Reading view

There are new articles available, click to refresh the page.

How Does Circle Make Money? The $668 Million Answer Hiding in Plain Sight

Circle earns interest on dollars that belong to you. It is legal, it is disclosed, and it is 95% of the business. Here is where that money goes, and what the other design looks like.

Dark title card reading “How Does Circle Make Money?” with the statistic that 95.2% of Circle revenue is interest on dollars that are not Circle’s, alongside a panel showing $324.6M paid to the distributor and $0 paid to USDC holders.
Circle earned $667.7M in reserve income in Q2 2026. USDC holders received none of it.

Last quarter, Circle earned $667.7 million.

Not from fees. Not from software. From interest on dollars that USDC holders handed over and were never paid a cent on.

That is not an accusation. It is line one of a public filing.

95.2% of Circle’s Q2 2026 revenue came from interest on money that was not Circle’s.

“How does Circle make money” has a boring answer and an interesting one. The boring answer fits in a sentence.

The interesting one is about who the float belongs to, and whether it has to work this way.

So, How Does Circle Make Money? It Earns Interest on Your Idle Dollars

Answer first, then the receipts.

  • You send $1 to Circle. Circle mints 1 USDC.
  • Circle parks your dollar in short-term US Treasuries and bank deposits.
  • Those instruments pay interest. Circle collects it.
  • You hold a token worth exactly $1. Indefinitely.

The Q2 2026 numbers, from Circle’s own results:

  • Total revenue and reserve income: $701.3 million
  • Reserve income alone: $667.7 million, or 95.2% of the total
  • Transaction revenue: $5.3 million
  • USDC in circulation at quarter end: $73.3 billion
Horizontal bar chart of Circle Q2 2026 revenue showing reserve income at $667.7 million or 95.2 percent, other revenue at $28.3 million, and transaction revenue at $5.3 million.
Reserve income is 95.2% of Circle’s Q2 2026 revenue. Everything else is a rounding error.

Circle has never hidden this. Reserve income accounted for 95% to 99% of total revenue in 2022, 2023 and 2024. The model is not a secret. The model is the product.

One statistic reframes the whole thing. USDC settled roughly $14.8 trillion in onchain volume in Q2, up 151% year over year. Circle booked $5.3 million in transaction revenue from all of that movement.

The float is the business. The movement is the marketing.

Where Does USDC Reserve Income Actually Go?

Here is the part most explainers skip.

Circle does not keep most of it. In Q2 2026 the company recorded $410.4 million in distribution and transaction costs. Of that, $324.6 million went to Coinbase.

The structure, in plain terms:

  • Coinbase collects 100% of reserve income on USDC held on Coinbase.
  • Coinbase collects 50% of residual reserve income on USDC held everywhere else.
  • The agreement, signed August 2023, was confirmed renewed on the same terms through 2029 on the August 5, 2026 earnings call.
Bar chart splitting Circle Q2 2026 reserve income into $324.6 million Coinbase distribution, $85.8 million other distribution and transaction costs, $290.9 million retained by Circle, and $0 paid to USDC holders.
The yield moves. It just moves sideways, to the distributor rather than the holder.

In 2025, Coinbase-linked distribution costs hit $1.4 billion, roughly 51% of Circle’s total revenue and reserve income for the year.

So the money does move. It just moves sideways.

Your dollars generate the yield. The distributor collects it. You keep a token worth a dollar.

Why Doesn’t USDC Pay You Yield? The GENIUS Act Answer

This is where people direct their annoyance at the wrong party.

Section 4(a)(11) of the GENIUS Act bars permitted payment stablecoin issuers from paying holders any form of interest or yield for simply holding the coin. Cash, tokens, other consideration, all of it.

The OCC’s February 2026 proposed rule goes further, presuming that yield routed through affiliates and third parties is also prohibited unless the arrangement can be justified.

Circle is not choosing to withhold anything. US law forbids a payment stablecoin issuer from passing reserve income to holders.

That line is now the loudest fight in US financial policy. Banks want it enforced strictly, arguing that pass-through rewards drain insured deposits and shrink credit.

The digital asset industry argues Congress deliberately left third parties out of scope.

Every major GENIUS implementing rule across the OCC, FDIC, Treasury and FinCEN was still pending finalisation as of mid-2026, while the OCC noted private forecasts of payment stablecoin issuance reaching $500 billion this year.

Half a trillion dollars of float, and the entire policy argument is about who is allowed to earn on it.

Which surfaces the real question, and it is an engineering question rather than a moral one:

If a dollar instrument cannot legally pay its holder, what would one look like that can?

What Happens When Protocol Revenue Goes Back to the Holder Instead?

Sky Protocol was built around the opposite answer.

USDS is not a payment stablecoin issued by a company sitting on your cash. It is an overcollateralized stablecoin generated onchain against governance-approved collateral.

Users retain non-custodial control of their holdings throughout. There is no issuer holding your float.

Supply USDS to the savings module and you receive sUSDS, which programmatically accrues the Sky Savings Rate. No lockups, no exit fees, no application form.

Flow diagram comparing two models. The payment stablecoin issuer model routes user dollars through T-bills to $667.7 million of reserve income and $0 to the holder. The onchain capital allocation model routes USDS through the Sky Agent Network to protocol surplus and back to holders via the Sky Savings Rate.
Same dollar, two destinations. The design decides who earns on the float.

The receipts, from the Q2 2026 quarterly report published by Sky Frontier Foundation:

  • Cumulative Sky Savings Rate distributions to holders crossed $250 million on June 29, 2026
  • $17.49 million accrued through sUSDS in the month of June alone
  • sUSDS closed Q2 at $5.52 billion, up 149% year over year, the largest rate-bearing stablecoin by supply
  • Sky Protocol generated $107.35 million in Gross Protocol Revenue in Q2, a second consecutive quarter above $100 million
One design routes reserve income to distribution partners. The other routes protocol revenue to the people holding the instrument.

Who Sets the Sky Savings Rate, and Where Does the Money Come From?

Not from token emissions. Not from a marketing budget.

The Sky Agent Network is a group of independent capital allocators, including Spark, Grove, Keel, Obex and Osero, that borrow USDS from Sky Protocol at a governance-set Base Rate and deploy it into their own strategies across credit, lending and tokenized real-world assets. They keep their spread. They pay the Base Rate back.

Those payments, plus vault stability fees, real-world asset yield and Peg Stability Module fees, pool in the protocol’s surplus layer.

Sky Governance then sets the Sky Savings Rate as a separate parameter, calibrated against revenue capacity and reserve targets.

Two consequences worth sitting with:

  • The rate is variable and governance-set, not market-set. Mid-Q2 2026, governance moved it from 3.75% to 3.60% on purpose, to sustain the pace of reserve accumulation. No algorithm did that. People voted.
  • Governance can move the spread toward the holder. On July 23, 2026, the Sky Spread was cut from 0.1% to zero, ratified onchain.

Set that next to a distribution agreement that renews on identical terms for another three years.

Line chart of sUSDS supply rising from $2.22 billion in Q2 2025 to $3.78 billion in Q4 2025, peaking at $6.49 billion in Q1 2026 and settling at $5.52 billion in Q2 2026 after a governance-set rate adjustment.
sUSDS grew 149% year over year while paying out more than $250M to holders.

The network around it kept compounding through the quarter. Binance completed its upgrade from DAI to USDS with automatic one-to-one conversion of user balances.

Pendle Finance introduced fixed-rate access to sUSDS, which reached $55.94 million in TVL by late July at a 5.37% fixed rate.

Spark seeded $150 million into a shared stablecoin liquidity layer on Uniswap v4 and cleared $70 million in volume in its first three days.

Across the USDS and DAI complex, unique holders held broadly steady at 673,811.

Three Questions to Ask About Any Stablecoin You Hold

Steal these. They work on every issuer, including this one.

  • Who earns the interest on my balance? If the answer is “the issuer and its distribution partners,” you are the funding, not the customer.
  • Where is the revenue published, and how often? A quarterly attestation is not the same as a live balance sheet you can refresh.
  • Who can change the terms, and can I watch them do it? A private renegotiation and an onchain governance vote are very different accountability structures.

Most people have never asked question one. It is the one that decides where a few billion dollars a year ends up.

Can You Actually Verify Any of This? Yes, and That Is the Point

Stablecoin trust usually means trusting a quarterly attestation and a PDF.

Sky Protocol publishes two live surfaces instead:

  • financial.skyeco.com is the financial record: balance sheet, Gross and Net Protocol Revenue, Protocol Surplus, Sky Reserves, and the collateral backing USDS.
  • insights.skyeco.com carries the quarterly reports and monthly operational updates behind every figure above.
Scorecard of four Sky Protocol metrics for Q2 2026: $107.35 million Gross Protocol Revenue up 10.5% year over year, $12.32 billion Protocol Collateral up 45.5%, more than $250 million in cumulative Sky Savings Rate paid to holders, and 673,811 unique holders across USDS and DAI.
Four numbers, all refreshable in public, none of them requiring an attestation PDF.

Protocol Collateral reached $12.32 billion at Q2 close, up 45.5% year over year.

Sky Reserves sat at roughly 55% of the $150 million Solvency Reserve target that governance approved in March 2026, deliberately prioritising the buffer over near-term distributions.

You do not have to take any of those numbers on faith. You can open the dashboard and check them mid-sentence.

Two Designs, One Question: Who Is the Float For?

Circle’s model is legal, disclosed and, for a payment instrument, defensible. Payment rails are not savings products, and the GENIUS Act drew that line deliberately.

Still, $667.7 million a quarter is a lot of float to route past the people who supplied it.

The alternative is not “a higher number.” It is a different answer to the ownership question.

Sky Protocol is a capital allocation network where revenue lands with the protocol, and governance decides in public how much of it flows back to holders through the Sky Savings Rate. Every parameter is a vote, and every vote is onchain.

Idle dollars are never actually idle. Somebody is always earning on them.

The only question that matters is whether that somebody is you.

Your turn. If your stablecoin issuer earns roughly 3.5% on your balance and pays you nothing, is that a fee you agreed to or a fee you were never shown? Drop your answer below. I read every response.


How Does Circle Make Money? The $668 Million Answer Hiding in Plain Sight was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.

Are Stablecoins Actually Safe? A Straight Answer to an Awkward Question

Thirty-six of them have already died. The reason why is not the one you have been warned about.

Are stablecoins safe? Four numbers that reframe the question: 36 stablecoin collapses since 2022, $2.5 billion destroyed, 72% caused by backing that was not real, 4% by algorithmic death spirals.
The stablecoin safety debate, in four numbers. The failure everyone fears is the one that almost never happens.

Since 2022, 36 stablecoins have collapsed with measurable losses. Roughly $2.5 billion, gone. Six of those failures happened in 2026 alone.

Now here is the part that should change how you think about stablecoin safety.

The algorithmic death spiral everyone still talks about, the Terra scenario, accounts for about 4% of that damage. Backing that was never real accounts for about 72%.

People are afraid of the wrong thing.

That matters more every month. Around 269 million onchain addresses now hold a stablecoin balance, and the category sits near $308 billion, roughly 13% of all crypto by market value. This is no longer a niche question.

“Stablecoin” Is a Marketing Word, Not a Safety Rating

Ask how safe stablecoins are and you get an average. Averages are useless here.

A stablecoin is not one thing. It is a promise with a structure behind it. The promise is identical across every token. The structure is not.

Three tokens can all say one dollar and mean three completely different things:

  • One holds Treasury bills at a regulated custodian and publishes monthly attestations.
  • One holds crypto collateral worth more than the tokens it issued, visible onchain, around the clock.
  • One holds a sister token it printed itself and calls that a reserve.

Same peg. Same ticker format. Wildly different odds.

So the honest answer to “are stablecoins safe” is that the category tells you nothing.

The structure tells you everything. That is not a dodge. It is the actual finding sitting in four years of stablecoin failure data.

What Actually Kills a Stablecoin: The Data Nobody Quotes

Bar chart of stablecoin failure mechanisms showing share of total value destroyed: backing that was not real 72%, runs and thin liquidity 15%, oracle and smart contract exploits 9%, algorithmic death spiral 4%.
What actually kills a stablecoin. Misstated reserves take nearly three quarters of the money. Terra-style collapses take four percent.

Look at the record and the pattern shows up fast.

  • Backing that was not real. The largest cause of loss by a distance. Reserves misstated, illiquid, or quietly lent out.
  • Runs and thin liquidity. Real reserves, not reachable fast enough. USDC touched $0.8789 in March 2023 when $3.3 billion of Circle’s reserves froze at Silicon Valley Bank. It recovered, because the money existed.
  • Oracle and smart contract failure. In March 2026, an attacker minted 80 million unbacked USR tokens and walked off with roughly $24 million. The token fell 95% in hours.
  • Freeze and seizure. Not a depeg at all. Your balance is fine and simply not yours to move.
  • Algorithmic design failure. Terra’s UST, roughly $40 billion erased in May 2022. Famous, catastrophic, statistically rare.

Moody’s counted 1,914 depeg events through mid-2023. Almost all were tiny and brief.

The ones that actually cost people money were never about the peg. They were about what sat behind it, which is precisely why S&P Global now scores stablecoins on asset quality first.

Depegging is the result. It is never the cause.

The 2026 Stress Test Nobody Called a Stress Test

Line chart of total stablecoin supply from January to August 2026, peaking at $322.1 billion in May and falling roughly $14.5 billion by August, annotated to show the drop was a redemption event rather than a depeg.
The 2026 contraction was the sharpest since Terra. It was also not a depeg. Supply falling and a peg breaking are different events.

This year handed the category its first genuine squeeze in four years.

Stablecoin supply peaked near $322.1 billion in mid-May, then shed roughly $14.5 billion by early August. The sharpest contraction since Terra. You can watch the whole curve live on DefiLlama.

Here is the nuance most headlines skipped. That was a redemption story, not a depeg story.

USDT and USDC both held within about 0.1% of a dollar throughout. Tokens were being burned at a dollar, not dumped at ninety cents.

Supply shrinking and a peg breaking are completely different events. Confusing them is how people panic at exactly the wrong moment, and it happened at scale this summer.

What moved the money was policy, not fear. The GENIUS Act bars permitted payment stablecoin issuers from paying holders any yield.

So capital rotated toward structures that still can: tokenized Treasuries, and yield-generating stablecoins.

The Congressional Research Service lays out how narrowly that prohibition is drawn, and the White House Council of Economic Advisers has since questioned whether it achieves anything at all.

That rotation is not marginal. Yield-bearing stablecoins drove more than half of net stablecoin supply growth in Q1 2026.

The category is quietly splitting in two: tokens built to move, and tokens built to sit still and earn.

Five Questions That Tell You If a Stablecoin Is Safe

Checklist graphic of five questions to assess stablecoin safety: can I see the collateral now, is there more collateral than tokens, what absorbs the first loss, can anyone freeze my balance, has the structure survived a crash.
The four-minute stablecoin safety check. Most holders have never run it on the token they are holding.

Safety is checkable. It just is not checkable from a homepage. Ignore the marketing and ask these five instead.

  • Can I see the collateral right now, without asking permission? A live number, not a quarterly PDF.
  • Is there more collateral than there are tokens? One-to-one leaves zero margin for a bad day.
  • What absorbs the first loss? If nobody can answer that, the answer is you.
  • Can anyone freeze or seize my balance? Non-custodial is a structure, not a slogan.
  • Has this structure survived anything? A model that has never met a crash is a hypothesis.
If a token fails three of these, the yield is not compensation. It is a warning label.

What Verifiable Stablecoin Backing Actually Looks Like

Donut chart of USDS Protocol Collateral composition: Sky Agent vaults 40%, Peg Stability Module 38%, overcollateralized crypto vaults 22%, with $14.15 billion in Protocol Collateral against $11.48 billion in stablecoin supply.
What overcollateralization looks like when it is auditable. Roughly $1.23 of Protocol Collateral behind every dollar of supply.

Sky Protocol is worth walking through here, not as the only answer, but because every one of those five questions has a public answer.

USDS is overcollateralized by design. At the time of writing, skyeco.com shows $14.15 billion in Protocol Collateral against $11.48 billion in stablecoin supply. Roughly $1.23 sitting behind every dollar.

That collateral is not a slide in a deck. It splits across:

  • The Peg Stability Module, roughly 38%
  • Sky Agent vaults, roughly 40%
  • Overcollateralized crypto vaults, roughly 22%

Three structural controls matter more than any of the marketing around them:

  • Price data waits one hour in the Oracle Security Module before it takes effect. A manipulation attack has to hold a false reading for over an hour, in public, under governance observation.
  • Undercollateralized positions are liquidated through descending-price Dutch auctions rather than panic sales.
  • No sensitive parameter change goes live the moment a vote passes. The Governance Security Module enforces a delay on every one.

None of that requires trusting a press release. Every position is auditable at financial.skyeco.com.

When Something Breaks: The Order of Operations

Diagram of the Sky Protocol loss absorption waterfall in four ordered layers: Sky Agent risk capital, the Surplus Buffer, recapitalization through SKY issuance, and Emergency Shutdown as a last resort.
Who eats the first loss, in a fixed and published order. The sequence matters more than any reassurance.

Most protocols answer “what if you lose money” with reassurance. Sky Protocol answers it with a sequence.

  1. Sky Agent risk capital. Each Agent posts capital proportional to its exposure, sized by asset class using a Basel III (CRR) methodology. It absorbs the shortfall first.
  2. The Surplus Buffer. Protocol revenue accumulates here before distribution. In May 2026, Sky Governance raised the target to $150 million USDS.
  3. Recapitalization through SKY issuance. Requires an Executive Vote with a mandatory time delay.
  4. Emergency Shutdown. Last resort. USDS minting halts and every holder redeems directly against the remaining collateral pool at the then-current ratio.

Knowing the order is the whole point. Ambiguity about who eats the first loss is itself the risk, and Sky Governance publishes every parameter behind that sequence onchain.

Where sUSDS and the Sky Savings Rate Fit

Chart showing sUSDS supply growing from $2.22 billion in Q2 2025 to $5.52 billion in Q2 2026, up 149% year over year, alongside $250 million cumulative yield accrued to holders and $107.35 million Gross Protocol Revenue in Q2 2026.
Where the money is moving. Yield-generating stablecoins are gaining share while the wider category contracts.

The yield has to come from somewhere real, and that is the part worth understanding.

The Sky Agent Network is a group of independent capital allocators that access USDS liquidity and deploy it across diversified strategies, paying a Base Rate back to the protocol.

sUSDS is the yield-generating stablecoin that gives access to the Sky Savings Rate funded by that revenue.

It closed Q2 2026 at $5.52 billion, up 149% year over year. Cumulative yield accrued to sUSDS holders has passed $250 million since inception.

One structural detail most people get backwards: sUSDS holders access the Sky Savings Rate. They are not exposed to any single Agent’s performance. Losses run down the waterfall above, not into the rate.

The Sky Savings Rate itself is variable and set by Sky Governance, funded from Sky Protocol revenue rather than from a marketing budget.

It is published live rather than promised, which is a meaningful difference when rates move.

The Track Record Nobody Can Fake

  • Solvent through Black Thursday in March 2020, when ETH fell more than 60% in hours.
  • Zero exposure to the UST collapse and the FTX bankruptcy. Governance had never approved either as eligible collateral.
  • Held through the SVB week in March 2023, when depeg pressure reached the Peg Stability Module and the peg was restored without an emergency.
  • Zero exploits on the core protocol across seven years.
  • S&P Global assigned a B- rating in 2024, the first structured finance credit rating given to an onchain protocol.
  • Contracts under ongoing review by Certora, ChainSecurity and Cantina.

The economics are public too. Sky Protocol generated $107.35 million in Gross Protocol Revenue in Q2 2026, up 10.5% year over year, with a fifth consecutive quarter of Net Protocol Surplus.

For the wider context on why supervisors keep circling this category, the Bank for International Settlements remains the clearest read.

So, How Safe Are Stablecoins?

Not safe as a category. Some are about as safe as onchain dollars currently get. Some are a spreadsheet somebody is quietly hoping you never open.

The difference has never been the word printed on the token. It is whether the backing is real, visible, and larger than the liability, and whether somebody wrote down in advance what happens when things go wrong.

You can check all of that in roughly four minutes. Almost nobody does.

Which of those five questions have you actually asked about the stablecoin sitting in your wallet right now? Drop the token and your honest answer in the comments. I want to know how many of them pass all five.

Published by Sky Frontier Foundation. Nothing here is financial advice. Rates are variable and set by Sky Governance. Verify every figure at financial.skyeco.com.


Are Stablecoins Actually Safe? A Straight Answer to an Awkward Question was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.

❌