Singapore's central bank published its stablecoin consultation paper on Tuesday, September 1, and I read it between two founder calls the same morning. Most of it was expected. Reserves of at least 100% of tokens in circulation, custodied separately at licensed institutions, redemption in fiat within five business days. Sensible plumbing, written by people who have watched a redemption queue break.
Then there's the clause about interest. Issuers may not pay holders interest, or any other benefit tied to holding the token, because the Monetary Authority of Singapore wants these things used for payments rather than as investments.
That clause is the one worth sitting with.
It isn't a Singapore idea. The GENIUS Act says it in Washington, MiCA said it first in Brussels, and as of last week the three largest stablecoin rulebooks on earth have converged on one sentence: a payment token may not pay you. What almost nobody has written down is the obvious follow-on question. And if roughly $300 billion of dollar-denominated balances can no longer earn anything inside the wrapper they currently sit in, where exactly do those balances go?
Yield is not a product feature you can legislate away. It's a floor rate attached to a Treasury bill, and it exists whether or not a token is permitted to pass it through. So the interesting work in 2026 isn't fighting the ban. It's building the thing on the other side of it.
What the three rulebooks actually agree on
The convergence happened faster than most people in this industry expected, and it happened without any formal coordination between the regulators.
Singapore's proposal, out for consultation until October 16, 2026, has no implementation date attached yet. The American clock is much closer. GENIUS takes effect the earlier of January 18, 2027 or 120 days after the federal banking regulators finalise their rules, and permitted issuers are flatly barred from offering holders any form of yield or interest. In Brussels it landed first, back in 2024, when MiCA prohibited interest on both e-money tokens and asset-referenced tokens.
Same rule, three jurisdictions, one shared motive. Regulators are not worried about crypto here. They're worried about a yield-bearing dollar instrument sitting outside the insured banking system and pulling deposits out of it — a fear that has nothing to do with blockchains and everything to do with what happened to American thrifts when money-market funds arrived in the late seventies.
I keep coming back to that analogy because the industry treats these rules as a crypto crackdown, and they are not. They're a deposit-defence measure. Once you read them that way, the design constraints they create become much easier to predict.
The yield didn't get banned. The wrapper did.
Here is the distinction the headlines flattened, and it's the whole piece: no regulator has said a tokenized dollar cannot earn. They've said a payment token cannot earn.
Those are different sentences with completely different consequences. A tokenized money-market fund is a security. It's allowed to distribute the yield on its underlying paper, because that's what a fund does and there's eighty years of law describing how. Tokenized Treasuries and money-market products crossed $10 billion by May 2026, inside a broader tokenized real-world-asset market that reached roughly $33.5 billion by early July, up from about $12 billion a year earlier. That growth happened before any of these bans bit.
I wrote about this shift back in April, in Bridge Notes #4, when the flippening in what "crypto rates" means was still an argument rather than a chart. The regulatory convergence since then has turned my thesis from a preference into something closer to a forced migration. Every basis point that can no longer legally live in USDC has somewhere obvious to go, and the somewhere is a fund share that settles on the same rails, in the same wallet, with a different legal wrapper and a different regulator attached to it.
But a fund share is not a payment instrument. You can't hand one to a merchant and expect the transaction to clear the way a stablecoin transfer clears, and most compliance departments will not let you try.
So the bridge that has to get built — and this is the actual opportunity — is the one that lets a balance sit in the yield-bearing wrapper by default and convert into the payment wrapper at the instant of spending, atomically, without the user thinking about which of the two they're holding. Sweep accounts did this for retail brokerage in the nineties and nobody called it innovation, because the two legs of the trade sat inside one broker-dealer, under one regulator, on one ledger, and the whole mechanism amounted to a nightly batch job that swept idle cash into a fund and back out again before the customer noticed either leg. On-chain, in 2026, the same idea has to work across two legally distinct instruments, three rulebooks that were written independently of each other, custody arrangements that were designed for a world where the token was the terminal asset rather than a temporary state, and a settlement window measured in seconds rather than overnight. Which is usually where the good businesses are.
Does banning the wrapper actually move the deposits?
Probably not, and the regulators seem to know it, which is why the American proposal went further than the statute did.
The OCC's implementing proposal extends the interest prohibition beyond issuers to affiliates and third parties. Read that as what it is: an acknowledgement that the first workaround — issuer stays clean, exchange or distribution partner pays the "reward" — was already priced in by everyone who read the original bill. Closing it at the affiliate level is the regulator playing the second move rather than the first.
What it can't reach is the fund. A tokenized money-market share distributing its own yield isn't an affiliate arrangement dressed up as a rebate, and treating it as one would require rewriting securities law rather than banking rules. That's the seam, and seams are where capital flows.
Where the analogy breaks
I don't want to oversell the sweep-account comparison, because two things about it are genuinely uncomfortable.
The first is friction that regulation doesn't remove. A tokenized fund share carries transfer restrictions, eligibility checks and, in most structures today, a whitelist. Retail users in emerging markets holding stablecoins as a dollar-savings instrument, which is the largest real use of this technology and the one I care most about, will not clear those checks. For them the yield ban isn't a migration. It's a subtraction, and I haven't seen anyone propose a serious answer to that.
The second is that I've been wrong on regulatory timing before. In late 2024 I told a founder that MiCA's interest prohibition would push European stablecoin volume offshore within a year. It didn't. Volume stayed, issuers absorbed the economics, and the behaviour I predicted showed up in a different place entirely. Rules move slower than my models say they will, and users move slower than the rules.
What founders should do with this
Stop building products whose unit economics depend on passing reserve yield to a token holder. That business has a legislated end date in at least three jurisdictions and the American one is sixteen months out at the latest.
Build the conversion layer instead. The teams I'd back into 2027 are the ones treating the yield wrapper and the payment wrapper as two states of the same balance, with the switching cost driven toward zero — custody that holds both, settlement that swaps between them at spend time, compliance that survives all three rulebooks at once rather than picking a favourite. Across the 60-plus companies I've backed and the ones we look at now at PRIM3, the pitches that get my attention this quarter are the boring infrastructure ones, not the yield ones. That's a reversal from where I was in January.
And if you're issuing a tokenized instrument, assume the wrapper you choose is a permanent architectural decision rather than a launch-week detail. The project I co-founded, SHIFT, sits on the securities side of that line by design — Series Tokens issued under a Marshall Islands DAO LLC, custodied one-to-one, attested by Chainlink Proof-of-Reserves — and picking that structure cost us months we would not have spent if we'd optimised for speed. I'd make the same choice again. Disclosure noted: I have a stake in that outcome.
FAQ
Does the GENIUS Act ban all yield on tokenized dollars? No. It bars permitted payment stablecoin issuers, and under the OCC's proposal their affiliates and third parties, from paying holders interest on a payment stablecoin. Tokenized money-market funds and other securities are regulated separately and can still distribute the yield of their underlying assets.
When does Singapore's rule take effect? The MAS consultation closes on October 16, 2026, and no implementation date has been published. The American deadline under GENIUS is firmer: January 18, 2027, or 120 days after final regulator rulemaking, whichever comes first.
Why do regulators care whether a stablecoin pays interest? The concern is deposit flight. A yield-bearing dollar token held outside the insured banking system competes directly with bank deposits, and all three frameworks are written to keep payment tokens functioning as payment substitutes rather than savings products.
Every rule that closes one wrapper opens the next, and the businesses worth building sit in the gap between them.