The GENIUS Act's Real Deadline Is 2028, Not 2027
The headline number in Treasury's new stablecoin rulemaking is January 18, 2027. That is the wrong date to be watching.
From that day, under proposed rules released this week, no one may issue a payment stablecoin in the United States without a federal or state license. It is a clean, quotable deadline, and every trade publication led with it. It is also the easier of the two obligations buried in the proposal, because the population it binds is small, well-capitalised, and already lobbying.
The date that will actually rearrange the market is July 18, 2028. On that day, digital asset service providers may no longer offer or sell payment stablecoins to persons in the United States unless those coins were issued by a licensed issuer.
Read that again with an operator's eyes. The first deadline regulates who may print. The second regulates who may distribute — and distribution is where the money and the users are.
Enforcement flows downhill to whoever holds the customer
Regulating an issuer is hard when the issuer is offshore, well-lawyered, and has no US entity to serve papers on. Regulating the exchange that lists its token is trivial, because the exchange has an address, a banking relationship, and a licence it does not want to lose.
That asymmetry is the entire design. Treasury is not really trying to compel foreign issuers to submit. It is building a chokepoint at the point of sale, then giving the industry eighteen months of runway after the licensing regime opens to clear the shelves.
Every US-facing venue — centralised exchanges, brokerages, custodial wallets, payment processors, fintech apps with a buy button — becomes a compliance surface for someone else's balance sheet. They will need to know, per token, whether an issuer holds a licence, and to keep knowing it as licences are granted, amended and revoked.
Nobody has built that plumbing yet. It does not exist as a product.
The foreign issuer clause is a negotiating position
The proposal handles overseas issuers with deliberate care. Foreign-issued payment stablecoins generally may not be offered or made available by service providers unless the issuer can comply with lawful orders and there are reciprocal arrangements with US authorities.
Reciprocity is doing enormous work in that sentence. It means the question of whether a given foreign stablecoin remains buyable by Americans in 2028 is not fully a question about the issuer's reserves. It is partly a question about the issuer's home jurisdiction and whether that government has struck a deal with Washington.
The largest dollar-denominated stablecoin in the world is not issued in the United States. Everybody reading the proposal knows which balance sheet the reciprocity language is aimed at, and everybody also knows that the same issuer has spent two years buying US Treasuries, hiring Washington counsel, and positioning a domestic product for exactly this eventuality. The clause is less a ban than a lever.
Levers get pulled during trade disputes. That is worth sitting with, because it means the composition of the US stablecoin market in 2029 may depend on diplomacy that has nothing to do with crypto.
Definitions are where the fight will actually happen
Treasury framed the proposal as clarifying when a licence is required and how payment stablecoins may be marketed. That framing is generous. Definitional rulemaking sounds procedural and is anything but — the boundary of "payment stablecoin" determines which products fall inside the regime and which escape it entirely.
Yield-bearing tokenised cash equivalents have grown enormously this year while the plain stablecoin float has shrunk. If a tokenised money market fund share can be spent like a dollar but is regulated as a security rather than a payment stablecoin, capital will find that door quickly. The industry has demonstrated, repeatedly, that it optimises for whichever legal category carries the lightest burden.
The comment period runs 60 days from Federal Register publication, with submissions posted publicly on Regulations.gov. Two months to argue about definitions that will govern a market measured in hundreds of billions.
Expect the comment file to be enormous and expect the substantive fights to be about edges, not principles. Nobody is going to argue that stablecoin issuers should be unlicensed. Everyone is going to argue that their particular instrument is not a payment stablecoin.
The timing is politically convenient
This proposal lands into a soft market rather than a euphoric one. Stablecoin market capitalisation has fallen from roughly $321 billion in May to around $305 billion — about a 5% drawdown, and by one count the third largest contraction the sector has recorded. Bitcoin is grinding in the low sixties. The Fear and Greed Index sits in fear.
A shrinking market makes for quieter opposition. Firms fighting for survival lobby less effectively than firms printing money, and a Treasury that wants a demanding rule finalised has better odds proposing it now than it would have had at the top.
There is a fair counterpoint, and it belongs in the record. Licensing regimes with distribution bans are how mature financial markets have always worked, and the alternative — an open shelf where any issuer can reach American retail through any app — is precisely the arrangement that produced the collapses regulators cite. Rules that inconvenience incumbents are not automatically bad rules. Much of what Treasury is proposing is unremarkable by the standards of any other regulated payments market.
But two dates in the same statute create a gap, and gaps get arbitraged. Between January 2027 and July 2028 there will be a window in which licensed and unlicensed stablecoins circulate side by side, with the licensed ones carrying the compliance cost and the unlicensed ones carrying the yield. Eighteen months is long enough for that spread to matter and short enough that nobody will bother rebuilding their treasury operations twice.
One statute, two licensing doors
There is a second structural quirk that has drawn far less attention than it deserves. The regime contemplates both federal and state licences, which means the answer to "is this issuer licensed" will depend on fifty-plus supervisory regimes with different examination standards, different capital expectations and different appetites for enforcement.
State chartering has a long history in American finance, and most of it is a history of regulatory arbitrage. Trust charters in particular have been the preferred vehicle for crypto firms seeking a supervisory badge without a bank's obligations. If one state builds a fast, cheap, permissive path to a stablecoin licence, that state becomes the industry's front door, and the federal standard becomes the option nobody chooses.
Treasury cannot fix that through a definitions rulemaking. Congress wrote the dual track into the statute. Whether it holds will depend on whether federal supervisors are willing to publicly disagree with a state banking commissioner, and they usually are not.
What builders should be doing with the runway
The practical work is unglamorous. Anyone running a US-facing venue needs to start treating issuer licence status as a first-class attribute of every asset it lists — tracked, versioned, and auditable — rather than something a compliance analyst checks by reading a press release. The venues that treat this as a data engineering problem in 2026 will onboard the 2028 regime quietly. The ones that treat it as a legal memo will be delisting assets in a panic during the last quarter before the deadline.
There is a version of 2029 where this all works: a handful of licensed issuers, reciprocal recognition deals with the EU and the UK, dollar tokens moving through regulated rails at negligible cost, and the offshore float shrunk to the jurisdictions that chose not to negotiate.
There is another version where American users simply hold the unlicensed coin through a non-US front end and Treasury discovers what every gambling regulator eventually discovers about chokepoints. The proposal assumes the first outcome. It never explains why the second one will not happen.