The number that circulated was 4.4 billion. On the surface it read like a wall of dry powder: Sky Protocol, the entity formerly known as MakerDAO, disclosing a Peg Stability Module inventory exceeding $4.4 billion in USDC. Headlines translated it into "liquidity." Traders read it as "backstop." Almost nobody asked the only question that matters at settlement — when every USDS holder wants dollars in a bank account on the same afternoon, does the wall hold, or does it merely describe a room you cannot leave?
I have spent enough time on the receiving end of these disclosures to distrust round numbers. In 2017, while auditing the token distribution logic of fifteen pre-sale ICOs, I learned that the most impressive figure in a whitepaper is almost always the least auditable one. The $4.4 billion is real. It is also answering a different question than the market thinks it asked. The alpha isn't in the marketing; it's in the silenced code.
Context: what the PSM actually is
To understand why the number is misleading, you have to separate two things the market treats as one: the ability to convert a token inside a protocol, and the ability to exit the crypto system into a bank. Sky's PSM, running since 2020, is a classic "protocol-as-market-maker" design. Instead of routing swaps through an automated market maker with liquidity providers, slippage, and pool fees, the protocol holds its own inventory of USDC and quotes a fixed 1:1 rate against USDS. There is no external counterparty on the other side of the trade. The safety of the exchange depends entirely on the health of Sky's own balance sheet, not on the depth of some third-party pool.
That architecture matters for cost. A user swapping USDC for USDS through the PSM pays zero fee and eats zero slippage. A user taking the same route through an intermediate token pool pays the pool's pricing and the pool's fee. Same asset, same direction, different cost — because the routing is different. The exit price a user actually receives is a function of the path, not of the token. Most retail never notices this; institutional desks price it every day.
The second structural fact is the DAI–USDS converter. DAI and USDS share the same issuance source. Converting between them creates no new asset, admits no new collateral, and consumes no liquidity — it is a change of unit of account, an accounting-layer operation wearing the costume of a liquidity event. When supply statistics move because of this converter, nothing has entered or left the system.
Then there is the wider token set: USDS, DAI, USDD, and the legacy USDDOLD. Four tickers, overlapping issuance logic, a migration path, and a vault-borrowing path. On July 8, 2025, Sky's USDC–USDS route was integrated into a major DEX router. Since then, cumulative volume on that route has reached roughly $1.6 billion over about three months — a daily average in the seventeen-million range. That is genuine adoption. It is also a flow figure, and flow is not stock. I have watched this category of confusion destroy more positions than any exploit.
Core: a five-path accounting framework for stablecoin supply
The single most useful contribution buried in the Sky disclosure is not the $4.4 billion. It is the implicit demand that anyone reading a stablecoin supply chart first decompose where that supply came from. Stablecoin supply growth is not a synonym for capital inflow. It is the sum of five economically distinct paths, and only one of them brings new dollars into crypto.
Path one: Circle Mint. Bank dollars, held by a qualified corporate account, are wired in and USDC is minted directly against them. This is the only path that introduces genuinely new external fiat. It is the only path where supply growth corresponds to money entering the system from outside. Everything else is internal reallocation.
Path two: PSM conversion. USDC and USDS swap at 1:1 inside the protocol. No new external dollar appears. One holder's position is exchanged for another's. This is a transfer between existing participants, dressed up as activity.
Path three: the DAI–USDS converter. Shared issuance source, no liquidity limit, no new collateral. This is pure unit-of-account change. If USDS supply rises while DAI supply falls by the same amount, nothing happened except a rebrand migrating across the books.
Path four: USDD vault minting. Collateral is locked and a stablecoin debt is minted against it. This is on-chain credit creation, and it is structurally isomorphic to how commercial banks create deposits when they make loans. It does not require a new bank deposit for every new token. It requires a borrower and a collateral position. The spendability of that minted stablecoin is bounded by the instantly convertible inventory, not by the collateral's mark-to-market valuation.
Path five: USDDOLD migration. An equal quantity of new tokens replaces old ones. This is a position upgrade, a database update with a new name field. It is not issuance in any economic sense.
Line those five paths up and the market's favorite heuristic collapses. "Stablecoin total supply is rising, therefore capital is entering crypto, therefore price is going up" treats a five-variable identity as a single signal. In the current sideways tape, where direction is scarce and every participant is hunting for a leading indicator, that error is expensive. If USDS supply growth is entirely composed of DAI migration and internal PSM conversion, then the "inflow" the market is pricing is zero new external dollars.
Here is the part that should worry anyone running a supply-based signal: the cumulative $1.6 billion figure that got quoted as evidence of adoption includes double-sided, repeated conversions. A single holder bouncing USDC to USDS and back inflates the cumulative number while netting to nothing. Flow metrics count gross activity; net metrics count real change. The former cannot prove the latter, yet the two are constantly conflated in dashboards that were built for convenience rather than correctness.
I ran into the same trap in a different costume during the 2020 DeFi Summer. A Python script I wrote to track pool inefficiencies across two AMMs surfaced a $2.4 million arbitrage caused by a delayed oracle update. That trade worked precisely because I separated the real mispricing from the surface-level activity that everyone else was staring at. The lesson transferred directly: the headline number is rarely the signal. The routing underneath it is. Correlations are the lie; liquidity is the truth.
The deeper point is that the 4.4 billion figure itself is a confidence-display behavior. It says: "We can absorb exits." But the number is the size of inventory on a specific path, not the size of the redemption the system can settle in full. Those are different quantities, and the gap between them is where the risk lives. Total collateral valuation is not the same as instantly redeemable inventory — and that distinction is the most frequently ignored metric in the entire stablecoin sector.
The DeFi angle nobody wants to price
Watch what is happening on the competitive side. Aave's GHO stablecoin pool has been running into liquidity scarcity, forcing the protocol to raise rates to attract capital back. This is not a coincidence; it is the same underlying problem wearing a different ticker. When a stablecoin's exit quality depends on the health of its own internal inventory, the interest-rate model that governs that inventory becomes a political instrument rather than a market signal. The rates these protocols quote have only a loose relationship to real supply and demand for dollars — they are administered numbers, tuned to defend a peg or refill a pool, and reading them as price discovery is a category error.
The competitive picture is now clear. Sky holds a PSM inventory above $4.4 billion and a DEX route that has processed roughly $1.6 billion cumulatively. Circle controls the only compliant direct fiat mint-and-redeem channel. Aave's GHO is scrambling to restore pool depth. What looks like a healthy competitive field is actually a hierarchy: everyone is fighting over the quality of the real exit channel, and only one participant owns the door to the banking system.
Contrarian: the 4.4 billion answers the wrong question
Here is the counterintuitive claim. The $4.4 billion disclosure is not primarily an operational fact. It is a narrative asset. It answers "what inventory sits on one exit path" and quietly invites the reader to conclude "the system can settle all redemptions." Those two statements are not equivalent, and the disclosure's framing depends on the reader not noticing.
Three specific overestimates are baked into the popular reading.
First, supply growth is read as new capital. As the five-path framework shows, most supply growth paths — PSM conversion, DAI migration, vault minting, token migration — produce no new external dollars at all. The market is systematically optimistic on a metric that is structurally upward-biased.
Second, the 4.4 billion is read as freely withdrawable user liquidity. It is not. It is bounded by the actual USDC inventory reachable on the routing path. If redemptions concentrate, the protocol faces peg pressure the moment that inventory drains, and the "44" is the ceiling of conversion capacity, not a guarantee of settlement.
Third, cumulative volume is read as adoption evidence. It is not. It double-counts round trips and cannot distinguish genuine growth from churn. The $1.6 billion is a flow statistic; the market wants it to be a stock statistic. It will not cooperate.
And the zero-fee, zero-slippage setting on the PSM is a fair-weather price. A fee of zero and a slippage of zero are not permanent commitments; they are the terms the protocol can afford when redemptions are light. Under stress, the rational response is to introduce fees or caps. Reading today's free conversion as tomorrow's exit assumption is exactly the kind of extrapolation that turns a benign statistic into a broken risk model. Due diligence is the only hedge against chaos — and due diligence here means asking what the terms look like on the worst day, not the best one.
There is a quieter structural risk underneath all of this. No matter how many on-chain paths exist, the final conversion back into bank dollars must pass through a compliant channel like Circle Mint, with its qualified-account gatekeeping and its AML and sanctions screening. The rich graph of on-chain conversion routes funnels down to a single choke point in the banking system. The chain's apparent freedom sits on top of a permissioned base layer that most participants never see, because they never try to leave.
Takeaway: watch the door, not the room
The only supply figure that cannot be faked by internal looping is the net mint-and-redeem flow through the compliant fiat channel. That is the number to track. If large net redemptions appear there, that is a real capital-exit signal, and it is stronger than any on-chain statistic, because it is the one place where the ledger and the bank statement have to agree. The ledger remembers what the marketing forgets.
So the questions worth carrying into next week are these. When USDS supply prints higher, will anyone bother to check whether DAI fell by the same amount — and if they did, what would that tell them about how much of the "growth" was ever money? When the PSM quotes zero fees, is that a price or a promise — and which one breaks first when the queue forms? And when the market finally demands a leading indicator it can trust, will it keep staring at the room, or will it learn to watch the only door out?
Scarcity is an algorithm, not a belief system. In stablecoins, the scarcest thing is not supply. It is an exit.
