent).

10. Valuation Discussion (Embedded Expectations)

No price target, no recommendation. This section frames what the current price embeds.

The price and the multiples. At $78.93, with ~248.6M economic shares (229.9M Class A + 18.7M Class B), the economic market cap is ~$19.6B; on ~266.7M diluted shares, ~$21.0B. Net of ~$2.3B corporate liquidity (and no debt), enterprise value is ~$17.3B. Against FY2025 figures (FACT/INTERPRETATION):

For context, at the $299 October-2025 peak the economic market cap was ~$60B+ (~$66B diluted) — roughly 115x Adjusted EBITDA and ~$57B EV on a ~$582M EBITDA base. The 74% drawdown has compressed the multiple from absurd to merely full.

Embedded-expectations / reverse logic. Reserve income ≈ float × rate. At ~$77B float and a 3.5% rate, run-rate reserve income is ~$2.7B; at a 39–40% RLDC margin, net revenue ~$1.07B; against $570–585M guided opex, run-rate Adjusted EBITDA ~$570M. So today’s ~$17.3B EV is ~30x a roughly flat run-rate EBITDA. To justify that, the market must underwrite a combination of:

1. Float compounding ~40% (management’s through-cycle target), and
2. Rates not collapsing — because each 100bp cut erases ~$387M of pre-tax income, and 40% float growth on $77B (~+$31B) at ~3.5% gross / ~39% retained adds only ~$420M of gross reserve income and ~$165M of net (RLDC) — i.e., a year of strong float growth roughly offsets a single 100bp rate cut. The two largely cancel. and
3. Arc / CPN / platform converting narrative into real, high-margin revenue to provide a second earnings leg the core engine cannot.

What the market is pricing correctly vs. incorrectly (INTERPRETATION).

• Correctly: the franchise survives and the float compounds through-cycle; the balance sheet is pristine; the GENIUS regime is a legitimizing tailwind; the drawdown removed the worst froth.
• Possibly incorrectly: (i) extrapolating float growth as if it adds to earnings when, at current rates, it largely offsets rate cuts; (ii) capitalizing Arc before any Arc revenue exists; (iii) under-pricing the 2026 Coinbase renegotiation as a live, asymmetric claim on half the P&L; and (iv) anchoring on Adjusted EBITDA that overstates true earnings by the recurring-SBC run-rate.

Scenario sketch (illustrative, not a target).

• Bear: Fed cuts 100–200bp through 2027, Coinbase share of circulation keeps rising (worse split), Arc revenue slips. Normalized EBITDA drifts toward ~$350–400M; a de-rating to ~15–20x EV/EBITDA implies an EV well below today’s — a materially lower equity value.
• Base: rates hold near current levels, float compounds ~30–40%, other revenue hits guidance, Arc is promising but immature. Roughly flat-to-modestly-higher normalized EBITDA at ~$570–650M; a ~25–30x multiple roughly supports the current zone.
• Bull: rates stabilize, the 2026 Coinbase renegotiation improves Circle’s split, CPN monetizes, and Arc becomes a genuine settlement layer with token economics. A second high-margin revenue leg justifies a re-rating well above today.

The honest read: at $78.93 the price sits at the upper end of a defensible fair zone, embedding base-to-bull outcomes while the dominant near-term variable (rates) points the other way and the second-largest (Coinbase) is unresolved.

11. Variant Perception

Consensus belief. Circle is the regulated, blue-chip stablecoin franchise — the “digital-dollar standard” — riding a multi-trillion-dollar TAM and a once-in-a-decade regulatory unlock (GENIUS), with optionality (Arc, payments, agentic commerce) that justifies a platform multiple. Sell-side targets cluster well below the peak but above today (e.g., Mizuho Neutral, PT cut to $85 in June 2026), implying “good company, fair-to-full price.”

Strongest bull case. USDC’s network effects are real and compounding — transaction-volume leadership (~80% on-chain), accelerating enterprise adoption (Meta/Stripe/DoorDash/Visa-commerce), and the fact that feared bigtech coins didn’t materialize all suggest a winner-take-most standard. GENIUS cements the duopoly and removes legal risk. The float compounds ~40% through-cycle, and Arc + CPN + agentic payments add high-margin revenue legs that transform Circle from a rate-spread business into a diversified internet-financial platform. At ~30x EBITDA off a $19.6B base — a fraction of the peak — you are buying the category leader after a 74% reset, with a debt-free balance sheet and a regulatory moat.

Strongest bear case. Circle is a moatless narrow bank whose only product is a zero-yield commodity dollar, whose revenue is set by the Fed (and falling), and which pays ~60% of that revenue to a single counterparty (Coinbase) that holds the customer, the trademark, and 2026 renegotiation leverage. Float growth largely offsets rather than adds to earnings at current rates. The platform story (Arc/CPN) is pre-revenue narrative used to justify a software multiple on a financial-spread business. Insiders cashed out ~$1.6B at the IPO/follow-on and have sold steadily since, with zero open-market buying. ~30x an Adjusted EBITDA that overstates true earnings, on an asset whose earnings power a single 100bp rate cut can cut by two-thirds, is a full price for a structurally pressured intermediary.

The 3–5 assumptions that decide it:

1. The path of short-term rates. The single most important variable; the model is leveraged long-duration.
2. The Coinbase economics (2026 renegotiation + on-platform mix). Half the P&L depends on a counterparty’s leverage.
3. Whether float growth can outrun rate compression — i.e., whether ~40% CAGR is achievable through-cycle and whether it adds to earnings or merely offsets cuts.
4. Whether Arc / CPN monetize into a real second high-margin revenue leg, or remain option value.
5. Whether the regulatory tailwind (GENIUS) net helps the incumbent or net invites larger competitors.

What would falsify each side. Bull falsified if: Q2–Q4 2026 prints show reserve income falling YoY despite circulation growth (rate cuts winning), the Coinbase renegotiation holds or worsens Circle’s split, and Arc revenue keeps slipping. Bear falsified if: the 2026 renegotiation materially improves Circle’s retained share, CPN/Arc generate disclosed, growing, high-margin revenue, and float compounds through a rate floor such that RLDC grows double-digits YoY for several quarters.

12. Fact vs. Interpretation Table

13. Open Questions

1. Coinbase 2026 renegotiation — terms, timing, and whether Circle’s retained share improves, holds, or worsens. The single largest swing variable on economics; not disclosed.
2. Arc token economics — revenue-recognition timing, Circle’s stake, incentive-cost drag, and whether Arc generates material disclosed revenue. Management deferred guidance to “next call.”
3. Aggregate post-lockup insider selling — exact $ by holder across 200+ Form 4s (sampled only here); and whether any insider has made open-market purchases.
4. FT Partners litigation — outcome, and whether any reserve has been booked against the ~$133M claim.
5. Normalized run-rate SBC and dilution — the steady-state SBC figure and net dilution after buyback offset (none today).
6. Reserve-income rate path — Circle’s reserve duration/positioning and how quickly the 3.5% return rate re-prices if the Fed cuts further.
7. On-platform mix trajectory — how high “USDC on Circle’s platform” (18% in Q1 2026) can climb, since it materially improves the retained margin.

14. What Must Be True (Bull and Bear)

Bull case — what must be true:

• USDC’s network effects prove durable and winner-take-most, so adoption keeps compounding the float at ~40% through-cycle (falsified if circulation growth stalls below ~20% for several quarters absent a crypto crash).
• Short-term rates stabilize near current levels — float growth then flows through to earnings rather than offsetting cuts (falsified if reserve income falls YoY for two-plus consecutive quarters despite circulation growth).
• The 2026 Coinbase renegotiation holds or improves Circle’s economics, and on-platform mix rises (falsified if the renewed terms reduce Circle’s retained share or the Coinbase share of circulation keeps climbing).
• Arc and CPN convert to disclosed, growing, high-margin revenue, building a credible second leg (falsified if “other revenue” stays <10% of total and Arc revenue keeps slipping through 2027).

Bear case — what must be true:

• Rates fall enough that reserve-income compression overwhelms float growth (falsified if the reserve return rate stabilizes and RLDC grows double-digits YoY).
• The Coinbase dependency proves a binding tax — half the economics stay (or grow) with a counterparty holding renegotiation and trademark leverage (falsified if a 2026 renegotiation materially improves Circle’s split).
• The platform optionality stays narrative — Arc/CPN remain immaterial to revenue (falsified by disclosed, scaling, high-margin Arc/CPN revenue).
• Post-GENIUS entrants (banks, bigtech, Stripe) peel off high-value captive float (falsified if USDC share of circulation rises and no large captive ecosystem launches a credible rival coin).

The cleanest single falsification test: track reserve income YoY against circulation YoY over the next 2–4 quarters. If reserve income grows roughly in line with circulation, the rate environment is benign and the bull math works. If reserve income lags circulation materially (or falls YoY while circulation rises), the rate cut is winning and the ~30x multiple is exposed.

15. Source Appendix

The full, dated source list appears in the Source Appendix below. Primary sources: Circle Internet Group FY2025 Form 10-K (filed 2026-03-09); Forms 10-Q for Q2/Q3 2025 and Q1 2026; DEF 14A (2026-04-01); Form S-1/A (2025-06-02); Forms 3/4 (insider transactions); earnings-call and conference transcripts (Q2 2025–Q1 2026, plus Bernstein/Citi/Digital Assets Symposium); SEC EDGAR XBRL; public market-data sources (reconciled to filings). Cross-read: public Coinbase Global (COIN) FY2025 disclosures.

This is an independent fundamental research note. The body contains no investment recommendation and no price target; valuation is discussed solely as embedded expectations and scenarios. The Claude's Take block above is a separately-labeled subjective opinion and general information only — not investment advice.

APPENDIX A — Standard Diligence Questionnaire

Circle Internet Group, Inc. (NYSE: CRCL) | Report date: 2026-06-10

Supplemental diligence questionnaire. Fact/Interpretation labels where material.

General

What thoughtful questions have other investors asked about this company? The recurring institutional questions, visible across the Q4 2025 / Q1 2026 calls and sell-side notes: (1) Is the RLDC margin sustainable, and what drives it? — management concedes it has “many puts and takes,” with on-platform mix (Coinbase) and other-revenue growth the swing factors. (2) How much USDC usage is real economic activity vs. trading/arbitrage? — management points to Visa commercial-transaction analytics (~60% commercial share) and CPN B2B flows, but the honest answer is a meaningful share is still exchange collateral and trading. (3) What does the Arc token do for Circle’s financials? — deferred to a future call. (4) Rate sensitivity — quantified in the 10-K. (5) Coinbase renegotiation — the elephant; largely unaddressed publicly. (INTERP)

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? Near a rate-cyclical high on the margin axis (reserve return rate fell 5.0%→3.5% and is likely to fall further if the Fed cuts) but a volume-cyclical mid-point (float at record $77B but flat sequentially amid a ~45% crypto drawdown). Net: earnings power is more likely to compress than expand near-term if rates fall. (INTERP)

Driven by external environment or internal actions? Overwhelmingly external — ~96% of revenue is the Fed funds rate applied to a float whose size is driven by crypto-market and adoption cycles. Internal actions (distribution deals, product launches) move the share of economics and the long-run float, not the rate. (FACT/INTERP)

How stable are revenues? Structurally volatile: revenue = float × rate, both exogenous. Float churns enormously (~$150B minted/redeemed per quarter vs. $77B balance). (FACT)

Outlook for products/services? Core USDC float guided to ~40% CAGR through-cycle; other revenue guided to $150–170M FY2026; Arc/CPN/agentic are the growth options. (FACT)

How big will this market be? Stablecoin market grew ~32% YoY; bulls project trillions as payments, collateral, and machine commerce migrate on-chain. Large, growing, global — but the profit pool is capped by the rate spread and the distribution tax. (FACT/INTERP)

Business Quality & Competitive Moat

Is the industry getting more or less competitive? More — GENIUS legitimizes the category and explicitly invites banks/bigtech; Stripe/Bridge, PYUSD, RLUSD, tokenized MMFs, and Coinbase’s own ambitions all loom. (FACT)

How profitable is the business (ROIC, ROE)? GAAP loss in FY2025 (IPO SBC); normalized net income ~$300–380M on ~$2.7B tangible equity → normalized ROTE ~12–14%. Conventional ROIC not yet meaningful post-IPO recap. (INTERP)

How profitable is the industry — competitors, barriers? A duopoly (USDT+USDC ~90%+) with high today margins (54% Adj. EBITDA on net revenue) but commodity product, near-zero holder switching costs, and competition that bids the spread away to distributors. Barriers to global-scale operation (licensing, banking, liquidity) are real; barriers to issuing a coin are low post-GENIUS. (FACT/INTERP)

Can the business be easily understood? Yes — it is a narrow bank earning NII on a stablecoin float, paying out ~60% to distributors. The complexity is in the platform roadmap and the Coinbase contract mechanics. (INTERP)

Undermined by foreign low-cost labor? Not applicable. The competitive threat is regulatory/distribution, not labor cost. (INTERP)

Do brands matter? Moderately — “USDC the regulated standard” carries institutional trust, but the trademark is licensed to Coinbase with an assignment trigger. (FACT/INTERP)

Nature of competition? Distribution-based (paying exchanges/wallets/fintechs to carry the token) and regulatory (compliance posture), not product or price. (INTERP)

Customers’ switching costs? Near-zero for holders (redeem and re-mint); higher for integrated distributors — but Circle pays heavily for that stickiness. (FACT)

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? The USDC brand/network and the platform (Arc/CPN) option value are not capitalized — genuine off-balance-sheet value if they monetize. (INTERP)

Off-balance-sheet liabilities? The ~$75–77B of stablecoin deposit liabilities are on the balance sheet (matched by segregated reserves). The unreserved ~$133M FT Partners claim is a contingent liability. (FACT)

How conservative is the accounting? Reserves are conservatively held (~88% in a 2a-7 government MMF, ≤3-month duration). Revenue recognition is straightforward. The main aggressive presentation is the Adjusted EBITDA add-back of recurring SBC. (FACT/INTERP)

How CapEx-hungry? Light — ~$70M/yr, mostly capitalized software. Asset-light model. (FACT)

Capital Allocation & Management

How much FCF, and how is it used? Corporate OCF $542M FY2025; capex ~$70M; no dividends, no buybacks — all retained for growth/M&A. (FACT)

Significant recent acquisitions? Hashnote (~$100M, USYC), Malachite ($15M), Centre 50% buyout from Coinbase ($209.9M, 2023). Small relative to enterprise value. (FACT)

Buying back shares? No repurchase program. (FACT)

Issuing large amounts of stock to insiders? Yes — ~$516M one-time IPO RSU vesting; ongoing SBC ~$200M+/yr; ~27M unvested awards + ~45M reserved. Recurring dilution. (FACT)

Compensation policy of directors/management? STIP = 70% Adjusted EBITDA / 30% growth milestones; equity vests on time + a one-time liquidity condition. No TSR, ROIC, or per-share metric — incentives reward scale, not per-share value. CEO total $16.3M (notable $4.1M security perk). (FACT)

Motivations of management? Founder-led, vision-driven (Allaire), pursuing a platform/“internet financial system” ambition. Aligned with growth and the long-run franchise; less obviously aligned with near-term per-share capital discipline. Heavy insider monetization (~$1.6B) at the IPO/follow-on and steady post-lockup selling temper the alignment read. (INTERP)

Valuation & Market Data

ADR, MLP, or K-1 issuer? No — US C-corp, common stock, three classes (A/B/C). (FACT)

Dividend policy? None; none anticipated. (FACT)

How profitable is the business? High-margin on net revenue today (54% Adj. EBITDA); GAAP loss FY2025 on IPO SBC; normalized net income ~$300–380M. (FACT/INTERP)

Is net income diverging from cash from operations? Yes, and instructively: GAAP net loss vs. $542M OCF, because of non-cash SBC and the geography of reserve flows (financing). OCF is the cleaner read of corporate economics, but note ~$270M of RSU tax-withholding cash sits in financing. (FACT)

Risks & Downside

What would cause the stock to decline? Fed rate cuts (the dominant lever), an adverse Coinbase renegotiation, a crypto-market downturn shrinking the float, new bank/bigtech entrants, Arc/CPN failing to monetize, continued insider supply, or an adverse FT Partners outcome. (FACT/INTERP)

Risk of catastrophic loss? A total loss is unlikely (debt-free, ~$2.3B net cash, fully-collateralized pass-through reserves, real franchise). A severe equity impairment is plausible from sustained rate cuts + adverse Coinbase terms against a full multiple. The reserve/depeg tail (2023 SVB precedent) is low-probability, high-severity. (INTERP)

Chance of a total loss? Low. (INTERP)

Recent News & Events

Has the business environment changed recently? Yes, materially over 24 months: IPO + follow-on (2025), GENIUS Act (July 2025), OCC trust approval (Dec 2025), Arc testnet → token presale (May 2026), falling rates, and a ~74% stock drawdown. (FACT)

Significant acquisitions? Hashnote (Jan 2025). (FACT)

Change in accounting policies? Adjusted operating expenses redefined from Q1 2026 (excludes SBC payroll tax and certain one-time items). (FACT)

Recent changes — new markets, facilities, management? New products (CPN, StableFX, Agent Stack, cirBTC planned), Korea/EM expansion, CCO promotion (Razzaghi, Sept 2025), Heath Tarbert President/CLO. (FACT)

APPENDIX B — Source Appendix

Circle Internet Group, Inc. (NYSE: CRCL) | Report date: 2026-06-10

Primary sources prioritized. All SEC filings accessed via EDGAR (CIK 0001876042). Earnings-call and conference transcripts from company investor relations and public transcript sources. Accessed 2026-06-10/11.

Primary — SEC filings (Circle Internet Group, Inc.)

Primary — Earnings-call & conference transcripts

Primary — quantitative data

Secondary

Notes on data reliability

• Economic share count (248.6M = 229.9M Class A + 18.7M Class B) reconciled to the Q1 2026 10-Q cover; diluted 266.7M per the 10-Q EPS footnote.
• Management commentary (calls, conferences) is treated as a hypothesis and validated against filings and financials.