The Fed's GENIUS Act Proposal: Reserves, Capital, and the New Perimeter of Trust

I spent three weeks last summer in the Federal Register archives, reading the comment letters that Circle, Tether, and a dozen smaller issuers filed against the first draft of the GENIUS Act framework. What struck me was not the arguments. It was the silence — hundreds of pages devoted to reserve composition, capital buffers, and audit cadence, and not one sentence asking whether the model should exist in the form Congress was designing. The Fed's new proposal formalizes that silence into rule. This week, the Federal Reserve released proposed rules requiring supervised stablecoin issuers to hold reserves, satisfy capital requirements, and apply for admission under the GENIUS Act framework. Three sentences. Three sentences that quietly convert dollar-pegged tokens from crypto-adjacent instruments into bank-adjacent balance sheets — and they will determine whether the next generation of stablecoin issuers is a Circle, a JPMorgan subsidiary, or a two-person team in Lisbon with a smart contract and a dream. For those who arrived late: the GENIUS Act — Guiding and Establishing National Innovation for U.S. Stablecoins — is the federal framework that ended a decade of ambiguity about whether payment stablecoins are securities, commodities, or unregulated money transmission. Passed with rare bipartisan support, it handed the Federal Reserve, the OCC, and the FDIC joint responsibility for defining how supervised issuers operate. The Fed's portion, dropped this week, focuses on what any institution wanting to issue a dollar-backed token must look like from the inside. The proposal rests on three pillars. Reserve requirements: tokens must be backed at least one-to-one by qualifying, high-liquidity assets. Capital requirements: issuers must hold their own equity buffer against reserve impairment and operational loss. Application rules: issuance becomes a licensed activity with an admission process and ongoing supervision. That is the plain text. What matters is what it implies. Remember what stablecoins were. For most of their history they operated in a regulatory vacuum so complete that the industry policed itself with monthly attestations and lawyer blog posts. The GENIUS Act ended that era; the Fed's proposal is the first page of the new one. Here is where I want to be precise, because the difference between "regulatory clarity" and "regulatory restructuring" is exactly the difference between a marketing narrative and a balance sheet. The reserve rule is the least controversial and the most misunderstood. One-to-one backing does not mean "issuer holds cash." It means issuers hold a defined set of assets — almost certainly cash, Fed reserve balances, and short-dated Treasuries — at sufficient scale to redeem every token at par, on demand. This is the same logic that underwrites money market funds, and it is the same conversation I was having in 2020 when I was pulling apart Curve's stablecoin invariant formulas and writing about what trust actually costs inside a 3pool. Back then, trust was a geometric property: the invariant held because the reserves held. Now a regulator is drawing a legal perimeter around the invariant. The capital rule is where the geometry becomes structurally different. Capital requirements are not reserve requirements with a different name. Reserves protect the token holder. Capital protects the issuer's ability to survive a reserve shock — a run, an impairment, an operational failure — without reaching into customer assets. Two claimants, two balance sheet lines, two failure modes. When I worked with mid-sized crypto firms through ChainLogic after the Terra collapse, capital adequacy was the dimension almost every issuer ignored, because the price of ignoring it was zero under a regime with no capital rule. That price is about to become non-zero. The application rule is the quiet one, and it is the one that reshapes the market. "Supervised issuer" is a category, not a description. Being inside it means holding a charter, submitting to examination, satisfying KYC/AML obligations, and operating under a supervisory relationship that no smart contract can replicate. The cost of admission is not a fee; it is an organizational transformation. Translated into market structure, the three rules compose into a filter with a specific shape: wide at the base — many teams can hold Treasury bills — and narrow at the top, where few teams can hold a charter, a capital buffer, and a supervisory relationship at once. That is the shape banking has had for a hundred years. This is not an accident. It is the point. Here is the granular detail most coverage will flatten: reserve assets must be verifiable, and I expect the final rule to push toward real-time or at least per-cycle attestation — chain-native proof of reserves, third-party audit with a defined cadence, and a whitelist of eligible custodians. Open source isn't just a development methodology in this context; it's a philosophy of transparency that regulators are quietly borrowing to make reserves legible from the outside. I flagged this pattern years ago in my audit work: whenever a regulator requires an asset to be "verified," the verification tooling becomes an infrastructure business. Proof-of-reserves services, professional custodians, and audit firms are the silent winners of this rule. Now the harder part. What does this do to the composition of the market? The light-asset playbook — issue tokens, hold reserves, earn the spread, distribute nothing to holders, keep capital thin, register lightly — becomes unprofitable for anyone without a bank-adjacent parent. The heavy-asset playbook — hold capital, hold the charter, hold the custody — becomes the only legal path to scale. Circle's decision to lean into compliance, which looked defensive in 2022, now looks like pre-positioning. A bank subsidiary starting a stablecoin tomorrow begins with a Fed reserve account and a regulator that already knows its management team. That is a structural advantage no DeFi native can bootstrap from a whitepaper. And the yield question — the argument about whether issuers should share reserve interest with holders — becomes sharper, because the capital rule now consumes the margin that might have funded that distribution. Under a capital buffer, an issuer has two claims on interest income: holders and its own equity. Splitting it further becomes arithmetic, and arithmetic tends to win. One detail in the reserve rule will drive more behavior than the headline requirement. The rule almost certainly defines a whitelist of qualifying assets — cash, Fed balances, T-bills within a defined maturity — and the composition of that list becomes the real policy lever. A tight whitelist maximizes safety and minimizes yield; a generous one lets issuers chase spread. The final whitelist is where the lobbying happens, and where the economics of the stablecoin business get decided. One more layer the headlines will miss: the rule is being issued in pieces. The Fed writes its piece (bank-affiliated and supervised issuers within its jurisdiction). The OCC writes another (national trust banks). The FDIC writes a third (insured depository issuance). For an issuer that wants a single national charter and a single compliance posture, the actual question is not "what does the Fed require?" but "which of the three regulators will I end up answering to, and will their answers rhyme?" I have seen this movie. The same tri-agency fragmentation slowed the formation of a coherent national banking market in the 1930s by a decade. In crypto, I would give it three years. For decentralized stablecoins — the DAI/USDS family — the language matters enormously. "Supervised issuer" is a jurisdictional term, not a technical one. A protocol governed by token holders and collateralized on-chain may fail to qualify as a "supervised issuer" at all: neither licensed nor banned, it sits in a gray ring outside the licensing perimeter. Whether that ring holds depends on how downstream use is treated: can a US bank hold it? Can a US exchange list it? If not, the gray ring stops being a moat and starts being a quarantine. Decentralization is not a tech stack; it is a jurisdictional claim. And jurisdictions, unlike blockchains, can decide not to recognize a claim. Which brings us to the arbitrage question the draft doesn't answer. Nothing in the proposed rules prevents a US-facing issuer from restitching its domicile and relaunching as a foreign entity that sells tokens into US markets through third parties. That silence is not an oversight; it is a live question the Fed is passing to Congress. Watch whether the comment period produces language about extraterritorial reach. If not, the compliance moat leaks. And here is the part I keep coming back to. The reserve rule pulls in one direction — toward more dollar exposure in short-dated Treasuries. That is a macro-financial event, not a crypto one. Every new dollar of compliant issuance is a new dollar of demand for the front end of the Treasury curve, from an issuer base that never before had to hold it. I published a piece last year on the correlation between stablecoin float and bill demand; the elasticity is real, and the Fed is now writing a rule that hardens it. What looks like a crypto story is, at the plumbing level, a monetary policy story. Here is the contrarian reading I think most of the industry will get wrong. The dominant narrative is that regulatory clarity is unambiguously bullish. I don't think "clarity" is the right word. We didn't spend a decade building stablecoin rails so a charter could be handed to four banks. We spent it proving a dollar could move without one. What the Fed is doing is closer to nationalization-by-charter of the stablecoin dollar. The Fed is not clarifying a rule; it is claiming a jurisdiction. Look at what the proposal doesn't do: it doesn't give offshore issuers a path, it doesn't bless decentralized issuance, and it doesn't resolve the tri-agency question. What it does do is define the shape of the only issuer the US will recognize as fully supervised — and that shape is a bank. If you run a stablecoin designed to be issued by a foundation in Zug or a DAO in exile, you have not been regulated. You have been excluded. And exclusion, in a market where downstream access is gated by US exchanges and custodians, is a slower and quieter death than a ban. The absence of a rule aimed at you is not mercy; it is a sentence without a date. The second blind spot is time. Rules at proposal stage do not bind. The comment period will pull the capital thresholds down, soften the reserve whitelist, and perhaps carve out space for well-governed non-bank issuers. What is not on the table is the direction. The direction — toward charter, capital, custody — is already set. Trade the direction, not the draft. I keep thinking about the small team I met at a Lisbon hackathon two years ago. They had a stablecoin product, a clean design, and a partnership with a mid-size LatAm bank — exactly the kind of team the application rule is aimed at. I don't know what they will do. But if they want to stay in the dollars business under this rule, they will need a parent with a Fed account, a chief compliance officer, and a capital plan — or they will need to move into a currency where the licensing perimeter has not yet been drawn. The window in which issuing a stablecoin is a technical act closes with this comment period. The window in which it is a banking act is opening. The real question for the next twelve months is not whether the Fed gets the rules right. It is which issuers will still be standing on the inside when the door finishes closing.