Notice: Editorial analysis with quotes frozen at the close of September 2, 2026, and business figures taken from Q2 2026 reports filed with the SEC between August 5 and August 14, 2026. This does not constitute financial advice or a recommendation to buy or sell any security. CleanSky does not receive commissions or referral payments from any of the companies mentioned.
Circle (CRCL) closed on September 2, 2026, at $88.63, 186% above the $31.00 at which it priced its shares in June 2025; Gemini (GEMI) closed at $4.29 compared to its $28.00 offering price, down 85%. These are 271 percentage points of divergence within the same batch of crypto IPOs—the wave of public listings spanning from May 2025 to January 2026. Timing does not explain it either: Figure and Gemini debuted just one day apart, yet by the close of September 2, 2026, they were separated by 117 percentage points. The six listed companies fall into two blocks that do not align with who is making money: the two trading above their offering price earn from idle balances—reserve yields, credit—while the four trading below earn from executed transactions. The six rows—offering price, day one close, September 2, 2026 close, and primary revenue driver—follow below, along with the exact point where this distinction stops explaining everything.
How much is each 2025 crypto IPO worth compared to its offering price?
The table uses three prices per company: the offering price, found in the final 424B4 prospectus filed in EDGAR, the SEC's public archive; the first-day close, representing what those who entered at the debut paid; and the close on September 2, 2026. Percentages are calculated on both bases: for Gemini, they differ by two points.
| Company (Debut) | Offering ($) | Day 1 Close ($) | Sep 2, 2026 Close ($) | Change vs. Offering | Change vs. Day 1 Close | Primary Driver (% of declared base, Q2-2026) |
|---|---|---|---|---|---|---|
| Circle · CRCL (Jun 5, 2025) | 31.00 | 83.23 | 88.63 | +185.9% | +6.5% | Reserve interest income — 95.2% of total revenue |
| Figure · FIGR (Sep 11, 2025) | 25.00 | 31.11 | 33.19 | +32.8% | +6.7% | Aggregate credit items — 65.2% of total net income |
| Bullish · BLSH (Aug 13, 2025) | 37.00 | 68.00 | 32.64 | −11.8% | −52.0% | Subscription, services, and data — 67.7% of adjusted income |
| eToro · ETOR (May 14, 2025) | 52.00 | 67.00 | 31.85 | −38.8% | −52.5% | Stock, commodities, and FX trading — 61.7% of net contribution |
| BitGo · BTGO (Jan 22, 2026) | 18.00 | 18.49 | 7.07 | −60.7% | −61.8% | Digital asset sales — 97.0% of gross revenue |
| Gemini · GEMI (Sep 12, 2025) | 28.00 | 32.00 | 4.29 | −84.7% | −86.6% | Trading — 39.0% of total revenue |
Figure (FIGR) closed on September 2, 2026, at $33.19, 32.8% above its $25.00 offering. Bullish (BLSH) closed that day at $32.64, 11.8% below its August 2025 offering of $37.00. eToro (ETOR) closed at $31.85, 38.8% below its May 2025 price of $52.00. BitGo (BTGO), which debuted on January 22, 2026, and is therefore outside the 2025 cohort, closed at $7.07, 60.7% below its $18.00 offering.
Between the offering price and the first-day close, the signs do not change: the same two are up on both bases and the same four are down. Against the debut's opening price, one row does flip: Gemini has lost 84.7% from its offering and 88.4% from its $37.01 opening price, while Figure, up 32.8% from its offering, is down 7.8% for anyone who bought at its $36.00 opening.
Why do Circle and Figure charge based on balance (float) while Gemini, Bullish, eToro, and BitGo charge per transaction (take rate)?
A cloud storage service bills per gigabyte stored per month: as long as the files are there, the bill arrives even if no one opens them. An API bills per call; if the client stops invoking it, the amount drops to zero that same month. One depends on a balance; the other, on frequency.
The six listed companies are distributed along this line. Circle collects interest on the reserves backing USDC in circulation: $73.3 billion as of June 30, 2026, up 19% year-over-year. Figure collects interest, origination fees, and capital gains from selling the loans it originates. Gemini, Bullish, eToro, and BitGo charge when someone buys, sells, moves, or holds assets. The first model is called float: third-party money resting on the intermediary's balance sheet that generates a yield the intermediary keeps. The second is the take rate: the percentage the intermediary retains from every transaction passing through its platform.
The limitation of float is that its engine is external: it depends on interest rates and how much balance the client chooses to leave, and the former can change in a single central bank meeting. The limitation of the take rate is different: it is charged on a flow that can vanish in a single quarter. The consequence is visible in the July 2026 indicators published by eToro in its 6-K (the form used by foreign companies listed in the U.S. to report results): 1.4 million crypto transactions in the month, down 73% year-over-year, with an average transaction size of $182, down 50%. In a balance-based business, such a drop takes quarters to appear in the income statement; in a movement-based business, it appears on the top line the following month.
How much of Circle's reserve yield does Circle keep, and how much does Coinbase take?
Collecting the toll on the balance and retaining it are separate operations, and Circle's 10-Q quarterly report for Q2 2026 quantifies this unambiguously. Gross reserve yield was $667.7 million out of total revenue of $701.3 million: 95.2% of the business. From there, $410.4 million goes to distribution and transaction costs, and the document identifies the primary recipient: "during the three months ended June 30, 2026 and 2025, we incurred $324.6 million and $332.3 million, respectively, in distribution costs related to our agreements with Coinbase."
In the second quarter of 2026, Coinbase took 48.6% of the gross yield from USDC reserves without issuing the stablecoin or managing its reserves: it charges for the balance its clients maintain on its platform, according to the collaboration agreement described in the 10-Q itself (how the Circle-Coinbase contractual split works). Circle retained $257.3 million, or 38.5% of the $667.7 million gross reserve yield for Q2 2026. After $254.5 million in operating expenses, the operating income from continuing operations stands at $34.4 million, with a net income of $48.2 million. For every $100 in interest generated by USDC reserves, five dollars end up as Circle's operating profit.
The true measure of float appears when comparing Circle with the other company on the list that collects interest on client balances.
| Toll on Balance, Q2-2026 ($ millions) | Circle | BitGo |
|---|---|---|
| Interest-generating base | 73,300 (USDC in circulation) | 4,600 (segregated stablecoin reserves) |
| Gross income from that base | 667.7 | 38.8 |
| Distribution and transaction costs / sponsor issuer fee | 410.4 | 35.7 |
| Retained | 257.3 | 3.1 |
| Percentage retained | 38.5% | 8.0% |
BitGo has float—$4.6 billion in segregated client stablecoin reserves as of June 30, 2026, compared to $3.3 billion at the end of 2025—and its Stablecoin-as-a-Service line grew by 148.0% to $38.8 million. It retains 8.0% of what it generates; Circle retains 38.5%.
Why do BitGo's $4.329 billion in revenue end up as $42 million?
BitGo's 10-Q declares $4,329.4 million in total revenue for the quarter, up 79.6% year-over-year. The figure is correct yet incomparable to Circle's $701 million or Gemini's $45 million: $4,197.5 million of that total consists of gross digital asset sales, with a matching cost of $4,190.4 million. The margin for that activity in the quarter was $7.1 million.
Subtracting each cost from its corresponding revenue—digital asset sales, staking fees, and the stablecoin sponsor issuer fee—BitGo's business nets out to approximately $42.5 million for the quarter: $27.5 million from subscriptions and services, $7.1 million from asset sales, $3.9 million from staking, $3.1 million from stablecoins, and $0.8 million from interest. This is an internal calculation based on the breakdown in Note 3 of the 10-Q. A second figure, $139.0 million, subtracts only the cost of digital asset sales and leaves the other two matching costs in. With this real size, the net loss of $19.0 million and the adjusted EBITDA of −$4.2 million for the quarter weigh much more heavily on $42.5 million than on $4,329 million.
What does Gemini's 10-Q say about its trading revenue and transaction losses?
Gemini is the hardest hit of the six, and its income statement offers no relief. Total revenue grew to $45.5 million from $33.3 million, but trading revenue fell 14.8% year-over-year to $17.8 million: the exchange saw a 38.2% drop—from $20.2 million to $12.5 million—mostly driven by retail (−$7.9 million, a 42% decrease), partially offset by the OTC desk, which executes large bespoke trades outside the order book, growing from $0.6 million to $4.7 million. Trading now accounts for only 39% of revenue. What is growing is the credit card: $16.2 million compared to $4.9 million a year ago, a 3.3x increase—now 35.6% of the total.
The problem lies in the expense line. Transaction losses for the quarter were $20.147 million compared to $3.553 million in the same quarter of 2025: a $16.6 million increase, or 467%, according to the 10-Q table. These losses exceed the entire trading revenue for the period by $2.4 million, and the document further details a $4.1 million discrete fraud reserve for the first half of the year. With $122.4 million in operating expenses, the operating loss was $76.9 million and the net loss was $107.7 million. The credit card contributed $16.2 million in revenue for the quarter and $16.1 million in provision for expected credit losses.
Do the Bitcoin price drop or profitability explain why Circle rises and Gemini falls?
Three alternative explanations compete with the revenue breakdown, and all three fall short against the data from this window.
- Crypto market sensitivity (beta). Between September 2, 2025, and September 2, 2026, Bitcoin went from $111,201 to $77,350 (−30.4%) and Coinbase from $303.56 to $174.96 (−42.4%). This explains why listed exchanges are falling, but not why Circle and Figure are trading 186% and 33% above their offering prices.
- Profitability. Bullish reported a net loss of $280.0 million under IFRS (adjusted profit of $14.3 million) and is down 11.8%. Gemini lost $107.7 million and is down 84.7%. By net profit, five of the six rows fall on the expected side; only eToro—with $53 million in GAAP profit, up 77% year-over-year, a net contribution of $229 million, and 4.28 million funded accounts, yet trading 39% below its offering—breaks the pattern. The revenue breakdown orders all six, though with n = 6, the difference between the two explanations is a single row.
- Listing timing. Figure debuted on September 11, 2025, with Bitcoin at $115,508, and Gemini on September 12 with Bitcoin at $116,102: a 0.51% difference between the two benchmarks. One year later, they are separated by 117 percentage points relative to their respective offerings.
Why is Bullish down 12% while Gemini is down 85% if both live off trading?
The limit of the revenue breakdown distinction is significant. These are six companies: a sample size that describes a directional split and little else. It orders who is above and who is below the offering price, but not the magnitude within each block, where dispersion is massive.
Bullish is down 11.8% and Gemini is down 84.7%, yet both are trading platforms. However, only 32.3% of Bullish's adjusted revenue comes from trading: the other $62.7 million, out of $92.6 million in adjusted revenue, comes from subscription, services, and data—CoinDesk, indices, Consensus, the CoinDesk conference—a record figure that grew while digital asset sales fell from $58.6 billion to $32.6 billion. This protection is neither interest nor credit: it is a subscription fee, a third toll model that this article's thesis does not cover and which worked during the quarter, even if under IFRS the result was a net loss of $280.0 million (−$1.78 per diluted share) compared to a $108.3 million profit in the same quarter of 2025. The price also reflects factors the breakdown does not capture, such as Bullish's acquisition of Equiniti, the British share registrar, expected to close in early 2027. With Bullish above $37.00, the clean separation between the two blocks would cease to exist.
eToro presents the symmetrical problem. Grouping the items from its 6-K income statement, its net contribution of $229.4 million is split into 61.7% from stock, commodities, and FX trading, 20.6% from interest, 12.2% from eToro Money, and 5.5% from crypto; this is an internal calculation, and eToro's definition also moves about $1.5 million of staking from crypto to interest, which would leave crypto at 4.8% and interest at 21.2%. By either measure, the crypto contribution fell from $29.3 million to $12.5 million in a year, a 57% drop, and interest remains far from the 50% threshold that defines Circle and Figure.
What is happening with the Kraken and Grayscale IPOs after the year of Circle and Gemini?
The next cohort is already acting accordingly. Grayscale filed its public S-1 on November 13, 2025, to list on the NYSE under the symbol GRAY and has not yet debuted; the financial press places the window in Q4 2026 at the earliest. Payward, Kraken's parent company, filed its S-1 confidentially in November 2025 and froze the plan on March 17, 2026, due to market conditions; as of August 28, 2026, there is no subsequent public filing in EDGAR. Neither is a pure float business.
The test that could topple the revenue breakdown distinction is written and verifiable via stock quotes. The reading will be falsified the day one of the four transactional companies closes above its offering price—Bullish, at 11.8% below, is the candidate—or if Circle or Figure close below theirs. It would also be falsified if Q3 reports, arriving in November, show that Circle's reserve yield or Figure's credit items fall below 50% of their revenue: in the Q2 10-Q, they stand at 95.2% and 65.2%. Revenue breakdown does not cause the price; at the close of September 2, 2026, and with six observations, it is the variable that orders the signs better than market sensitivity, better than profit, and better than the listing schedule.
Related articles: What happens to Circle's reserve yield if the law bans its distribution. The race for the federal bank charter: Circle, Ripple, BitGo, and Paxos. How companies are valued before going public in perpetual markets. Price discovery before and after an IPO. Monitor your positions and stablecoins on CleanSky — portfolio tracking, lending protocols, and crypto card comparison, without investment recommendations.