The GENIUS Act, Risk, and the Honest Limits

Lesson 1 built the mechanism, Lesson 2 the business case. Both rested on an assumption worth examining: that the issuer is holding what it claims, and will redeem when asked. For most of stablecoins' history, nothing required that. You were trusting a private company's word about reserves it need not disclose in any standard form.

That is what changed. The GENIUS Act of 2025 created a federal framework for payment stablecoins in the US, turning a private arrangement into a defined, supervised instrument.

The significance is easy to state. Lesson 1 showed the peg is a credit and liquidity question, not a technology one. A law that dictates reserve composition, redemption, and disclosure is therefore not adjacent to the mechanism, it is the mechanism, written down and made enforceable.

This lesson covers:

• What the Act actually requires, precisely.
• The yield ban, which quietly determines the entire business model.
• What risks survive the rules.
• What remains unsettled.

One framing to carry: regulation here is not primarily a constraint on stablecoins, it is what makes them usable by serious institutions. A treasurer cannot hold a claim on an issuer whose reserves are unverifiable. The rules are the on-ramp for the very adoption Lesson 2 described.

The GENIUS Act's requirements map directly onto the failure modes of Lesson 1. Take them in order.

Definition. A payment stablecoin is a digital asset issued for payment or settlement and redeemable at a predetermined fixed amount. The redemption promise is written into the definition itself.

Full reserves, 1:1. Issuers must hold at least one dollar of permitted reserves for every one dollar issued. Not fractional, not "substantially backed."

Permitted reserves are enumerated, and the list is deliberately narrow: coins and currency; deposits at insured banks and credit unions; short-dated Treasury bills; repos and reverse repos backed by T-bills; government money market funds; central bank reserves; and similar government-issued assets approved by regulators.

Who may issue. Banks and credit unions through subsidiaries, or nonbanks. But nonbanks are restricted to financial firms unless the Stablecoin Certification Review Committee (SCRC), the Treasury Secretary plus the chairs of the Federal Reserve and the FDIC, unanimously finds they pose no risk to the banking or financial system and will comply with requirements.

Disclosure. Issuers must publish their redemption policy, provide monthly attestations of reserve composition, and supply monthly CEO and CFO certifications of those reports.

Classification. Payment stablecoins are explicitly excluded from securities or commodities designation, resolving years of jurisdictional argument.

Read as a set, the Act is not regulating a technology. It is regulating a promise to pay a dollar, exactly as Lesson 1 predicted it would have to.

One provision looks minor and shapes everything: issuers of payment stablecoins cannot pay interest or yield to customers who hold them.

To see why it matters, follow the money. An issuer holds a dollar of Treasury bills for every token. Those T-bills earn interest. The holder gets none of it. So the issuer's business model is straightforward: it keeps the float. At scale, that is a large, low-risk income stream, earned on money customers gave them at zero cost.

So why ban paying the holder? Because of what a yield-bearing stablecoin would become. An instrument that is redeemable at a dollar, pays interest, and is issued by a nonbank is functionally a bank deposit or a money market fund, without being regulated as either. That would pull deposits out of banks (which lend them into the economy) and into an instrument outside that framework. The ban keeps the stablecoin a payment instrument rather than a savings product, and preserves the boundary between payments and deposit-taking.

The consequences are worth naming plainly:

• Holding stablecoins costs you the risk-free rate. In a positive-rate world, a business holding balances forgoes real income.
• This reinforces Lesson 2's framing. Stablecoins are for moving money, not parking it. It is why tokenized treasuries and money market funds are a natural complement: settle in the stablecoin, hold value in the yield-bearing thing.
• It explains issuer incentives. Issuers are motivated to maximize float, meaning balances that sit still, while the ban ensures holders would rather not.

The design is coherent, and the tension is deliberate.

The Act closes real gaps. It does not make stablecoins risk-free, and a business should hold the residual risks clearly.

• Redemption availability, not just solvency. This is the USDC lesson from Lesson 1, and no reserve rule fixes it. Reserves sit in banks; banks keep hours and can fail. A token that settles 24/7 is backed by assets reachable only on weekdays. The Act mandates that reserves exist, not that they are reachable at 2am Sunday.
• Issuer and bank concentration. You now hold a claim on a specific issuer, whose reserves sit at specific banks. That is counterparty risk, and it is the risk that bit USDC. It is not diversified away by the token being on a chain.
• Run dynamics remain. Even fully-reserved instruments can be run on, because redeeming first is rational if you fear a queue. Full backing makes a run survivable; it does not make it impossible.
• Attestation is not audit. Monthly attestations, with CEO and CFO certification, are a genuine improvement over nothing. They are still periodic snapshots of a portfolio that changes daily, and an attestation is narrower than a full audit.
• The ends are still slow. Lesson 2's honest caveat: the on-chain leg is fast, but the on-ramp and off-ramp sit in the banking system, with its hours, compliance checks, and fees.
• AML and sanctions. Public chains are transparent, which helps investigators and also means your counterparties and balances are visible. Businesses inherit screening obligations, and the compliance cost is real and easy to omit from a cost comparison.

None of these is disqualifying. All of them belong in an honest evaluation.

Assemble the picture across the cursus.

The settled column is substantial, and it is mostly mechanism: a stablecoin is a fully-reserved claim, redeemable at par, whose peg is enforced by arbitrage and whose usefulness comes from settlement being the transfer itself. That is durable knowledge, and it will outlast any particular issuer.

The unsettled column is mostly institutional. The sharpest open question is the one Lesson 2 raised: whose liability wins? Banks are building tokenized deposits, keeping the value inside the banking system. Issuers are building stablecoins, holding reserves outside it. Both settle on-chain, both are fast, and they are competing for the same corporate balances. The GENIUS Act legitimized the second without settling the contest.

And note what is not in the settled column: how much real payment volume exists. Lesson 2 showed why the headline trillions do not answer it. The honest position is that the mechanism is proven, the regulation is now real, and the adoption is genuinely underway but far smaller than the numbers suggest.

Close with something usable. If a business is considering a stablecoin, the questions follow directly from the mechanism, and none of them are about blockchains.

• Who is the issuer, and under what regime? A GENIUS-compliant US issuer is a defined, supervised entity. An offshore issuer under no comparable regime is a different proposition wearing the same word.
• What is in the reserves, and where? Permitted assets are short-dated and government-backed for a reason. Ask which banks hold the deposits, because that is your concentration risk (Lesson 1's SVB lesson).
• What is the redemption policy? The Act requires it to be published. Read it. Redemption is the peg (Lesson 1), so its terms, timing, and conditions are the product.
• What do the monthly attestations actually say? They are required and certified by the CEO and CFO. Periodic and narrow, but far better than a claim on a website.
• What is the all-in cost? On-ramp, off-ramp, network fees, FX, and compliance. Compare against your actual corridor, not against a worst-case wire.
• Whose liability do you want? Issuer (stablecoin) or bank (tokenized deposit)?

The throughline of the whole cursus: a stablecoin is a claim on an institution that settles like software. The software half is genuinely new and genuinely useful, especially where correspondent banking is worst. The claim half is old finance, and it is where all the risk lives. Evaluate the second half with the tools you would use for any other counterparty, and the technology stops being the interesting question.

The Act mandates 1:1 permitted reserves, restricts who may issue, and requires published redemption policies with monthly certified attestations, while banning yield to holders so the issuer keeps the float; redemption availability, issuer concentration, run dynamics, and slow on and off ramps remain.