A compliance lawyer I met at a fintech panel in Miami told me, after the room had cleared and the catering was being broken down, that the GENIUS Act is the best thing that ever happened to two specific people in the industry.
He did not want to be named. The two people are the stablecoin issuer's CFO and the exchange treasury head. Not the holder. Not the saver. Not anyone reading this.
The premise of the law is clean enough that you can write it on a napkin. Issuers of payment stablecoins cannot pay yield to holders. The reasoning is that a stablecoin paying yield starts to look like a money market fund, and money market funds live under a regulatory frame the law explicitly chose not to extend to stablecoins. Cleaner the instrument looks as money, easier it is to defend it as money. That is the official story and it is internally consistent on its own terms.
The unofficial story is what I want to walk through. Because the yield does not stop existing the moment the holder stops receiving it. The Treasury bills sitting in the reserve still pay their coupon. The cash equivalents still earn at the policy rate. The money does not vanish. It just moves down the org chart and onto a different income line.
So I spent the last six weeks tracing where it goes. Not because the answer is technically obscure — it is not, the public 10-Q from the publicly traded issuer makes it visible if you know which line to look at — but because almost nobody in the retail conversation has done the math out loud. And the people who have done the math sit in two specific seats and have no incentive to publicize the number.
The Yield Did Not Disappear. It Was Redirected to the Issuer.
Let me give the strongest version of the counter-argument first, because it deserves to be taken seriously.
The vast majority of stablecoin holders were never receiving yield in the first place. USDC, USDT, USDP — none of them paid yield to the wallet that held the token. The yield-bearing wrappers were a separate category and a small fraction of supply. So the argument from the issuer side is: nobody is losing anything they actually had. The GENIUS Act is just codifying a state of affairs that already described 95% of the float. The number is approximate, I do not have the precise breakdown handy from a grounded source, and I will not invent one.
That concession is real. The structural critique survives it anyway, and here is why.
Before the law, the issuer had at least some competitive pressure to share float yield with the holder. PayPal's PYUSD launched with a rewards program. Several DeFi-native dollar tokens distributed yield directly through rebasing or staking mechanics. The pressure was visible in product roadmaps because the cost of capital for an issuer was implicitly set by what the holder could earn somewhere else. The law removed that pressure. Now every payment stablecoin is statutorily required to look the same on the yield axis, which means none of them have to compete on it. The issuer captures the entire spread between zero (paid to the holder) and the policy rate (earned on the reserve) with statutory cover.
I want to be careful here. I am not saying the issuer's economics are unreasonable. They run real operating costs, real banking relationships, real audit and compliance overhead, and the float yield funds all of that. What I am saying is that the redistributional question — does any of that float yield flow back to the people whose dollars made it possible — is now answered structurally and forever in the negative for payment stablecoins. That is a policy choice, not an accident.
And the policy choice was made without most retail holders understanding it was being made.
The Exchanges Built a New Margin Line and Are Not Itemizing It.
This is the part of the trace that surprised me, and it is the part the Miami lawyer was pointing at when he named the exchange treasury head as one of the two winners.
Exchanges hold enormous stablecoin float on behalf of customers. Binance runs $18.5 billion in daily volume across 1,850 listed pairs, and the stablecoin balances sitting in customer wallets at any given moment are substantial. Bybit runs $9.2 billion daily across 970 pairs. Bitget runs $6.1 billion. OKX runs $4.9 billion. MEXC runs $3.8 billion across 2,400 listed pairs — wider catalog, lower per-pair depth, same structural fact.
The customer's stablecoin balance, when it is sitting idle in the exchange wallet, is functionally identical to a deposit. The exchange is not contractually required to pay yield on that balance in most cases. And the exchange has the same banking and Treasury access the issuer has — sometimes through the issuer's own product, sometimes directly. The float earns. The customer does not see it.
Now layer the staking and earn products on top. Every one of those five exchanges — Binance, Bybit, Bitget, OKX, MEXC — supports staking products on their feature lists. The staking and earn surface is where the exchange selectively redistributes some of the float yield back to the customer, on the exchange's terms, in the exchange's branded product, with the exchange taking the spread between what the underlying earns and what the customer is paid. It is the same yield. Different wrapper. Different distribution rules. Different counterparty risk, because now the customer is exposed to the exchange's solvency, not just the issuer's.
The GENIUS Act does not regulate this layer. It regulates yield paid directly on the stablecoin itself, not yield paid on a custodial product denominated in the stablecoin. The exchange compliance teams know this. The exchange product teams know this. The retail user, for the most part, does not.
So the practical effect is: the law pushed the yield distribution decision from the issuer to the exchange. The exchange now decides who gets to earn and how much, and the exchange takes the spread. That is not a small change. That is the most profitable layer of the stack quietly consolidating around custodial control.
Self-Custody Is the Cleaner Answer Now, and Even the Custodians Know It.
Here is the conclusion I did not expect when I started writing this piece.
If the yield is gone from holding the stablecoin itself — really gone, statutorily gone — then the economic argument for keeping balances on an exchange has weakened significantly. The convenience is still there. The trading access is still there. The fiat ramps are still there. But the implicit float yield that the exchange captures is now the customer's only opportunity cost for self-custody, and the customer cannot see the number.
Ledger and Trezor and the GridPlus Lattice1 will not pay you anything to hold your stablecoin. That is the honest version of self-custody and it is the only version that makes sense in the new regime. The hardware wallet returns nothing because there is nothing to return. The exchange returns selectively, sometimes, on its own product surface, with a spread it does not disclose. The qualified custodian — Coinbase Custody operating as a NY DFS Trust Company, Fidelity Digital Assets operating as a NY DFS Trust, Anchorage Digital operating under the OCC's first federal trust charter for crypto — sits in the middle, offering institutional-grade custody for clients who care about regulatory clarity more than about the basis point spread.
I have been skeptical of self-custody maximalism for a long time. The "not your keys, not your coins" line is true at a categorical level and bad advice for active traders who need exchange access for execution. That has not changed. What has changed is the storage-layer math for the buy-and-hold stablecoin balance, the operational treasury, the savings reserve. For those use cases, the GENIUS Act made the hardware wallet honest in a way it was not honest before. There is no yield you are giving up by moving the balance to cold storage, because there was no yield the exchange was passing through to you anyway, and now there cannot be.
That is the part the custodians know. The Anchorage and Fidelity teams I have read public commentary from have been quietly positioning around qualified custody for institutions specifically because they understand the exchange custodial product just lost its yield-arbitrage justification for the customer who can do the math.
The retail customer cannot do the math yet because the math is not itemized. That is the asymmetry the law created.
This started as a piece about the GENIUS Act and turned into a piece about where money flows when a statute removes the most visible recipient from the distribution. The conclusion I did not see coming: the law made self-custody more economically rational for the storage layer, not less. Whether the issuer-side and exchange-side yield capture will draw its own regulatory scrutiny next cycle — or whether the industry has now successfully insulated the most profitable layer of the stack from public disclosure — is the question I cannot answer with the public filings yet. If you work somewhere this is itemized, write.
FAQ
Does the GENIUS Act apply to yield-bearing stablecoins like sDAI or Ondo's USDY?
The premise of the law is that payment stablecoins cannot pay yield to holders. Whether a given yield-bearing dollar token is classified as a payment stablecoin under the statute, or as a different instrument that lives under a different regulatory frame, depends on the issuer's filings and the regulator's posture in the months following enactment. The clean answer is that products marketed as payment stablecoins are constrained, and products structured explicitly as securities or money market shares are not — but the boundary will be contested.
If exchanges still pay yield on stablecoin balances through earn products, what changed?
The legal locus of the yield decision changed. Before, an issuer could compete on float-yield distribution. Now they cannot. The exchange "earn" product is a custodial offering layered on top of the stablecoin, distinct from the stablecoin itself, and the exchange controls the spread and the eligibility rules. The customer no longer has access to issuer-level yield as a competitive baseline, only to exchange-curated products with exchange-set terms.
Is self-custody on a Ledger or Trezor actually safer for stablecoin holdings now?
Safer is the wrong frame. The risk profile is different. Hardware wallets remove exchange solvency risk and remove the exchange's discretion over your balance. They add operational risk — lost seed phrase, firmware compromise, user error. What changed with the GENIUS Act is the opportunity cost: previously you were arguably giving up some access to yield distribution by self-custodying. Now you are not, because the yield distribution to holders has been statutorily closed.
Are qualified custodians like Coinbase Custody or Anchorage Digital affected?
Qualified custodians sit in a different layer than retail exchange wallets. Coinbase Custody operates as a NY DFS-regulated Trust Company, Fidelity Digital Assets as a NY DFS Trust, and Anchorage Digital under an OCC federal trust charter — the first granted to a crypto-native firm. Their fee structure is explicit and their clients pay for custody directly, so the float-yield-capture dynamic is structurally different from a retail exchange wallet. The institutional customer can see the number.
Where does the float yield show up in the issuer's financials?
For the publicly traded stablecoin issuer, it appears as interest income on reserve holdings in the income statement. The reserves are typically a mix of short-duration Treasury bills, repo, and cash equivalents, and the yield earned on that portfolio funds operating costs and net income. Before the GENIUS Act, there was implicit competitive pressure to share some of that yield with holders. After the act, that pressure is structurally absent.
Does this apply outside the US?
The GENIUS Act is US statute and binds US-domiciled issuers and US persons. Stablecoin issuers operating offshore for non-US users sit under whatever framework their domicile imposes. The practical effect, given that the dominant payment stablecoins are US-linked and used globally, is that the US regime sets the default for most global liquidity. Other jurisdictions — the EU under MiCA, Singapore under MAS guidance — have their own yield rules that may or may not align.
What should a long-term stablecoin holder actually do with this information?
Decide whether the convenience of exchange custody is worth the float yield the exchange now captures without itemizing. For balances you trade actively, the answer is probably still yes. For balances you hold as a savings reserve or operational treasury, the math has tilted toward self-custody or qualified custody where the cost is explicit. The choice is yours and the data to make it is not currently visible to retail, which is itself part of the problem.
Will the yield-capture layer be regulated next?
That is the open question I closed the piece with. Public filings will show the issuer side of the capture if you read them. The exchange side is harder to extract because the spread is not itemized in customer-facing disclosure. Whether regulators turn their attention there in the next legislative cycle depends on whether the disparity becomes politically visible — and as of now, it is not.