Circle is responsible for issuance, while Coinbase controls the users: who actually profits from stablecoins?

Author: Darko, IOSG

Everyone is watching who issues stablecoins, but from a business model perspective, the real focus should be on who controls the distribution channels.

The crypto industry has spent a decade searching for a product that even outsiders would use. Stablecoins are the answer. In the past year, mainstream financial institutions have finally understood it.

The key change is not in technology, but in institutional attitudes. The "GENIUS Act," which took effect on July 18, 2025, establishes a federal regulatory framework for payment stablecoins in the United States for the first time. Subsequently, the OCC has promoted or approved national trust bank licenses for Circle, Ripple, and Paxos. In June 2026, Visa, Mastercard, and others joined the Open Standard, which consists of over 140 companies, and launched Open USD. Dollar tokens have been functioning normally since 2014; there has been no sudden technological breakthrough. The real "Aha!" moment is that banks and card organizations finally understand what stablecoins can do.

As of September 28, 2026, DefiLlama shows that the total market capitalization of stablecoins is approximately $306.6B, with USDT around $183.7B and USDC around $75.4B. Compared to the mid-July 2026 data used in the original text, the total has slightly increased, and USDC has also grown. Citi's benchmark forecast remains at $1.9T by 2030.

Beyond enthusiasm, two things need to be clarified.

First, the trading volume in the title seriously overestimates real usage. BCG and Allium estimate that the total transfer amount of stablecoins will be about $62T in 2025; after excluding bot trading, routing, and internal transfers, only $4.2T remains, of which the amount actually used for purchasing goods and services is about $350B-$550B. The real payment scale is one to two orders of magnitude lower than the headline.

Second, supply contraction and usage innovation can occur simultaneously. In June 2026, the supply of stablecoins decreased by $7.7B, marking the largest single-month decline since 2022, but the decline was only 2.4%, and there was no decoupling. In the same month, the adjusted transfer volume reached $1.79T, a year-on-year increase of 63%. Fewer dollar tokens are circulating, but they are moving faster.

There are three reasons behind this: The crypto market is reducing risk. BTC and ETH have fallen, with over $4B flowing out of spot Bitcoin ETFs. When traders deleverage, they redeem trading collateral, but corporate wages and operating funds do not disappear simultaneously. The market is preemptively responding to the interest-bearing restrictions of the "GENIUS Act." The relevant restrictions will take effect on January 18, 2027, and some funds have shifted to tokenized government bonds that can still provide direct yields. As of September 24, 2026, the scale of tokenized U.S. government bonds is approximately $14.9B, down from about $16B at the end of July. Therefore, the direction of "funds continuously flowing from stablecoins to tokenized government bonds" is worth noting, but recent data is not showing unilateral growth. The velocity of money is increasing. Standard Chartered estimates that stablecoins are turning over an average of 6 times per month, double that of two years ago. The same payment volume now does not require the same amount of stablecoin balance to be held. Only the first point indicates that the demand for stablecoins is still influenced by the crypto cycle, and it may not be the most important factor.

The third point is the most noteworthy because it does not completely align with the interests of issuers. Issuer revenue is roughly equal to "held funds × yield." The more widespread the payments, the faster the funds turn over; the stablecoin balance required for each dollar of payment actually decreases.

Trading collateral sits there, continuously earning money for the issuer; operating funds keep flowing, contributing less in held yield. If the payment thesis holds, issuers will see more usage, but they may not earn more from each unit of usage.

Thus, the question is no longer whether stablecoins will grow, but:

If stablecoins become the default transmission method for internet dollars, which layer can retain the profits?

Issuance creates a revenue pool, distribution determines how it is shared.

1. Where does the money for a stablecoin come from, and who takes it?

There are five layers of participants making money in the stablecoin ecosystem, corresponding to three business models: reserve interest is an interest rate business, fund flow fees are a trading volume business, and infrastructure is a SaaS business.

This is the most asymmetric part. The issuance side creates the largest profit pool, while the distribution side decides where the funds sit, but the public chain responsible for transfers is actually the cheapest layer in the value stack.

2. Issuers earn interest but cannot retain all profits

Circle has gone public and has the most complete disclosures, making it suitable for analysis.

These numbers indicate three things. Revenue follows held funds, not payment volume. Circle earns interest during the period funds are held, not fees when dollars move. In Q2 2026, USDC on-chain transaction volume grew by 151%, but it did not directly increase reserve income. What truly constrains revenue is interest rates, not competition. USDC's average circulation increased by 25%, but reserve income only grew by 5%. The reserve yield dropped by 66 basis points, consuming most of the incremental growth from the balance increase. This is more like the impact caused by a decline in SOFR rather than USDC losing market share to competitors. Less than half is retained by the issuer. Circle retained 39% after deducting distribution costs in 2025, rising to 41.2% in Q2 2026. The proportion taken by distributors has not continued to rise, but the revenue pool they share is shrinking as interest rates decline.

3. Whoever controls the users controls the revenue share

The following four sets of data become increasingly concerning for issuers. Circle obtained the license, but Coinbase still hasn't lost a penny According to the 2023 agreement, Coinbase can receive most of the reserve income generated by USDC on its platform and a portion of the reserve income from USDC outside the platform. Circle paid Coinbase $324.6M in related distribution costs in Q2 2026; the total cost paid to major distributors in 2025 was about $1.66B.

This three-year agreement entered the renewal window in August 2026. Circle confirmed on the earnings call on August 5 that the agreement would be renewed under the original terms until 2029.

This is the strongest evidence in the entire text. Circle had already obtained a federal trust bank license, the payment network was growing, and there were multiple distribution partnership cases in the market that gave 90%-100% of reserve income to the channels. Two months ago, Coinbase joined the Open USD alliance, which competes with USDC, and Circle's stock price fell about 17% that day. However, none of this changed the renewal terms.

The reason is simple: customers are in the hands of the distributors. Whether the contract can be renewed mainly depends on performance thresholds, which does not mean that the issuer has the initiative to reprice.

In Q2 2026, the average USDC balance in Coinbase's products reached a historic high of $20B, accounting for over 30% of the quarter-end USDC circulation. Nearly one-third of Circle's revenue base is concentrated on a single competitor's platform, and the next complete renewal window will not come until 2029. Hyperliquid: Trying to issue its own token to capture interest, ultimately returns to USDC In September 2025, Hyperliquid launched USDH, hoping to retain the reserve income corresponding to billions of USDC on its platform. USDH only reached a maximum scale of about $21M and ceased operations on June 20, 2026.

The subsequent arrangements are more interesting. There is about $6B USDC on Hyperliquid. Coinbase counts it as part of its platform balance, obtaining corresponding reserve income, and then returns about 90% to Hyperliquid. Why is Coinbase in the middle instead of Circle doing it directly? Because Coinbase has transferable channel profits. By recognizing the balance as platform funds, Coinbase first obtains a higher proportion of income and still has room to return to Hyperliquid. Circle's economic interests remain basically fixed in both scenarios, and the renewal locks in another three years. USDG does not make money from issuance but gives profits to the channels Paxos is responsible for issuing the Global Dollar, with a partner network that includes Robinhood, Kraken, Galaxy, and Mastercard. Partners can retain the vast majority of reserve income; if the balance stays on their platform, the proportion can reach up to 100%. The issuer here acts more like a service provider rather than a profit center.

This model will self-reinforce. Robinhood Chain launched on July 1, 2026, designating USDG as the only native stablecoin, issuing $178M in the first week, and supporting a 7% Earn product. Since channels can retain profits, they will naturally direct their own chains, brokerage businesses, and 27 million accounts toward the stablecoin that is most willing to pay. Open USD: Over 140 companies band together to turn stablecoins into a channel business Open USD was launched in June 2026, backed by over 140 participants, including Visa, Mastercard, and Coinbase. After deducting management fees, it distributes reserve income to distribution channels. USDG proves that channels can take away income through contracts; Open USD directly incorporates this arrangement into product design and includes large card organizations.

Coinbase, while completing the renewal as the largest distributor of USDC, also joined Open USD. Its disclosures show that in the first half of 2026, the combined stablecoin trading volume of USDC and Coinbase accounted for 79% of stablecoin trading volume, up from 55% in 2025. Distribution channels can remain neutral regarding the currency, which issuers cannot do.

4. With interest rates dropping by 300 basis points, issuance profits evaporate by nearly 80%

Issuer revenue can be simplified to: held funds × yield, minus the contractual channel share, minus relatively fixed operating costs.

Assuming an issuer has $100B in held funds, with 50% of the yield shared with channels, and annual operating costs of $600M:

If the deposited funds remain unchanged and the yield decreases by 300 basis points, it will erase about 79% of the operating profit. To continue earning $1.9 billion at a 2% yield, the deposited funds must increase to about $250 billion. In other words, the scale needs to grow by 150% just to return to the starting point.

This is no longer a hypothesis. Circle's actual reserve yield for Q2 2026 was 3.48%, which is right around the second line. The assumption of a 50% channel share is even conservative: total revenue for the quarter was $701 million, with distribution, transaction, and other costs amounting to $412 million.

Of course, Circle has not been idle.

At the end of the quarter, the annual transaction volume of the Circle Payments Network reached $14.7 billion, a quarter-over-quarter increase of 76%; as of July 31, it had risen to $23 billion. The company raised its full-year non-reserve revenue guidance to $310 million-$330 million. However, $242 million of that came from a one-time token presale, not sustainable revenue. The payment network had not yet started charging fees at that time, so the real focus should be on the fee rates, not the transaction volume.

Circle is also expanding downstream. At the time of writing, Arc was planning to launch on September 16; as of the date of this update, the Arc public mainnet has launched as scheduled, with over 100 institutions and ecosystem builders participating on the first day, including BlackRock, DTCC, ICE, Mastercard, Standard Chartered, and Visa, all serving or participating in validation and integration. Circle is using its still ample deposited funds income to purchase a position in the settlement layer.

Tether seems like an exception, but the data from Q2 2026 indicates that direct comparisons can be misleading. Tether reported operating profit of about $1.5 billion for the quarter, with a USDT issuance of about $184.6 billion, holding about $115 billion in U.S. Treasury bonds, and without the massive channel share like Circle. However, the operating profit emphasized by the company does not include the market value fluctuations of its held gold and Bitcoin, both of which saw significant declines that quarter. Excess reserves dropped from $8.23 billion to $4.11 billion. Supply increased, but the buffer was halved due to asset price fluctuations.

The same accounting standards also amplified the 2024 profit often cited by outsiders. Of the reported $13 billion profit at that time, about $5 billion came from unrealized gains. Excluding the market value fluctuations on both sides, Tether resembles a business earning $1 billion-$1.5 billion per quarter from Treasury bond spreads: large in scale, sensitive to interest rates, and bearing balance sheet risks that regulated issuers typically cannot take on.

The real dividing line is not "regulated" versus "offshore," but rather self-distribution versus rented distribution. Circle rents channels, with terms locked until 2029; Tether had already built its network before the arrival of regulation. Banks and payment companies that directly control customers can usually retain profits far exceeding 41%.

V. No matter how fast public chains run, they usually can only earn a fraction

Transferring on Solana or Base often costs less than a cent. Compared to the interest generated by storing one dollar in reserves for a year, the transfer fees are almost negligible. Public chains only charge when funds move, while issuers earn money every day the balance exists.

These figures can only be used for directional comparisons, as the data standards from different providers are not consistent, and Tron is particularly susceptible to statistical methods.

Tron is the most enlightening exception. BCG estimates that in 2025, Tron processed about $235 billion-$375 billion in real economic stablecoin payments, surpassing other public chains. The reason is not throughput, but that certain cross-border channels have already formed standards around Tron: exchange support, wallet integration, deep USDT liquidity, and long-established user habits.

Even so, Tron does not own the customers. Withdrawal services bring users in, wallets provide interfaces, and Tether provides dollars. Tron is the settlement track of this channel, not the distributor.

Neutral blockchain space struggles to achieve high profits because public chains are always competing on who is cheaper. Only when liquidity and usage habits make it difficult for users to leave does the track gain pricing power. The biggest winner is hidden off-chain: the U.S. Treasury According to the 2023 agreement, Coinbase can receive most of the reserve income generated by USDC on its platform and a portion of the reserve income from USDC outside the platform. The distribution costs Circle paid to Coinbase in Q2 2026 amounted to $324.6 million; the total cost paid to major distributors for the entire year of 2025 was about $1.66 billion.

This three-year agreement entered the renewal window in August 2026. Circle confirmed on the August 5 earnings call that the agreement would be renewed under the original terms until 2029.

This is the strongest evidence in the entire text. At that time, Circle had already obtained a federal trust bank license, the payment network was growing, and there were multiple distribution partnership cases in the market that allocated 90%-100% of reserve income to channels. Two months prior, Coinbase joined the Open USD Alliance, competing with USDC, and Circle's stock price dropped about 17% that day. However, none of this changed the renewal terms.

The reason is simple: customers are in the hands of the distributors. Whether the contract can be renewed mainly depends on performance thresholds, which does not mean that the issuer has the initiative to reprice.

In Q2 2026, the average USDC balance in Coinbase products reached a historical high of $20 billion, accounting for more than 30% of the USDC circulation at the end of the quarter. Nearly one-third of Circle's revenue base is concentrated on a single counterparty's platform, and the next complete renewal window will not come until 2029. All the participants mentioned earlier are actually competing for the same income pool: the interest on reserve assets. Where does the interest come from? The answer usually does not appear in the stablecoin value chain diagram.

The "GENIUS Act" requires issuers to hold cash or U.S. Treasury bonds with maturities not exceeding 93 days. It not only regulates stablecoins but also legally creates demand for short-term U.S. Treasury bonds.

As of the end of Q2 2026, Tether held about $115 billion in U.S. Treasury bonds, claiming to be the largest non-sovereign holder. Circle's approximately $79 billion in reserves is mostly placed in funds managed by BlackRock. The IMF pointed out that the combined holdings of U.S. Treasury bonds by the two companies have already surpassed those of Saudi Arabia.

About 99.8% of the stablecoin supply is denominated in U.S. dollars, so regardless of which layer ultimately takes the value, the growth of stablecoins will drive T-bill demand. Standard Chartered envisions that the scale of stablecoins will reach $2 trillion by 2028, corresponding to up to about $1 trillion in new U.S. Treasury bond purchases. Such buyers barely existed five years ago and do not demand maturity premiums.

But two restrictions need to be added. First, stablecoins are still small compared to the approximately $7 trillion money market fund industry, acting as marginal buyers that are continuously growing, but not yet dominant. Second, the impact of fund inflows and outflows on yields is not symmetrical. BIS research found that an inflow of $3.5 billion could reduce the yield on three-month Treasury bonds by about 2-2.5 basis points; an equivalent outflow could push it up by 6-8 basis points. Redemptions have about three times the impact on risk-free rates compared to subscriptions. Therefore, the $7.7 billion supply contraction in June is not just a sentiment indicator.

VI. The truly stable business lies in deposits, withdrawals, FX, and compliance

Deposits and withdrawals and FX. On-chain transfers are cheap, but entering and exiting the on-chain world is not. BCG estimates that exchange deposit and withdrawal fees are about 0.1%-1%, professional service providers charge 1%-3%, and crypto ATMs can go as high as 7%. Emerging market cross-border channels also face higher FX spreads. These are the widest trading profits in the entire value chain and are concentrated in the areas with the most rigid demand: acquiring dollars. Compliance, custody, and payment orchestration. Regulated entities cannot simply access a wallet to launch business. List screening, transaction monitoring, key management, custody, reserve services, and auditable reports mostly need to be outsourced. While it may not be as eye-catching as reserve income, it is more stable: long contract cycles, high switching costs, and it does not rely on interest rates being maintained at 5%, nor on whether a particular channel is renewed.

This layer may not capture the largest profit pool, but regardless of which stablecoin ultimately prevails, it can charge fees.

VII. After issuers can no longer pay interest, channels become more valuable

The "GENIUS Act" prohibits issuers from directly paying interest solely because users hold coins, but it does not clearly prohibit independent exchanges or wallets from subsidizing rewards with their share of reserve income.

The legislative intent is to keep stablecoins as payment tools rather than turning them into deposit substitutes. However, the actual market effect is different: when issuers cannot use income to compete for balances, competition shifts to channel shares that control users. Issuers pay distributors, who then decide whether to give a portion of that to users.

Thus, the ban on interest payments for holding coins actually protects channel profits. USDG and Open USD were designed to take advantage of this space.

This advantage is also controversial. Banking groups want to expand the restrictions to third-party rewards, and some 2026 market structure bills are attempting to more broadly limit passive income. The boundaries are still not defined.

Banks also face choices: issue their own and bear the costs, provide custody and reserve services for others' stablecoins, join alliances, or watch deposits flow out. My baseline judgment is that most of the announced bank stablecoins will ultimately become alliance products or infrastructure collaborations. Issuance has clear scale effects, and few individual banks have large enough distribution networks to independently establish liquidity.

VIII. Conclusion: Stablecoins will commoditize, but the entry points will not

Visa does not issue cards, lend, or directly collect interchange; these tasks are completed by banks. Visa earns money from the network.

Creating payment tools does not equate to controlling profits. Value will ultimately flow to the party that controls the scarce control points: acceptance networks, distribution channels, liquidity, or customer relationships.

Stablecoins are rapidly reconstructing this framework. Three conclusions can be drawn from this.

Issuance will gradually standardize.
The GENIUS Act has not made issuance easier; approval, liquidity, redemption infrastructure, and trust remain barriers, and Circle's federal license is indeed valuable. However, regulatory requirements will compel issuers to hold similar reserve assets, and disclosure rules will converge, with each regulated dollar stablecoin promising the same thing. Differentiation will shift to the tokens themselves. Circle obtained its license before the renewal window but did not change the terms for Coinbase.

Declining interest rates will compress the profits from idle funds, and this has already begun.
The reserve yield has dropped by 66 basis points, resulting in only a 5% increase in reserve income for a 25% average circulation growth. Any issuer model calculated at a 5% interest rate should be recalculated at 2%. Circulation growth can offset interest rate cuts, but it must occur before channel sharing and fixed costs erode profits. More challenging is that the payment scenarios that most drive adoption require the least balance to support each dollar of transaction volume.

Distribution channels have bargaining power.
Coinbase's renewal is the most informative stablecoin contract event of 2026, and the information comes from "nothing has changed." Circle enters the renewal window with a federal license, a growing payment network, and publicly comparable cases where 90%-100% of the revenue is transferred to the channels; Coinbase holds over 30% of USDC balances and has joined competitive alliances, with no obligation to renegotiate. Ultimately, the agreement extends under the original terms until 2029.

For asset allocators, the following criteria can be used:

• Conduct stress tests on pure issuance models with lower yields and higher channel sharing.
• Treat the renewal date as an informational event rather than merely a risk event. Automatic renewal mechanisms are usually more favorable to channels.
• Separate one-time income from sustainable income. Token sales and listing incentives can beautify performance during the transition period; what truly matters is the fee rate corresponding to transaction volume.
• Focus on companies that control the entry points for deposits and withdrawals and specific payment channels, especially in emerging markets where dollars are scarce and FX spreads are wider.
• Compliance, custody, and trading infrastructure are least sensitive to interest rates but still need to account for regulatory risks and cyclical risks.
• Do not treat transaction volume as income. USDC processed $14.8 trillion in one quarter but still primarily earns from static balances.

If this is truly the "Aha!" moment for currency, the insight gained from the AI industry is not that foundational models will necessarily commoditize, but that capabilities diffuse quickly, prices will be driven down, and bargaining power will ultimately shift to scarce distribution capabilities and customer relationships.

Stablecoins themselves may become increasingly interchangeable, but the paths for users to acquire, hold, and spend stablecoins will not.

Data Updates and Source Notes

• Circle Q2 2026 Financial and Operational Data: https://www.circle.com/pressroom/circle-reports-second-quarter-2026-results
• Circle Arc Mainnet Launch: https://www.circle.com/pressroom/circle-launches-arc-mainnet-an-economic-operating-system-for-the-internet
• Circle Q2 2026 10-Q: https://www.sec.gov/Archives/edgar/data/1876042/000187604226000248/crcl-20260630.htm
• Coinbase Q2 2026 10-Q: https://www.sec.gov/Archives/edgar/data/1679788/000167978826000088/coin-20260630.htm
• Circle-Coinbase Collaboration Agreement: https://www.sec.gov/Archives/edgar/data/1876042/000187604226000062/coinbasecollaborationagree.htm
• OCC License and Condition Approval Records: https://www.occ.gov/topics/charters-and-licensing/interpretations-and-decisions/2026/interpretations-and-actions-jan-2026.html
• Open USD Announcement: https://www.fireblocks.com/blog/open-usd-stablecoin-infrastructure-partner
• Tether Q2 2026 Reserve Verification: https://tether.io/news/tether-posts-strong-q2-performance-generates-1-5b-net-operating-profit-maintains-4-11b-reserve-buffer-and-expands-gold-holdings-to-more-than-146-tons/
• Stablecoin Market Capitalization and Supply: DefiLlama, September 28, 2026 snapshot, https://defillama.com/stablecoins
• Tokenized U.S. Treasuries: RWA.xyz, September 24, 2026 snapshot, https://app.rwa.xyz/treasuries

Note: Different data platforms may have varying statistical criteria for stablecoin supply, on-chain transaction volume, and public chain transaction fees. Tether data comes from BDO's verification report and is not a U.S. GAAP audit. Data from Citi and Standard Chartered are forecasts and should not be viewed as definitive results.