IOSG partner Darko argues that the most important question in the stablecoin business is not simply who issues the coin, but who controls distribution. In his view, crypto spent a decade looking for a product that people outside the industry would actually use. Stablecoins are that product. What changed over the past year was not the technology itself, but the attitude of major financial institutions.
The article points to the GENIUS Act, signed into effect on July 18, 2025, as the first federal framework for U.S. payment stablecoins. After that, the Office of the Comptroller of the Currency moved forward with or approved national trust bank charters for Circle, Ripple and Paxos. In June 2026, Visa and Mastercard joined the Open Standard, a group of more than 140 companies, and helped launch Open USD. Dollar tokens had worked since 2014, the piece says. The real "Aha!" moment came when banks and card networks finally understood what stablecoins could do.
Stablecoins are growing, but headline transfer numbers overstate real payment use
As of Sept. 28, 2026, DefiLlama showed total stablecoin market capitalization at about $306.6B, including roughly $183.7B in USDT and $75.4B in USDC. The piece notes that these figures are slightly higher than the mid-July 2026 numbers used in the original draft, with USDC also moving up. Citi’s base-case forecast remains $1.9T by 2030.
Darko says two points matter beyond the enthusiasm.
First, headline transaction volume materially overstates real usage. BCG and Allium estimated total stablecoin transfers at about $62T in 2025. Once bot activity, routing and internal transfers are removed, that falls to $4.2T. Of that, only about $350B-$550B was tied to the purchase of goods and services. Real payment scale, in other words, was one to two orders of magnitude below the headline number.
Second, supply contraction and record usage can happen at the same time. In June 2026, stablecoin supply fell by $7.7B, the largest monthly decline since 2022. The drop was only 2.4%, and there was no depeg. In the same month, adjusted transfer volume reached $1.79T, up 63% year over year. Fewer dollar tokens were outstanding, but they moved faster.
Why supply fell while transfer activity rose
The article lists three drivers behind that pattern.
- Risk came out of crypto markets. BTC and ETH fell, while spot Bitcoin ETFs saw more than $4B in outflows. When traders deleverage, they redeem trading collateral, but corporate payroll and working capital do not disappear in parallel.
- The market started positioning ahead of the GENIUS Act’s ban on direct interest paid by issuers to token holders. That restriction takes effect on Jan. 18, 2027. Some capital shifted into tokenized Treasuries that can still offer yield directly. As of Sept. 24, 2026, tokenized U.S. Treasury products stood at about $14.9B, below the roughly $16B level at the end of July. The direction still matters, the article says, but recent data do not show a straight one-way increase.
- Velocity increased. Standard Chartered estimated average monthly turnover in stablecoins at about six times per month, twice the level from two years earlier. The same amount of payment activity now requires less idle stablecoin inventory.
Darko says only the first factor clearly shows that stablecoin demand still moves with crypto cycles, and even that may not be the dominant force. The third factor matters more because it does not fully align with issuer economics. Issuer revenue is roughly equal to idle balances multiplied by yield. As payments become more common and money turns over faster, each dollar of payment activity requires fewer stablecoins sitting still.
Trading collateral that remains parked can keep earning for the issuer. Working capital that keeps moving does not. If the payment thesis plays out, issuers may get more usage without earning more per unit of use. That shifts the central question. The issue is no longer whether stablecoins will grow, but which layer keeps the profits if stablecoins become the default way to move dollars across the internet.
The article’s answer is concise: issuance creates the revenue pool, distribution decides how it is divided.
Five layers in the stack make money, but not in the same way
The piece says five layers in the stablecoin stack earn money across three business models. Reserve income is an interest-rate business. Money-flow fees are a volume business. Infrastructure is a software-as-a-service business.
The asymmetry sits here. Issuers create the largest profit pool. Distributors decide where balances rest. The public chains that process transfers, by contrast, are often the cheapest layer in the value stack.
Circle shows that issuers earn on balances, not on transfer volume
Darko uses Circle, now public and more transparent than peers, as the clearest case study.
He draws three conclusions from the numbers.
Revenue follows idle balances, not payment volume. Circle earns interest while USDC sits in reserve-backed form, not a fee each time dollars move. In Q2 2026, on-chain USDC transaction volume rose 151%, but that did not directly lift reserve revenue.
The main constraint on revenue is rates, not competition. Average USDC in circulation grew 25%, yet reserve revenue rose only 5%. Reserve yield fell 66 basis points and absorbed most of the benefit from higher balances. The article says this looks more like the effect of lower SOFR than market share loss to rivals.
Issuers keep less than half. After distribution costs, Circle retained 39% in 2025, improving to 41.2% in Q2 2026. The share taken by distribution partners did not keep rising, but the pool they split is shrinking as rates move lower.
Whoever owns the user relationship owns the split
The article then walks through four examples. Each one, it says, should make issuers more cautious.
Circle got the license. Coinbase still did not give up economics.
Under a 2023 agreement, Coinbase receives most of the reserve income generated by USDC held on its platform and a portion of reserve income from off-platform USDC. Circle paid Coinbase $324.6M in related distribution costs in Q2 2026. In full-year 2025, it paid about $1.66B to major distribution partners.
The three-year agreement entered its renewal window in August 2026. On its Aug. 5 earnings call, Circle confirmed that the deal had been renewed through 2029 on the existing terms.
Darko presents this as the strongest evidence in the article. By then, Circle had already secured a federal trust bank charter, its payments network was expanding, and multiple comparable distribution deals in the market were handing 90%-100% of reserve income to channels. Two months earlier, Coinbase had also joined the Open USD alliance, which competes with USDC, and Circle shares fell about 17% that day. None of that changed the renewal terms.
The reason, the piece says, is simple: the customer sits with the distributor. Whether a contract renews may depend largely on performance thresholds, but that does not mean the issuer has real repricing power.
In Q2 2026, average USDC balances inside Coinbase products reached a record $20B, accounting for more than 30% of USDC in circulation at quarter-end. Nearly one-third of Circle’s revenue base was concentrated on a single counterparty platform, and the next full renewal window will not come until 2029.
Hyperliquid tried to issue its own coin, then came back to USDC
In September 2025, Hyperliquid launched USDH in an effort to keep reserve income associated with the billions of dollars in USDC on its platform. USDH peaked at only about $21M and stopped operating on June 20, 2026.
The follow-up arrangement was even more telling. Hyperliquid held about $6B in USDC. Coinbase counted that as on-platform balance, collected the associated reserve revenue, then returned roughly 90% to Hyperliquid. Darko asks why Coinbase sat in the middle instead of Circle dealing directly. His answer is that Coinbase had channel economics it could transfer. By classifying the balances as on-platform funds, Coinbase first captured a higher percentage of revenue and still had room to rebate most of it to Hyperliquid. Circle’s economics, by comparison, remained largely fixed in either setup, and the renewed deal locked that in for another three years.
USDG shifts the economics to the channel
Paxos issues Global Dollar, while the partner network includes Robinhood, Kraken, Galaxy and Mastercard. Partners can keep the vast majority of reserve income. If balances remain on their own platforms, the share can reach 100%. In this model, Darko says, the issuer looks more like a service provider than a profit center.
He argues that the design reinforces itself. Robinhood Chain launched on July 1, 2026, made USDG its only native stablecoin, issued $178M in the first week, and used it to support a 7% Earn product. Once a channel can keep the economics, it has a direct reason to route its chain, brokerage business and 27 million accounts toward the stablecoin willing to pay the most.
Open USD writes revenue sharing into the product itself
Open USD launched in June 2026 with backing from more than 140 participants, including Visa, Mastercard and Coinbase. After management fees, reserve income is distributed to channels. USDG showed that channels could claim the economics through contract design. Open USD, the article says, hard-codes that arrangement into the product and adds major card networks on top.
Coinbase renewed its role as the largest USDC distributor while also joining Open USD. Its disclosures show that USDC plus Coinbase partner stablecoins represented 79% of stablecoin transaction volume in the first half of 2026, up from 55% in 2025. Distribution channels can stay neutral across coins, Darko writes. Issuers cannot.
A 300-basis-point rate drop can wipe out most issuer profits
The article reduces issuer economics to a simple equation: idle balances multiplied by yield, minus channel sharing fixed by contract, minus relatively fixed operating costs.
In a hypothetical example, an issuer with $100B in balances, 50% of reserve income shared with channels and $600M in annual operating costs would lose about 79% of operating profit if yields fell by 300 basis points and balances stayed unchanged. To keep earning $1.9B at a 2% yield, balances would need to rise to about $250B. That means scale would have to increase 150% just to get profits back to the starting point.
Darko says this is no longer just a thought experiment. Circle’s actual reserve yield in Q2 2026 was 3.48%, almost exactly where the second row of that model would sit. The 50% distribution-share assumption may even be conservative. Circle posted $701M in revenue that quarter, while distribution, transaction and other costs totaled $412M.
The article adds that Circle is not standing still. At quarter-end, Circle Payments Network was running at $14.7B in annualized transaction volume, up 76% quarter over quarter. By July 31, that figure had reached $23B. The company raised full-year guidance for non-reserve revenue to $310M-$330M. Still, $242M of that came from a one-time token presale rather than recurring revenue. The payments network had not started charging fees at that point, so the metric to watch is pricing, not raw volume.
Circle is also pushing downstream. When the original article was written, Arc was still scheduled to launch on Sept. 16. By the time of the update, Arc public mainnet had launched as planned. More than 100 institutions and ecosystem builders participated on day one, while BlackRock, DTCC, ICE, Mastercard, Standard Chartered and Visa served as validators or integration partners. Darko says Circle is using still-strong reserve income to buy a place in the settlement layer.
Tether looks different, but simple comparisons can mislead
Tether appears to be an exception, the article says, but Q2 2026 numbers show why direct comparison can be misleading.
Tether reported about $1.5B in operating profit for the quarter, had about $184.6B in USDT outstanding, and held about $115B in U.S. Treasuries. It also does not have a distribution-sharing burden on the scale Circle does. But the company notes that its operating profit excludes mark-to-market changes in gold and Bitcoin held on its own balance sheet, and both assets fell sharply in the quarter. Excess reserves dropped from $8.23B to $4.11B. Supply increased, but the buffer was cut in half by asset-price moves.
The same accounting treatment also affects the often-cited 2024 profit figure. Of the reported $13B, about $5B came from unrealized gains. Strip out mark-to-market swings in both directions, and Tether looks more like a Treasury carry business earning $1B-$1.5B a quarter: very large, rate-sensitive and willing to take balance-sheet risk that regulated issuers usually cannot.
The real dividing line, Darko argues, is not "regulated" versus "offshore" but owned distribution versus rented distribution. Circle rents its channel and is locked into terms through 2029. Tether built its network before the current regulatory era arrived. Banks and payment firms that directly own customer relationships can often keep well above 41% of the economics.
Public blockchains move money cheaply, but usually keep only scraps
On Solana or Base, transfers often cost less than one cent, the article says. Compared with the interest generated by holding one dollar of reserves for a year, transfer fees are almost negligible. Public chains earn only when money moves. Issuers earn every day balances sit still.
Darko adds that these numbers should be used directionally because data providers do not use identical methodologies for stablecoin supply, on-chain volume and chain fees. Tron is especially sensitive to the way activity is counted.
Even so, Tron is the most instructive exception. BCG estimated that Tron handled about $235B-$375B in real-economy stablecoin payments in 2025, ahead of other chains. The reason was not throughput alone. Certain cross-border corridors had already standardized around Tron because of exchange support, wallet integrations, deep USDT liquidity and established user habits.
But even in that case, Tron does not own the customer. Deposit and withdrawal services bring users in. Wallets provide the interface. Tether provides the dollar. Tron is the settlement rail, not the distributor. Neutral blockspace struggles to command high margins because public chains compete constantly on price. Only when liquidity and user habits make switching hard does the rail gain pricing power.
The biggest winner may be the U.S. Treasury
All of the players above are fighting over the same pool: interest earned on reserve assets. The source of that interest usually does not appear in a standard stablecoin value-chain diagram.
The GENIUS Act requires issuers to hold cash or U.S. Treasuries with maturities of no more than 93 days. That means the law does more than regulate stablecoins. It also creates legal demand for short-dated U.S. government debt.
At the end of Q2 2026, Tether held about $115B in U.S. Treasuries and described itself as the largest non-sovereign holder. Circle had about $79B in reserves, most of them in funds managed by BlackRock. The IMF has said the two companies together hold more Treasuries than Saudi Arabia.
About 99.8% of stablecoin supply is denominated in dollars, the article notes, so growth in the sector will increase T-bill demand regardless of which layer captures the economics. Standard Chartered projected that stablecoins could reach $2T by 2028, implying as much as $1T in additional Treasury demand. Buyers like this barely existed five years ago, and they do not ask for term premium.
Darko still adds two constraints. First, stablecoins remain small compared with the roughly $7T money market fund industry. They are a growing marginal buyer, not a dominant one. Second, inflows and outflows affect yields asymmetrically. BIS research found that a $3.5B inflow can push the three-month Treasury yield down by about 2-2.5 basis points, while an outflow of the same size can push it up by 6-8 basis points. Redemption shocks the risk-free rate roughly three times as much as subscriptions do. That is why the $7.7B supply contraction in June is more than a sentiment signal.
The steadier businesses sit in ramps, FX, compliance and custody
The article says the most stable economics in the stack may sit outside issuance itself.
On- and off-ramps, plus FX
On-chain transfers are cheap, but entering and leaving the on-chain world is not. BCG estimated exchange deposit and withdrawal fees at roughly 0.1%-1%, specialist service providers at 1%-3%, and crypto ATMs as high as 7%. In emerging-market cross-border corridors, those costs can be layered with wider FX spreads. According to Darko, that is where some of the widest transaction margins in the entire value chain sit, concentrated in the part of demand that is hardest to avoid: getting dollars.
Compliance, custody and payment orchestration
Regulated institutions cannot launch a business simply by plugging into a wallet. Sanctions screening, transaction monitoring, key management, custody, reserve services and auditable reporting are often bought from outside providers. Those revenues may not look as flashy as reserve income, but they are steadier. Contracts last longer, switching costs are higher, and the model does not depend on rates staying at 5% or on a single distribution contract getting renewed.
This layer may not capture the biggest pool of profits, the article says, but it can charge fees no matter which stablecoin wins.
Once issuers cannot pay interest directly, channels become more valuable
The GENIUS Act bars issuers from paying interest simply because users hold a stablecoin. It does not clearly prohibit independent exchanges or wallets from using their own share of reserve income to subsidize rewards.
The legislative goal was to keep stablecoins functioning as payment tools rather than deposit substitutes. Darko argues that the market effect may be different. If issuers cannot use yield to compete for balances directly, the contest moves to revenue sharing with channels that control users. Issuers pay the distributor. The distributor then decides whether to pass part of that back to users.
In that sense, the ban on direct interest can end up protecting channel margins. USDG and Open USD are built to take advantage of that opening.
The boundary remains contested. Banking groups want the restriction expanded to third-party rewards, and some 2026 market-structure bills also sought broader limits on passive yield. The line is still unsettled.
Banks face their own choices: issue a coin and bear the cost, provide custody and reserve services for someone else’s stablecoin, join a consortium, or watch deposits migrate away. Darko’s base case is that many announced bank stablecoins will eventually turn into consortium products or infrastructure partnerships. Issuance has real scale effects, and a single bank rarely controls a distribution network large enough to build liquidity on its own.
Conclusion: stablecoins may commoditize, but access points will not
The article closes with a Visa analogy. Visa does not issue cards, extend credit or directly collect interchange. Banks do that. Visa makes money from the network.
Building the payment instrument does not mean controlling the profit pool. Value tends to flow to whoever controls the scarce choke point: acceptance networks, distribution, liquidity or the customer relationship. Stablecoins are now rebuilding that structure quickly.
From there, Darko draws three conclusions.
Issuance will gradually standardize. The GENIUS Act does not make issuance easy. Approvals, liquidity, redemption infrastructure and trust remain barriers, and Circle’s federal charter still has value. But regulation pushes issuers toward similar reserve assets and similar disclosures, while every regulated dollar stablecoin promises the same thing. Differentiation shifts above the token. Circle secured its federal license before the renewal window, yet that did not change Coinbase’s terms.
Lower rates compress idle-balance profits, and that has already started. A 66-basis-point decline in reserve yield turned 25% growth in average circulation into only 5% growth in reserve revenue. Any issuer model built on a 5% rate environment, the article says, should be rerun at 2%. Circulation growth can offset falling rates, but only if it outpaces pressure from channel sharing and fixed costs. The harder part is that the payment use cases driving adoption need the least balance support per dollar of transaction volume.
Distribution channels hold the bargaining power. Darko describes the Coinbase renewal as the most informative stablecoin contract event of 2026, precisely because nothing changed. Circle entered the renewal window with a federal charter, a growing payments network, and public comparables in which 90%-100% of reserve income was handed to channels. Coinbase controlled more than 30% of USDC balances, had joined a rival alliance, and had no obligation to reprice. The agreement was extended through 2029 on the same terms.
The article finishes with a screening framework for allocators:
- Stress-test pure issuance models with both lower yields and higher channel sharing.
- Treat renewal dates as information events, not just risk events. Automatic renewal structures often favor the channel.
- Separate one-off revenue from recurring revenue. Token sales and launch incentives can flatter transition-period numbers. The key metric is take rate on payment volume.
- Focus on companies that control on- and off-ramp access and specific payment corridors, especially in emerging markets where dollars are scarcer and FX spreads are wider.
- Compliance, custody and trading infrastructure are less rate-sensitive, though they still carry regulatory and cycle risk.
- Do not mistake transaction volume for revenue. The article notes that USDC processed $14.8T in one quarter and still depended mainly on money made from idle balances.
Darko’s final point is that if this really is the "Aha!" moment for money, the lesson from AI is not simply that foundation models commoditize. It is that capabilities spread fast, prices get pushed down, and bargaining power ends up with scarce distribution and customer relationships. Stablecoins themselves may become increasingly interchangeable. The paths by which users acquire them, hold them and spend them may not.
Data update and sources
The source list in the article includes Circle’s Q2 2026 financial and operating results, the Arc mainnet launch announcement, Circle’s Q2 2026 Form 10-Q, Coinbase’s Q2 2026 Form 10-Q, the Circle-Coinbase collaboration agreement, OCC charter and conditional approval records, the Open USD announcement, Tether’s Q2 2026 reserve attestation, the Sept. 28, 2026 DefiLlama snapshot for stablecoin market cap and supply, and the Sept. 24, 2026 RWA.xyz snapshot for tokenized U.S. Treasuries.
The article also notes that data providers do not use identical methodologies for stablecoin supply, on-chain transfer volume and blockchain fees. Tether data comes from a BDO attestation rather than a U.S. GAAP audit. Forecasts from Citi and Standard Chartered should not be treated as certain outcomes.