Circle (CRCL) — The Token Sale Made the Quarter. The Distribution Bill Sets the Value.
Circle is worth $55.95 a share on our model, against a $75.22 close. The third quarter's revenue growth was a one-time token sale: without it, revenue fell 7.7% and operating income was $26.5 million. The value turns on how reserve income is split with the partners that distribute USDC. Binance's fee moves that split against Circle from the fourth quarter; a higher Fed path and the GENIUS rules work the other way. The price needs USDC to grow about 35% a year from 2028.
In July we valued Circle at $50 and called retention the decisive variable: the share of reserve income Circle keeps after paying the platforms that hold USDC for their users. Four months on, the stock is at $75.22, third-quarter revenue rose 17%, and the share paid away fell slightly, to 62.2%. Both readings are incomplete. Most of the growth was a sale of ARC tokens that happens once, and the company's guide says the fourth quarter brings a step down in what Circle keeps, as Binance's new fee runs for a full quarter. Against that, the Fed raised rates in September and projects 4.1% through 2027, which lifts the income on USDC's reserves. This note re-prices Circle on all three.
What this note answers
- What did the quarter earn once the token sale is taken out?
- How is reserve income now split between Circle and its distributors, and what does the fourth-quarter guide imply?
- What are USDC's float and the rate path likely to do, and what is the ARC position worth?
- What is Circle worth after the print, and why does the number move from July's $50?
- What does the $75.22 price require?
- What would change the view, and when do we find out?
Terms used throughout, defined once. USDC: Circle's US-dollar stablecoin, a token redeemable one for one in dollars and backed by cash and short-term Treasuries. Reserve income: the interest Circle earns on those reserves. Reserve return rate: reserve income as a yield on average USDC. Distribution ratio: distribution and transaction costs as a share of reserve income, the part paid to Coinbase, Binance and other partners that hold or promote USDC. RLDC: revenue less distribution costs, Circle's measure of what it keeps. ARC: the token of Arc, Circle's own blockchain; the presale sold 807.5 million of 10 billion tokens to institutions at $0.30. Proof of stake: a system in which token holders lock up tokens to run the network and earn rewards. GENIUS Act: the 2025 US law that regulates payment stablecoins; its rules take effect on 18 January 2027. Cost of equity: the annual return a shareholder should require. FCFE: free cash flow to equity.
The case for the price, in the company's terms
The strongest reading of this quarter comes from Circle's own filings and call. It goes like this.
- USDC is winning the use that matters. USDC carried 79% of all stablecoin onchain transaction volume in the third quarter, from 66% a year earlier, on about 27% of supply. Onchain volume rose 49% to $13.9 trillion, a record $208 billion was minted and redeemed, and meaningful wallets rose 16%. On management's argument, balances follow usage, so circulation lags but follows.
- The rules arrive in January, and they came out friendlier. The GENIUS rules take effect on 18 January 2027. The OCC's final rule of 21 October dropped the proposed presumption that partner reward arrangements pay prohibited yield. Management says large companies and banks have been waiting for exactly this before building on digital dollars.
- Distribution is being bought on purpose, at the best venues. Binance (more than 300 million registered users) signed for five years and invested $100 million in Circle stock. Under the Hyperliquid arrangement, that venue's USDC balances sit on Coinbase's platform. Circle calls these investments designed to grow circulation and RLDC dollars over time, and paid no upfront cost for Binance.
- The rest of the platform is scaling fast. Circle Payments Network (CPN) reached $41.3 billion of annualized payment volume at quarter-end and more than $58 billion by 30 October, from $23 billion at 31 July. Tazapay adds $25 billion of volume and payout rails in more than 100 markets. Management expects material CPN revenue in 2027.
- Arc is live, and Circle owns a quarter of it. Mainnet launched on 16 September with 11 founding validators and more than 100 institutions and applications; about $1.5 billion of USDC, cirBTC and EURC sat on Arc by late October. The presale valued the network at $3 billion. Circle's 25% allocation, 2.5 billion tokens, is carried at nil under GAAP and would be worth $750 million at the presale price.
- Rates went up, not down. The Fed raised its target in September, and its projections put the policy rate at 4.1% through 2027. On Circle's filed sensitivity, each 100 basis points on the reserve yield is worth about $368 million a year to Circle after distribution.
What this note tests, and what it cannot. It tests claims 2, 3, 5 and 6 directly: the rule text, the fourth-quarter guide's arithmetic on distribution, the ARC terms in the 10-Q, and the rate path against Circle's own sensitivity. It tests claim 1 against the float, which is what earns reserve income, and claim 4 against the revenue line it feeds. It cannot test whether usage turns into balances on any timetable: no filed data link the two, and the 40% multi-year growth framework the company keeps is not a forecast with a date. The bull case carries that claim at close to the company's pace.
What the quarter earned
Total revenue and reserve income was $864.9 million, up 17% on the year. Of that, $181.8 million was ARC presale revenue: about 75% of the $242.2 million presale, recognized as the tokens were minted and delivered and the public mainnet launched. Excluding it, revenue was $683.1 million, down 7.7%. Reserve income fell 8% to $657.6 million: average USDC rose 8% to $73.3 billion, but the reserve return rate fell 61 basis points to 3.5%. Other revenue excluding ARC was $25.5 million, down 11%, after $41.6 million in the first quarter and $33.6 million in the second; the company cited fewer blockchain integrations and lower subscription and services revenue.
The margin expansion is the presale. RLDC was $453.3 million, a 52.4% margin, which the release shows as 1,291 basis points up on the year. Excluding the presale, which carries no distribution cost, RLDC was $271.5 million, a 39.7% margin, against 39.5% a year earlier. Operating expenses rose 16% to $244.9 million, so operating income of $208.3 million becomes $26.5 million without the presale, against $81.0 million a year earlier, and that is before any cost the presale itself carried.
Net income was taxed at about 1%. Income tax was $2.2 million on $228.2 million of pre-tax income: federal and state loss carryforwards and stock-compensation deductions shelter US income, and Circle keeps a valuation allowance on some US tax assets. Net income was $226.0 million, $0.84 a diluted share. Other income fell to $19.9 million from $72.1 million, which a year ago included a $56.2 million fair-value gain. We value every year at a 25% tax rate. The federal loss carryforward ($478.8 million at the end of 2025) would add at most $0.36 a share at the 21% federal rate, less whatever 2026's income has used, so we leave it out.
Adjusted EBITDA has two definitions, and the presale dominates both. Adjusted EBITDA was $316.7 million, up 85%. Since the first quarter it also excludes employer payroll taxes on stock compensation, which adds $2.7 million this quarter ($313.9 million on the prior definition) and $5.0 million a year ago. That change is small; the presale is not: excluding it, adjusted EBITDA was $134.9 million, down 21% on the year.
Exhibit 1 — Where reserve income went, six quarters
How to read it. Each bar is one quarter's revenue in millions of dollars, Q3-2025 to Q3-2026, and the fourth quarter as the company's guide implies it (pale). From the bottom: grey is reserve income paid to distributors, dark gold is reserve income Circle kept, light gold is other revenue excluding ARC, black is the ARC presale. The label above each bar is the distribution ratio. Compare the dark gold segments: what Circle kept from reserves was flat to lower through the year, and the presale is the only new block.
The split, the float and the rate
The fourth-quarter guide says the split moves against Circle. The company kept its full-year guide: other revenue $310–330 million and an RLDC margin of 41.7–43.7%, both including the presale. The CFO added the figures that matter. Excluding ARC, the full-year margin should land "near the midpoint of our original guidance range" of 38–40%, and Q4 is the first full quarter of both the Hyperliquid arrangement and the renewed Binance partnership. He pointed analysts to "the step down that we anticipate in Q4". Nine-month RLDC margin excluding ARC was 40.8%. With fourth-quarter reserve income of about $680 million (average USDC of $72.5 billion at a 3.75% reserve rate) and other revenue at the midpoint of the $130–150 million full-year range excluding ARC, a 39% full year needs a fourth-quarter margin of about 34%. That is a distribution ratio of about 69.6%, against 62.2% in the third quarter; at 40% it is 65.5%, at 38% 73.7%. Put another way: holding the third quarter's terms and allowing for the higher rate, Circle would keep about $233 million a year more than the guide implies, about 32 basis points on the float. The guide was restated five weeks into the quarter, with the cause named.
Coinbase renewed on the same terms; the cost comes through other partners. Payments to Coinbase were $331.3 million, 50.4% of reserve income (50.8% a year earlier), and distribution costs fell 8.5% on the year on lower reserve income. USDC on Circle's own platform rose 36% to $13.9 billion, 18.1% of circulation on a daily average (+462 basis points), and Circle keeps all the reserve income on those balances. That mix did not lower the ratio by much in the third quarter, and the guide adds a new payer in the fourth. July's bear case put the ratio at 68% partly on a worse Coinbase renewal. Coinbase renewed on the same terms (its own third-quarter 10-Q says so; under Circle's agreement a renewal runs three years, which would run to August 2029), yet the fourth-quarter ratio our arithmetic on the guide implies is above July's bear.
What we use. The base case holds the ratio at 69.6% at a 3.75% reserve rate: the fourth-quarter ratio our arithmetic on the guide implies, held flat, because Binance's fee is in the run-rate from that quarter and no filed figure or guide puts 2027 below it. Circle argues that the new agreements grow circulation and RLDC dollars, and its own-platform share rose 462 basis points in a year. For a year, though, new incentive arrangements have offset that mix: the ratio moved only from 62.9% to 62.2%, and the 10-Q says Circle expects distribution expense to increase as it adds distributors. So the base stays flat through and after the 2029 renewal, and the bull case carries the improvement: it starts at 65.5% (the 40% end of the guide) and reaches 62% by 2031 as balances grow outside incentive programs. The bear uses 72% (near the 38% end) and adds two points at the 2029 renewal.
What Circle keeps when rates move. Circle's filed sensitivity says 100 basis points on the reserve yield changes reserve income by $739 million and distribution costs by $371 million, so Circle keeps about half of any rate change, more than the 37.8% of reserve income it kept on average in the third quarter. The model applies that split year by year. It is measured on today's contracts, and the Coinbase and Binance terms renew in 2029 and 2031: if Circle keeps only 37.8% of a rate change from 2030, fair value is $58.70; at 61.8%, $53.35. The effect is small, and it runs against intuition because the long-run reserve rate of about 3.07% (the 3.2% policy rate less the 13bp gap) sits below the 3.75% reference: a smaller share of a rate change is also a smaller share of the fall.
The rate path is higher than July assumed. July's base case used a 3.0% reserve rate for 2027 on expected Fed cuts. Instead, the Fed raised its target to 3.75–4.00% in September and held it in October, with two members voting for another rise; the median projection is 4.1% at the end of 2026 and 2027, 3.9% in 2028, 3.6% in 2029 and 3.2% in the long run. Circle's reserve rate has run about 13 basis points below effective fed funds (3.54% against 3.67% in the third quarter), so the base case earns 3.97% in 2027 and 3.07% from 2030. The bear uses the low end of the Fed's projections and the bull the top of its central tendency.
The float is flat, and October gave back the summer. USDC ended September at $73.9 billion, flat on the year, and its share of US-dollar stablecoins fell 183 basis points to 27%. It rose in August and September, then fell to $71.9 billion at the end of October and $71.3bn on 3 November, below June. All US-dollar stablecoins on DeFiLlama's count are about $295 billion, down 3% in twelve months; Circle's 27% share uses a narrower CoinMarketCap set. Open USD, the consortium coin that Coinbase backs, had about $0.9 billion outstanding in early November, 0.3% of the market; Circle's 10-Q does not mention it. The company's 40% multi-year growth framework is unchanged, and the CFO said growth "will not follow a straight line".
Exhibit 2 — USDC in circulation, month-ends, and the 2027 averages we use
How to read it. The gold line is USDC in circulation at each month-end in billions of dollars, December 2024 to October 2026. The bars at the right are the average USDC each scenario carries for 2027, with July's base case ($95 billion) dashed for comparison. Read the right edge against the line: the base case asks for modest growth from today's level, and July's base sits above any month-end so far.
How long the float grows faster than the economy. Each stage is set by its driver. The adoption stage runs from 2028 through 2031: the GENIUS rules start in January 2027, and the five-year Binance term, the longest distribution term now in force, runs to September 2031. The rates differ by case. USDC's own history gives the range: from its June 2022 peak ($55 billion) to its March 2026 peak ($80 billion) it grew about 10% a year, the run from $24 billion at the end of 2023 was far faster, and the last twelve months were flat. The base case uses 18% a year, roughly the market doubling by 2031 with USDC's share held; the bull uses 35%, close to management's framework; the bear uses 4% and stops at the 2029 Coinbase renewal. Growth then fades in a straight line to the economy's rate by 2038 (2033 in the bear case).
| Driver | Stage | Years | What sets the length | Evidence |
|---|---|---|---|---|
| USDC in circulation | Base year | 2027 | the first year all terms now in force run for twelve months | $71.3bn on 3 Nov; base $77bn, bear $65bn, bull $92bn average |
| USDC in circulation | Adoption | 2028–2031 | the end of the Binance term (Sep 2031), the longest distribution term in force; bear to the Aug 2029 Coinbase renewal | peak to peak about 10%/yr; last 12 months 0%; company framework 40%; base 18%, bull 35%, bear 4% |
| USDC in circulation | Convergence | 2032–2038 | straight line to 4.0% (3.0% bear) | above-median growth rarely persists beyond a decade (Chan, Karceski & Lakonishok, 2003) |
| Reserve return rate | Fed path | 2027–2029 | the Fed's dated medians, 4.1% / 3.9% / 3.6%, less 13bp | September 2026 projections |
| Reserve return rate | Long run | 2030 on | the Fed's 3.2% long-run median | base 3.07%, bear 2.77%, bull 3.47% |
| Distribution ratio | Terms in force | 2027–2029 | the Coinbase term to Aug 2029; Binance in the run-rate | base 69.6% at a 3.75% rate (the fourth-quarter guide); bear 72%; bull 65.5% |
| Distribution ratio | After the renewal | 2030 on | the mix of balances outside incentive programs | bear 74%; bull 62% by 2031; base flat |
| Other revenue ex-ARC | CPN and Tazapay | 2027–2031 | the Tazapay close (2027) to the end of adoption | FY26 $130–150M; base $200M then 25%/yr |
| Costs | Investment | 2027–2031 | the same stage as the revenue they pay for | FY26 GAAP opex guide $973–1,056M; base $1,060M then 8%/yr |
The ARC position is its own asset. A token sale is revenue when recognized, and it does not recur: the rest of the presale ($60.4 million) sits in deferred revenue against an obligation the 10-Q ties to Arc's move to proof of stake, and further sales depend on Circle choosing to sell. So we take ARC out of every earnings run-rate and value the position once. Circle's allocation is 25% of the 10 billion tokens; the whitepaper puts 60% in an ecosystem pool that covers token sales, grants and growth programs, so we treat the presale as coming from that pool. The base case values Circle's 2.5 billion tokens plus 0.5 billion of future ecosystem sales at the $0.30 presale price, discounted three years at the cost of equity (proof of stake in 2027, then at least a year of lock-up by analogy with the presale buyers' terms) and taxed at 25% on a nil basis, less a 10% chance that buyers use their right to be repaid if Arc has not moved to proof of stake by 8 May 2028: $444 million. The mark is not independent: an affiliate of IDG Capital, a holder of more than 5% of Circle, bought $25 million of the presale, and the token does not trade. The bear values the tokens at zero and assumes half the presale is repaid (−$121 million); the bull values 4 billion tokens at $1.00 ($2,079 million). Staking rewards, which the CFO named as Circle's main expected recurring revenue from Arc, are paid in ARC, so they sit in the token's value, not a separate line.
What Circle is worth after the print, and why the number moved
How we value it. Circle earns a spread on money that is not its own: the $73.6 billion of reserve cash and the $73.5 billion owed to USDC holders match each other and sit outside capital, and most of its $3.86 billion of equity is corporate liquidity of $2.97 billion. So we value free cash flow to equity driver by driver from 2027 to 2040, with growth landing on 4.0% (3.0% in the bear case), the Fed's long-run projections of 2.0% real growth plus 2.0% inflation, and count the liquidity once, at face. Operating capital is what remains, about $0.9 billion, including $0.7 billion of goodwill and intangibles from acquisitions, which count because they were bought.
| Bear | Base | Bull | |
|---|---|---|---|
| 2027 average USDC | $65bn | $77bn | $92bn |
| USDC growth, 2028–2031 | 4% (to 2029) | 18% | 35% |
| Policy rate 2027 / long run | 3.1% / 2.9% | 4.1% / 3.2% | 4.4% / 3.6% |
| Distribution ratio at a 3.75% rate | 72% → 74% | 69.6% | 65.5% → 62% |
| 2027 other revenue ex-ARC | $130M | $200M | $330M |
| 2027 costs (GAAP, stock pay included) | $950M, cut to break even from 2028 | $1,060M | $1,150M |
| 2027 net income | −$390 million | $77 million | $456 million |
| USDC in 2031 | $75bn | $149bn | $306bn |
| ARC leg | −$121 million | $444 million | $2,079 million |
| Value a share | $9.71 | $30.75 | $152.60 |
The bear case is close to the liquidity floor of $10.75 a share: it loses money in 2027, and we assume management then cuts costs to stay near break-even, as it did with a 5% workforce reduction in the third quarter. The bull case caps the operating margin at 60% of RLDC, between Mastercard's 57.6% and Visa's 61.2% operating margins in their latest fiscal years; it binds from 2031. At 67% fair value is $59.84; with no cap, $65.75. Terminal value is 60% of the base case and 55% of the bull's operating value.
The cost of equity. The rate is a judgment about Circle's risk now. Sixteen monthly returns since the June 2025 listing give an own beta of 1.07 with a standard error of 2.07. Taken at face value that gives about 9.8%, close to the rate the price implies, but a standard error that wide cannot tell a beta of 1 from a beta of 3, and the decline after the IPO dominates it. So the peers set the read, and choosing them is the judgment. Reserve income behaves like a rate-sensitive float, which points to payments and broker reads (Visa 0.79, Interactive Brokers 1.33, PayPal 1.38); balances that move with the crypto cycle and high fixed costs point to Coinbase (3.17), Robinhood (2.37) and Block (2.56). The six-name median of 1.87 gives 13.4% on a 4.95% ten-year Treasury yield and a 4.5% equity risk premium. We use a beta of 1.8, slightly under it, because income tied to short-term rates argues against the top of the crypto-side reads: 13.0%. The terminal rate is 10.4%, a beta of 1.2, at the top of the stable range, because a mature USDC still carries rate and crypto-cycle exposure. July stated no rate; its multiple implied about 15.9%, a beta near 2.4, between Robinhood's 2.37 and Block's 2.56. The peer median gives $54.88; one point either way on the explicit rate gives $58.79 and $53.32.
The weights. We keep 25 / 50 / 25, as a decision on its own condition. The evidence since July pulls both ways. The float was flat and the split moved against Circle, which is the bear case's direction; the rate path rose and the GENIUS rules arrive in January with a friendlier final text, which is the bull's. Nothing observable today favors one tail over the other, so equal tails stand. A tilt of five points of weight toward either tail gives $48.81 or $63.10; the tripwires below say what would move them.
From July's $50. July's engine, re-run unchanged, gives $49.76 (published as $50): 2027 net income at a 25% tax rate times 35 (base) or 38 (bull), plus corporate liquidity, with the bear at 2.5 times gross profit, weighted 25 / 50 / 25 on 285 million shares. Exhibit 3 moves one input at a time to today's value on that engine. Liquidity at 30 September, including the presale cash and Binance's $100 million, adds +$1.67. A smaller float takes away $26.04 (2027 average USDC of $65 / $77 / $92 billion, against July's $60 / $95 / $150 billion), and the Fed's higher path adds +$20.82. The distribution ratio with Binance and Hyperliquid in the run-rate takes away $6.00. Other revenue excluding ARC takes away $11.14: July's base carried $320 million for 2027 against an annualized $102 million today. Costs add +$2.50, and the ARC position adds +$2.50. On July's method, the quarter's data give $34.06.
Then the method. July used 285 million shares to allow for dilution while also counting stock pay as a cost, which charges it twice; today's diluted count plus the Tazapay stock, 276.2 million, with stock pay as a cost, adds +$1.08. On our engine July's whole method equals a flat cost of equity of 15.9%: a multiple of 35 on earnings this close to a trough is the same as requiring 15.9% a year from the stage model's cash flows. At that rate the switch of engine is neutral by construction, so the change from July is the rate. Moving it to the argued 13.0% in every year adds +$10.45, and letting it converge to 10.4% from 2031 to 2038 as USDC matures adds +$10.36. The result is $55.95.
Exhibit 3 — From July's $50 to today's value, one factor at a time
How to read it. The first bar is July's fair value a share, re-run on its own engine; the last is today's. Each bar between is the change in dollars a share from moving one factor to today's value, in the order shown: gold adds, grey subtracts. The bars between the dotted lines are the quarter's data on July's method; the pale black bars after them are the method changes and the cost of equity. The dashed line is the $75.22 price. The data steps lower the value; the engine change is neutral at the rate July's multiple implies, and the two cost-of-equity steps raise it.
| Step | Kind | $ a share | Fair value after |
|---|---|---|---|
| Corporate liquidity at 30 Sep (cash, corporate-held USDC, investments, digital assets), including the presale and Binance cash | data | +1.67 | 51.43 |
| Average USDC, FY2027 (bear $65bn / base $77bn / bull $92bn, from $60bn / $95bn / $150bn) | data | -26.04 | 25.39 |
| Reserve return rate, FY2027, on the Fed's September path (2.97% / 3.97% / 4.27%, from 2.75% / 3.00% / 3.25%) | data | +20.82 | 46.20 |
| Distribution ratio, FY2027 at each case's 2027 rate, with Binance and Hyperliquid in the run-rate (77.7% / 68.5% / 63.6%, from 68% / 65% / 63%) | data | -6.00 | 40.20 |
| Other revenue excluding ARC, FY2027 ($130M / $200M / $330M, from $180M / $320M / $550M) | data | -11.14 | 29.07 |
| Costs, FY2027 ($950M / $1,060M / $1,150M, from $900M / $1,060M / $1,250M) | data | +2.50 | 31.57 |
| ARC position as its own leg, net of the repayment right (new asset; July had no lever) | data | +2.50 | 34.06 |
| Shares: today's diluted count plus Tazapay (276.2M) with no forward dilution, because stock pay is already an expense (from 285M) | method | +1.08 | 35.15 |
| Method: multiple of FY2027 earnings replaced by stage FCFE, 2027-2040, at the flat 15.9% rate July's multiple implies on this engine (neutral by construction); bear moves from a gross-profit multiple to the same engine | method | +0.00 | 35.15 |
| Cost of equity, explicit years: from the 15.9% July's multiple implies to the argued 13.0%, flat in every year (July stated no rate) | cost of equity | +10.45 | 45.59 |
| Cost of equity, terminal: 13.0% converging to 10.4% from 2031 to 2038 as USDC matures | cost of equity | +10.36 | 55.95 |
| Weights: 25 / 50 / 25 kept, argued on their own condition (no change) | weighting | +0.00 | 55.95 |
| Fair value 49.76 → 55.95 | +6.20 | 55.95 |
Each row re-runs the same engine with one more factor at today's value, in the order shown. July's engine is run unchanged ($49.76, published as $50). Rows sum to the move. Kinds: data (filed figures, the Fed's September path and the 3 November close, on July's 35 / 38× multiples and the bear case's 2.5× gross profit), method (the share count with stock pay charged once, and the change from a multiple to the stage model at the flat rate July's multiple implies, neutral by construction), cost of equity (two rows: to the argued explicit rate, then its convergence to the terminal rate) and weighting (unchanged). The workbook's Attribution sheet carries every row.
What the price requires. At $75.22, the weighted value equals the price only at a flat cost of equity of 9.4%, about what an average stock with a beta of 1 requires. Holding our rates, the base case needs USDC to grow about 35% a year from 2028 to 2031, to about $256 billion, against 18% in our base case and about 10% a year from peak to peak. Or it needs the base case's distribution ratio at about 45%, a smaller share for partners than in any year Circle has reported (50.3% in 2023). Or the bull case needs about 41% of the weight, with the bear kept at 25%. Two routes come closer: if every case's adoption stage lasts two years longer, to 2033, the weighted value is $78.41, above the price; if the policy rate holds at 4.1% for good instead of falling to 3.2%, it is $71.84. In July's own frame, a 35-times multiple on 2027 earnings, the price needs about $122 billion of average USDC in 2027, 71% more than today; July's figure was $124 billion, and the float has not moved toward it.
Exhibit 4 — The base case's USDC against the path the price needs
How to read it. Each line is average USDC in circulation in billions of dollars, 2027 to 2035. Gold is our base case; the dashed grey line is the base case with its 2028–2031 growth raised until the base value equals the price, fading to 4% by 2038 as in the model; the dotted black line is the bull case. Read the gap between the gold and grey lines: it is the extra float the price assumes over the base case.
Relative check. At the price, Circle trades at about 271 times the base case's 2027 earnings and about 15 times its 2027 RLDC, net of liquidity. The median trailing price-to-earnings ratio of six listed payment and brokerage companies is 31.7; on base 2027 earnings that gives $21.16, but 2027 is a low year in every case, and at our terminal rate the multiple implies about 7.8% perpetual growth. We show it as a cross-check, not a weighted method.
Return on capital
Exhibit 5 — Return on capital against the cost of capital, five quarters
How to read it. The gold line is Circle's trailing-twelve-month return on operating capital at each quarter-end from September 2025 to September 2026: operating income excluding the ARC presale and the IPO stock charge, taxed at 25%, over average equity less corporate liquidity (reserve cash and holder deposits are excluded on both sides; goodwill and intangibles are included). The dashed black line is the 13.0% WACC, the test; the dotted grey line is the 13.0% cost of equity, a reference. The diamond at the right is the return on new capital the terminal value assumes (25%); the return earned on new capital over the window is not meaningful because capital grew under 5%. Read the gold line against the dashed line.
Circle's return on capital is falling toward its cost. It was 25.5% in September 2025, 30.0% in December and 14.4% in September 2026, against a WACC of 13.0%: a fading regime. The cause is operating income, not capital: capital stayed near $0.85 billion while quarterly operating income excluding ARC fell from $81 million to $27 million, as rates fell faster than the float grew and costs rose 16%. Including the presale, the September figure is 31%. If the fourth quarter lands on the guide's midpoint, the trailing return falls to about 8%, below the WACC. Five quarters are all the filings support, because equity less liquidity was negative before the June 2025 IPO. The WACC is the cost of equity to within a basis point, because Circle has no borrowing and its only debt-like claim is $17 million of leases. The terminal value assumes 25% on new capital, below the 2025–26 returns and well above the cost, because growing the float takes little capital but competition limits the spread.
The receipt: the triggers July left armed
July left four triggers. Two had closed on 31 August with the Coinbase renewal; one fired on its words in September and reversed in October; one has not fired on the reading we take.
Both legs were met by 30 September. The renewal leg: Coinbase's Q3 10-Q says it renewed its arrangement with Circle on the same terms; under the agreement a renewal runs three years, to August 2029. The supply leg: USDC rose in August ($72.1bn to $74.1bn) and September (to $74.6bn) on DeFiLlama, and CoinGecko agrees ($73.9bn, $74.4bn). October gave the gain back: $71.9bn at month-end, $71.3bn on 3 November, below the June level. The trigger named a rating change at July's $64.62; this note replaces it with a full rebuild, and on the rebuild the price sits above fair value. Retired.
Mainnet went live on 16 September and Circle booked $181.8M of ARC revenue, a fifth of the quarter's revenue. Read literally, that is Arc revenue at material scale. We read the trigger as written for its purpose, mainnet monetization: the $181.8M is about 75% of a $242.2M presale recognized as delivery milestones were met, a one-time sale of tokens, not fees, staking or other recurring network income, and the 10-Q discloses none of those. Scored as met on its words and not on its purpose, the same standard the renewal-plus-supply trigger is read on; the model is unchanged by it. Re-armed as crq3-arc-pos and crq3-arc-recurring.
Closed in our record on 31 August at the status-quo branch. Coinbase's Q3 10-Q says it renewed its arrangement with Circle on the same terms, and Circle's 10-K sets renewals at three years, so the next point is August 2029. Circle's Q3 10-Q puts Q3 payments to Coinbase at $331.3M, 50.4% of reserve income.
Closed on 31 August: the renewal resolved, so the condition could not be met. The stock's low after the July note was $57.84 intraday on 3 August; it did not reach the $50 fair value. The 52-week low of $49.90 was on 5 February 2026, before the July note.
July's other claims, re-tested. The $124 billion reverse solve: the bar is $122 billion today in the same frame, against a float that has fallen. Distribution at 63% of reserve income: 62.2% in the third quarter, about 69.6% on the fourth-quarter guide. Coinbase at about 51% of reserve income: 50.4% in the third quarter and 49.9% over nine months. Open USD contagion: the coin is about $0.9 billion and the Coinbase terms did not move, but Circle's distribution cost rose anyway, through Binance and Hyperliquid. The July addendum on the OCC charter: Circle National Trust opened on 24 July to offer custody to Circle and its affiliates, which touches none of the model's drivers. The stock did not reach the $50 fair value after the July note: its low was $57.84 intraday on 3 August. The 52-week low of $49.90 was on 5 February 2026, before the note.
What carries the story to the next update
- CRCL Q4-26 distribution ratio (by 2027-02-28). The base case holds distribution and transaction costs at 69.6% of reserve income at a 3.75% reserve rate, from the CFO's statement that full-year RLDC margin ex-ARC lands near 39%, the midpoint of the original 38-40% range: the ratio that guide implies for Q4 on our arithmetic. Reads: Q4-26 ratio <= 66.0% or >= 72.0% (rate-adjusted to 3.75% on the 10-Q's 50.2% marginal share). Silence: Between the bands = no change. A ratio moved by a one-off true-up the company names is read net of it. Model consequence: At or below 66%: base ratio 65.5%, FV $59.98. At or above 72%: base 72%, FV $53.58. Q3's 62.2% held in Q4: FV $63.22.
- CRCL average USDC in Q2-27 (by 2027-08-15). The base case averages $77bn in 2027 and grows 18% a year from 2028; GENIUS rules take effect on 18 January 2027. Reads: Q2-27 average USDC >= $85bn or <= $68bn. Silence: Between the bands = no change. A move in either direction carried by one partner's balances, a new deposit (as Hyperliquid's was in 2026) or a withdrawal, is read as one event: the base path changes only if the next quarter holds it. Model consequence: At or above $85bn: base FY27 $90bn and 25% a year, FV $69.81. At or below $68bn: $68bn and 10%, FV $47.11.
- CRCL FY2027 other-revenue guide excluding ARC (by 2027-02-28). The base case carries $200M of other revenue excluding ARC in 2027 (CPN, Tazapay after its close, integrations), from $130-150M in 2026. The CFO will guide 2027 at the Q4 call. Reads: FY2027 guide midpoint excluding ARC >= $300M or <= $150M. Silence: A guide that does not split out ARC is read with ARC token sales removed at the company's own split; if no split is given, the trigger waits for the Q1-27 10-Q. Model consequence: At or above $300M: FV $60.75. At or below $150M: FV $53.55.
- CRCL Arc moves to proof of stake by 8 May 2028 (by 2028-05-08). Presale buyers can ask for their money back ($242.2M gross) if Arc has not moved to proof of stake by 8 May 2028. The base case puts that at 10%, the bear at 50%. Reads: proof-of-stake transition completed (or repayment rights exercised). Silence: No transition by the date = repayment assumed exercised in full until the company says otherwise. Model consequence: Met: the repayment right falls away in every case (bear tokens still at zero), FV $56.11. Missed and repaid, with the token leg at zero in every case: FV $52.50.
- CRCL recurring Arc revenue (by 2028-12-31). The CFO called staking rewards Circle's primary expected recurring revenue from Arc once proof of stake arrives. The model values the token holding, not a revenue line, so a disclosed price for ARC is the first test. Reads: ARC trades at >= $0.60 or <= $0.15 for 30 days after the first unlock; or staking revenue disclosed at >= $50M a year. Silence: No trading and no disclosure = no change; the presale mark stands as the base. Model consequence: At $0.60 (bull $2.00): FV $58.68. At zero: see crq3-arc-pos.
- CRCL: the Fed's longer-run rate (by 2027-12-31). Reserve income follows the policy rate: the model uses the September 2026 projections (3.2% in the long run; reserve rate 13bp below). Each 100bp on the whole path moves fair value by about $25 a share. Reads: longer-run fed funds median <= 2.9% or >= 3.6%. Silence: Inside the band = no change; the path is re-read at each update. Model consequence: At 2.9%: FV $45.43. At 3.6%: FV $61.59.
- CRCL: what would change the view (by 2027-11-30). The price needs the base case's USDC to grow about 35% a year from 2028, or the bull case to carry about 41% of the weight. Both legs use the scenarios' own 2027 averages: the bull's $92bn and the bear's $65bn. Reads: toward: average USDC >= $92bn in any 2027 quarter with the distribution ratio <= 66%; away: average USDC <= $65bn in any 2027 quarter with the ratio >= 72%. Silence: Neither by the Q3-27 print = no change; the weights are re-argued at that update. Model consequence: Toward: weights 0/50/50, FV $91.68. Away: 50/50/0, FV $20.23.
- CRCL Binance arrangement (by 2031-09-17). Five-year term from 17 September 2026; a monthly incentive fee on USDC held through the Modular Smart Contract Wallet service. Either side can end it early on specified events. Reads: termination, amendment of the fee, or Binance-held USDC disclosed. Silence: Silence = the terms run as modeled. Model consequence: An early end that removes the Q4 step: base ratio back toward Q3's 62.2%, FV $63.22, before any loss of balances.
- CRCL: Coinbase terms at the August 2029 renewal (by 2029-08-31). Renewed in August 2026 on unchanged terms; the bear case adds two points of distribution cost at the 2029 renewal. Reads: an amendment to the revenue split. Silence: Silence through August 2029 = renewed on the same terms. Model consequence: A two-point rise in every case: FV $52.27.
What would change our view
The view moves toward the price if the float and the split improve together: average USDC at or above $92 billion in any 2027 quarter, the bull case's own 2027 average, with the distribution ratio at or below 66%. That is the bull case's path arriving on schedule, and half the weight would move to it, giving $91.68. The view moves away if both worsen together: average USDC at or below $65 billion, the bear case's own 2027 average, in any 2027 quarter with the ratio at or above 72%, which would put half the weight on the bear case, giving $20.23. Short of either, the single readings in the tripwires apply: the fourth-quarter ratio (settled at the February print), the 2027 other-revenue guide (same call), average USDC in the second quarter of 2027, and whether Arc reaches proof of stake by 8 May 2028.
The call
We value Circle at $55.95 a share, 26% below the $75.22 close, with a range of $9.71 to $152.60. The quarter's growth was a token sale, and the guide says the split of reserve income moves against Circle as Binance's fee runs in full. Against that, rates are higher than July assumed, the GENIUS rules arrive in January, Arc is live, and Circle holds a token position July could not value. Measured on July's method, the rise from $50 comes from the required return, not from the quarter: July's multiple implied about 15.9% a year, and we argue 13.0% falling to 10.4%. The price asks for USDC growth of about 35% a year for four years, roughly double our base case. The next decisive readings both come at the February print: the fourth-quarter distribution ratio and the 2027 guide.
Sources & Method
Filings. Coinbase Global's Q3-2026 10-Q (0001679788-26-000122), for the renewal of its Circle arrangement on the same terms. Circle Internet Group 8-K of 4 November 2026, Exhibit 99.1 release (accession 0001876042-26-000309): income statement, balance sheet, quarterly table, non-GAAP reconciliations, guidance. Form 10-Q for the quarter ended 30 September 2026 (0001876042-26-000311): ARC presale and token accounting (Notes 2, 5, 11), deferred revenue, payments to Coinbase, distribution-cost changes, interest-rate sensitivity (Item 3), Tazapay, OCC trust bank, subsequent events (MiCA significant e-money token classification, effective 12 November 2026). 8-Ks of 22 September (Binance, 0001876042-26-000276), 8 September (Tazapay, -000267, with the share purchase agreement), 25 September (CFO and board, -000279), 11 May and 2 July (ARC presale, -000148, -000205). Q1-26 and Q2-26 10-Qs and releases, Q2-25 and Q3-25 10-Qs and releases, the FY2025 10-K (0001876042-26-000062) and its 10-K/A of 13 July 2026 (-000228), which adds only the audited statements of the Circle Reserve Fund and touches no July input. ARC whitepaper (arc.io), for the 60 / 25 / 15 allocation.
Call. Circle's Q3-2026 call of 4 November, 8:00 AM ET, in three independent transcriptions (Motley Fool, gloom.sh, TradingKey); quoted words match in at least two.
Market and macro. Prices: Schwab and Robinhood (the 3 November settled close, $75.22, agrees in both). The print-day session closed at $76.12, and the stock closed at $69.48 on 5 November; neither is an input. Monthly closes for betas: Schwab. USDC and stablecoin supply: DeFiLlama and CoinGecko. Rates: FRED (effective fed funds, SOFR, Treasury yields to 3 November). Fed Summary of Economic Projections of 16 September 2026. OCC final GENIUS rule of 21 October 2026, as summarized by Jones Day. Peer price-to-earnings ratios: Schwab, 4 November. Visa and Mastercard operating margins: SEC XBRL company facts from their latest 10-Ks.
Base rates. Chan, Karceski and Lakonishok, "The Level and Persistence of Growth Rates", Journal of Finance (2003), for how long above-median growth lasts.
Method. Free cash flow to equity, because Circle's reserve cash and holder deposits match and no capital ratio scales its earnings with equity; a model of excess returns on tangible equity would treat corporate liquidity as earning equity, so liquidity is counted once, at face, with no interest income in the earnings. The horizon is 2027–2040, with mid-year discounting from 3 November 2026; reserve income = average USDC × (policy rate − 13bp); retained reserve income from the distribution ratio at a 3.75% reference rate plus 49.8% of any rate change (the 10-Q's sensitivity); other revenue excluding ARC; costs on a GAAP basis with stock pay expensed; 25% tax; reinvestment of half of each year's RLDC growth; terminal value at the stable rate with reinvestment of growth over a 25% return on new capital; liquidity at face; ARC as its own leg. Shares: 276.2 million (256.2 million outstanding, 14.7 million dilutive securities in the third-quarter count, 5.3 million for Tazapay at the anchor price). The model kit (TON618_CRCL_Model.xlsx) carries every input, the stage table, the yearly cash flows, the attribution, the sensitivities and the return-on-capital series.
Notes to the financial statements not modeled. The fee dispute with a former financial advisor (Note 22) is not sized by Circle and is not priced here. Circle also says the tax treatment of stablecoins is uncertain and its effect "is not estimable"; we do not price it. Warrants for up to 6.6 million shares at $22.71–42.14 (Note 15), most of them subject to performance conditions, would lower fair value by about $0.67 a share if all were exercised. Operating leases are $17.2 million (Note 4). Strategic investments are carried at $124.4 million and counted in liquidity at that value (Note 8).
Limitations. Binance's fee is not disclosed; the fourth-quarter ratio is our arithmetic on the company's guide and depends on our estimate of fourth-quarter reserve income. ARC does not trade and its unlock schedule is not published; an affiliate of a holder of more than 5% of Circle bought about 10% of the presale at the same $0.30, so the mark is not fully independent. Five quarters of return on capital cannot show a full cycle. The bull case's float is far above any USDC level to date; it rests on the company's framework, not on a filed forecast.
Disclosures
Information only. TON618 Capital. This report is for information purposes only. Nothing here is an offer to sell or a solicitation of an offer to buy any security, fund interest, or digital asset, and nothing here is personalized investment advice or a recommendation regarding any instrument.
Publisher's exclusion. All research is published solely as general, impersonal information of regular circulation. It is not tailored to the objectives or circumstances of any individual and is not issued in connection with compensation from any client. The Fund has no clients and distributes all research free of charge. On that basis it publishes in reliance on the publisher's exclusion from the definition of "investment adviser" under the Investment Advisers Act of 1940 (§202(a)(11)(D); cf. Lowe v. SEC, 472 U.S. 181 (1985)).
Registration & conflicts. TON618 Capital is not registered as an investment adviser or broker-dealer in any capacity. The Fund is a Bitcoin fund and may hold or transact in the securities or digital assets it discusses; material conflicts are disclosed where they exist. Ownership: the Fund holds no position, long or short, in CRCL or in any other security named in this report as of the report date, and has no economic interest in their price. Compensation: the Fund received no compensation from any party in connection with this report and charges nothing for it.
Use of AI. Artificial intelligence is used in the creation of this research. All methodology and data integrity are reviewed and approved before publication by TON618 Capital's Chief Investment Officer, Keyth Beck; errors may nonetheless occur, and readers should verify independently.
CFA. This report was prepared to align with CFA Institute analytical standards (methodology only). CFA® and Chartered Financial Analyst® are registered trademarks owned by CFA Institute. That reference describes the analytical framework applied; it does not imply the report was prepared, reviewed, or authored by a CFA charterholder, and the report is not issued, reviewed, endorsed, certified, or approved by — nor affiliated with — CFA Institute.
Risk & feedback. Past performance is not indicative of future results. Digital assets and equities are volatile and may result in total loss of capital. Corrections and feedback are welcome — please direct them to CIO Keyth Beck at keyth@ton618capital.com. Version 1.0.