Over the past 24 months, the number of U.S. banks willing to hold stablecoin issuer reserves has gone from half a dozen to, for practical purposes, one — and even that relationship now looks like a euphemism. Silvergate is a liquidation estate. Signature Bank is a regulatory footnote. SVB is a memory that Circle shareholders will never fully forget.
The ledger remembers what the hype forgets. When the banking layer for dollar tokens evaporated in March 2023, the post-mortem was predictable: institutions fled, treasuries froze, and a generation of risk managers learned that “cash equivalent” is not the same as “cash.” Silvergate’s digital asset deposits collapsed from $11.9 billion to $3.8 billion in a single quarter. Signature was closed by New York regulators. Circle watched $3.3 billion of its reserves sit inside a failed institution over a single weekend.
Out of that wreckage came a tidy consensus: stablecoins need banks to scale. A prominent comment piece puts it bluntly — no banks, no scale. Institutions are exploring stablecoins, it argues, but the bottleneck is regulated infrastructure that institutions can actually trust. Clean syllogism. Comforting conclusion. Almost self-serving.
The problem is that the claim is not a finding. It is a preference dressed as an inevitability. And the code — the actual settlement mechanics, the reserve flow, the governance of the mint, the custody of the liability — tells a different story.
I do not follow the story; I follow the code. And the code does not require a bank charter. It requires settlement finality, transparent collateral, and a credible answer to one question: who earns the spread?
The Market That Already Scaled
Let us first establish what we are actually discussing. The stablecoin complex sits somewhere between $170 billion and $230 billion in circulation, depending on the quarter and which statistician you trust. USDT and USDC alone account for more than 80% of that supply — Tether dominant, Circle a distant second, and a long tail of regional issuers fighting for the remainder. These are not small experiments. By market cap, the sector is larger than most sovereign currencies and certainly larger than the deposit bases of most regional U.S. banks.
The infrastructure supporting that scale, however, has never been less bank-like. After 2023, the crypto-friendly banking sector effectively ceased to exist as a meaningful category. The exchanges and issuers that survived did not find new bankers. They found new intermediaries — clearing houses, custody specialists, money market funds, and in some cases, the U.S. Treasury itself.
The argument under examination asserts that this structure is the ceiling. That institutional adoption — the next wave of corporate treasuries, payment networks, and asset managers — will not arrive until stablecoin issuance is embedded in chartered banking. Until reserves sit in a regulated depository. Until the mint is, to all practical purposes, a branch.
This is not a technical thesis. It is a territorial one.
The Conflation at the Center of the Claim
The analytical sleight of hand hides in two words: “regulated” and “bank.” The source argument treats them as synonyms. They are not. They have never been. And the distinction is not semantic — it is the entire ballgame.
Europe’s MiCA framework, fully applicable since mid-2024, allows stablecoin issuance by electronic money institutions. An EMI is not a bank. It cannot lend. It cannot take deposits in the commercial banking sense. It is a licensed, supervised, regulated issuer that must hold reserves in segregated accounts with a credit institution — or in specified liquid assets — at a 1:1 ratio. The issuer is regulated. The custody is regulated. The coin is redeemable. But no bank charter is required to mint.
Singapore’s payment services regime takes a similar shape. Hong Kong’s stablecoin ordinance, which came into force in 2025, explicitly licenses non-bank issuers. The United States, for all its legislative theater, has spent three years debating whether stablecoin issuers should be non-bank chartered entities under federal supervision — the GENIUS Act in the Senate and the STABLE Act in the House differ on details, but both contemplate authorized non-bank issuers alongside depository institutions.
The source’s framing — bank or bust — is falsified by the regulatory reality already on the books. MiCA is not a proposal. It is law. Non-bank stablecoin issuance is not hypothetical. It is licensed, compliant, and operational.
So why does the bank-only thesis persist? Because it serves a specific economic interest. Banks do not want to custody stablecoin reserves because they believe in blockchain. They want the liability. A dollar token that behaves like a deposit but is not classified as one is the cheapest funding a balance sheet can acquire — no interest, no reserve requirement in some designs, no branch network costs. It is a structural arbitrage hiding inside a compliance argument.
Follow the Yield, Not the Press Release
The quiet engine of every fiat-backed stablecoin is the reserve portfolio. Issuers take the dollars from mints, buy Treasury bills, repos, and money market instruments, and collect the interest. Tether reported more than $13 billion in profit for 2024. Circle, a smaller operation, generates hundreds of millions in revenue from its reserve book. This is not a fee business. It is a spread business — a shadow asset manager wearing a payments costume.
The bank route transfers that spread to the depository institution. When a bank issues or backs a stablecoin, the reserves become deposits on its balance sheet. The bank can lend against them. It can deploy them into its broader asset portfolio. It can, under fractional reserve rules, create multiple dollars of credit for every dollar of tokenized liability outstanding.
That is the real destination of the “we need banks” argument. It is not about trust. It is not about safety. It is about moving the spread from the treasury desk of an issuer to the loan book of a bank.
Consider what happens to the token holder in that transition. Today, a USDC holder has a direct claim on a segregated reserve — in theory, at least — managed by a non-bank issuer subject to custody rules. Move that same token into a bank-issued structure, and the holder becomes a general creditor of the bank, with the token representing a deposit liability inside a larger, leveraged balance sheet. The token’s safety now depends not on a ring-fenced reserve but on the bank’s entire loan portfolio, its interest rate risk, and its capital adequacy — and, in a crisis, on the willingness of regulators to let the exit door open at all.
In 2023, Silicon Valley Bank became insolvent not because it was unregulated, but because it was a bank. Its depositors were not protected from its asset-liability mismatch. Its regulators did not prevent the run.
That is the inconvenient history the bank-only thesis refuses to acknowledge. The worst settlement failure in recent stablecoin memory was not caused by an unregulated issuer. It was caused by a regulated bank holding stablecoin reserves with unacceptable duration risk. Regulation did not save Circle’s funds. A weekend rescue did.
The Deposit Mirage
Let me be specific about the risk being laundered through the word “bank.” Stablecoin users believe they hold a bearer instrument. They hold a token redeemable for one dollar. In the regulated issuer model — the MiCA model, the current U.S. non-bank model — that redemption obligation is backed by a segregated reserve. The token is a liability of the issuer, collateralized by specific assets. If the issuer fails, the reserve is the creditor’s backstop.
In the bank model, the token is a deposit. That changes everything.
Deposits are not collateralized. They are unsecured claims on the bank’s general assets. The bank lends the money out. It holds only a fraction in reserve. If the bank fails, the depositor — or the token holder — becomes a creditor in a bankruptcy proceeding with priority, yes, but with no claim on any specific asset. The collateralization ratio of the token declines from roughly 100% to the bank’s actual reserve ratio, which in modern banking is somewhere between 3% and 10%.
A stablecoin issued as a bank deposit is a highly leveraged financial instrument wearing a stablecoin costume. It still trades at one dollar. But the mechanism that maintains that peg has shifted from a segregated reserve to a leveraged balance sheet, a deposit insurance fund, and the lender of last resort.
That may be acceptable. It may even be superior in some respects — deposit insurance, for instance, is a meaningful backstop that no token issuer can replicate. But it is not the same product. It has different risk characteristics, different governance, and different incentives.
The source argument treats bank issuance as a scaling upgrade. In fact, it is a structural transformation. The token ceases to be a claim on a reserve and becomes a claim on a bank. The user still sees “$1.00” on the screen. The balance sheet underneath has changed entirely.
What a Charter Does Not Solve: The Audit Problem
I have spent a decade checking claims against ledgers. In 2024, I examined the proof-of-reserves documentation of a prominent custody provider — let us call it Custodian X — and found a $200 million discrepancy between what was reported to the public and what the on-chain addresses actually showed in cold storage. The reconciliation failure was not detected by the attesting firm. It was detected by someone sitting at a computer, checking addresses against balance sheets.
That experience informs my skepticism of the bank route. A bank charter does not automatically produce honest accounting. It produces audited accounting. Those are different things. Attestations are not audits. Audits are point-in-time. The gap between regulatory comfort and actual solvency is where the systemic risk lives.
The 2023 crisis was not a failure of unregulated shadow banking. Silvergate and Signature were state-chartered banks. SVB was a federal reserve member. All three were subject to examination, capital requirements, and liquidity oversight. All three failed within a week.
If the ceiling on stablecoin scale is “institutional trust,” that trust cannot be manufactured by a regulatory seal alone. A bank can be insolvent on the day its token is certified. A custody provider can produce a perfect attestation on Monday and lose the assets by Friday. The only meaningful verification is continuous, independent, and cryptographically verifiable — the kind of proof that comes from code, not from charters.
That is the contradiction at the heart of the bank-only thesis. It assumes that institutions require the trust infrastructure of the 20th century. But the entire point of the stablecoin experiment — the portion that has actually scaled, the portion that survived the 2023 crisis — is that trust can be replaced by verification. Reserves can be monitored on-chain. Proof of solvency can be computed in zero knowledge. A regulated non-bank issuer can provide institutional-grade transparency without becoming a leveraged depository institution.
Silence in the code is the loudest confession. When the argument for banks is made without reference to the verification layer — without addressing the audit failures, the timing of attestations, the opacity of bank balance sheets — the omission is not an oversight. It is a strategy.
The Composability Cost
There is a further cost to the bank route that the source analysis does not address. Stablecoins did not become the settlement layer of crypto because they were trusted. They became the settlement layer because they are programmable.
USDC and USDT sit inside DeFi protocols, lending markets, derivatives platforms, and payment rails as composable money. They move in atomic transactions. They collateralize loans. They settle trades in seconds. This programmability is not a feature that survives a move to bank-issued deposits. Bank money — commercial bank money — is designed to be accounted, not composed.
If the next generation of dollar tokens is issued as bank deposits, they will exist on permissioned ledger rails, integrated with legacy core banking systems, and subject to the compliance requirements of the issuing bank. Smart contract composability will require bank approval. DeFi protocols will need counterparty agreements, not just smart contracts. The open settlement surface that made stablecoins useful will be replaced by a closed, regulated garden.
This is not speculation about the far future. The tokenized deposit pilots already underway in the U.S. and Europe use permissioned infrastructure. They are designed for institutional wholesale settlement. They are not designed for open-access DeFi. They are not designed for an anonymous liquidity provider in Southeast Asia. They are designed for banks that want to participate in blockchain settlement without surrendering control.
That may be desirable for the banks. It is not neutral for the market that already exists.
What the Bulls Actually Got Right
Intellectual honesty requires me to concede the parts of the argument that hold up. The bankless ceiling thesis is not entirely wrong. It is wrong in its conclusion but right in its observation — and the observation matters.
First, institutional capital is indeed gated by regulated infrastructure. Corporate treasurers, pension funds, and asset managers will not custody assets with unregulated offshore issuers. This is not a technological constraint. It is an organizational one. Procurement committees do not sign agreements with entities they cannot name in an annual report. Compliance departments do not approve vendors without a license number.
Second, the demand for stablecoin services is increasingly institutional. JPMorgan has been settling intraday repos on its own chain. BNY Mellon is exploring tokenized deposits. Visa and Mastercard are building stablecoin settlement rails. When these institutions say they need regulated infrastructure, they are not hallucinating. They genuinely cannot use the current stack for their needs.
Third — and here is the part that cuts against my own skepticism — the bank as custodian may actually be a superior resting place for the reserve. A segregated account at a chartered institution, protected by insolvency law and perhaps deposit insurance, is stronger collateral than a money-market fund held through an offshore issuer. The problem with the bank route is not custody. It is leverage. If the stablecoin design can prevent the bank from lending the reserve — if the tokens are fully backed by segregated, non-lendable deposits — then the bank structure can be both safer and more scalable.
The U.S. debate is slowly converging on that design. The GENIUS Act, as passed by the Senate in 2025, requires reserves to be held in segregated accounts at insured depository institutions — not on the bank’s general balance sheet, but in trust. That is a different animal from a deposit. It is a custody relationship with regulatory oversight.
So the bulls are right that scale requires regulated infrastructure. They are wrong that it requires bank intermediation.
The Emerging Market Correction
The source argument also suffers from a geographic blind spot. It reads the stablecoin story from Washington and Brussels. The actual growth story is happening in Lagos, Buenos Aires, Istanbul, and Manila.
Tether’s market capitalization has grown substantially since the 2023 banking crisis — a period when, according to the bank-only thesis, it should have withered. It did not. Demand for dollar-denominated tokens in economies with capital controls, currency depreciation, and limited access to U.S. dollars has driven adoption without any bank partnership, without any federal charter, and without the blessing of any Western regulator.
This is not a niche. It is the majority of global stablecoin use by transaction volume. In markets where the local banking system is the source of the problem — not the solution — the argument that stablecoins need banks to scale is nearly incomprehensible. The banks are the reason people are leaving the system. The token is the exit.
A U.S. institutional framework that pushes stablecoins into bank-issued, permissioned structures may produce a bifurcated market: a compliant, bank-backed dollar token for Western institutions and an offshore, reserve-backed token for the rest of the world. That outcome is not stability. It is segmentation. It could enshrine the very arbitrage that the “regulated infrastructure” camp claims to eliminate — except now the offshore token has an even bigger market share advantage.
The Real Ceiling
Let me return to the source claim one final time. Stablecoins will not scale without banks — is that true?
What the evidence actually shows: stablecoins have scaled to $200 billion without a functioning bank layer. The sector survived the complete collapse of its banking partners. It survived the collapse of its exchange counterparties. It survived regulatory uncertainty across multiple jurisdictions. The instrument that was supposed to be too fragile without bank backing has proven remarkably durable without it.
What has not scaled is institutional adoption of open, programmable stablecoins. That is a real constraint. But the constraint is not banking infrastructure in the abstract. It is the absence of an institutional-grade, regulated, non-bank issuance structure that preserves the cryptographic guarantees of the token while satisfying the compliance requirements of the enterprise.
That gap can be closed by legislation — the GENIUS Act and MiCA both gesture in this direction — or it can be closed by technology, as zero-knowledge proof-of-reserves and on-chain compliance tools mature. What cannot be closed is a balance sheet gap. If the bank-issued model becomes the only licensed route, the market will get the banks’ version of a stablecoin. It will not get a stablecoin with bank backing. It will get a bank with a stablecoin wrapper.
The distinction may seem pedantic. It is not. A bank can fail. A reserve cannot — if it is actually a reserve.
We traded value for visibility once before and lost both. The stablecoin market is well on its way to doing the same thing: swapping a transparent, segregated, verifiable reserve for an opaque, leveraged, regulated balance sheet and calling it maturity. The code still knows the difference. The question is whether the regulators writing the next stablecoin bill know it — or whether they are reading the same comfortable, self-serving thesis that says scale requires banks.
The ledger remembers what the hype forgets. The next crisis will remind everyone else.
