Metadata mismatch found. The Bank Policy Institute and The Clearing House have now walked into the OCC's stablecoin rulemaking twice โ once responding to the March 2026 proposal, again in July โ asking the regulator to bolt bank-grade fraud monitoring and liquidity backstops onto payment stablecoin issuers. Read the comment letters as compliance literature and they are genuinely good: specific, operational, technically literate. Read them beside TCH's operating portfolio and the document changes character. The institution asking the regulator to raise the wall around stablecoin payment rails also operates CHIPS and RTP. It is lobbying on the competitive value of the exact infrastructure those stablecoins would displace. That is not a footnote. That is the trade.
The OCC published its final rule on payment stablecoin issuance with an effective date of January 18, 2027 โ eighteen months after GENIUS Act signature, consistent with the statutory timeline. GENIUS already sets a floor: 1:1 reserve backing, disclosure obligations, AML/CFT program requirements, and a prohibition on paying yield to holders. BPI and TCH are not arguing against that floor. They are arguing for what sits above it: liquidity buffers sized like a bank's, fraud-risk systems sized like a bank's, reserve custody routed through supervised depositories. The reporting on this framed it as "regulatory capture." I'd frame it more narrowly and more usefully: it is a fixed-cost transfer, and fixed costs are the most reliable moat in finance.
One framing correction before we go further. Howey is the wrong lens and applying it here is a category error. Payment stablecoins under GENIUS are carved out as non-securities โ the fight is bank law and payment-system law, not investment-contract law. The relevant comparison is MiCA, which sets reserve and disclosure requirements but explicitly permits non-bank issuance. That is the delta. The US is roughly two years behind MiCA's full application, and the BPI/TCH ask is the mechanism by which Washington lands stricter than Brussels did โ not through statute, through sub-statutory rulemaking where the comment record is thin and the litigation risk is deferred.
Here is the mechanism, and it is boring in the way that effective things usually are. Compliance is a fixed cost. It does not scale with issuance volume the way variable cost does. A $50 billion issuer and a $500 million issuer file the same BSA/AML program, retain the same auditors, stand up the same transaction monitoring stack, and now โ under the BPI/TCH ask โ fund the same liquidity buffer and fraud-risk desk. Divide that by reserves and the small issuer's unit economics collapse while the large issuer's barely move. Nobody has to ban anyone. The rule only has to be expensive enough.
The 2020 AMM paper I keep coming back to makes the same point in a different domain. I spent that DeFi Summer arguing that Uniswap V2's constant product formula looked neutral and wasn't โ the impermanent loss sat in a place retail didn't look. Compliance complexity works identically. It looks neutral. It is designed. Two of the BPI's asks do the heavy lifting: fraud risk monitoring and liquidity assurance. Those are precisely the two items requiring balance-sheet capacity and a risk middle office โ the two things a lean crypto-native issuer structurally cannot assemble in a quarter. Everything else in the letter is reasonable. That's the tell. Effective capture wears the highest-quality technical clothing available.
Run the revenue model on the issuer side. A payment stablecoin is roughly a 100% reserve book of cash and short Treasuries, and the issuer keeps the float โ that is the entire business. I want to be precise here, because the lazy take is wrong: stablecoin issuance is not a Ponzi structure. The yield is genuine risk-free carry. That makes it better than most of what I audited in 2020, and it makes it fragile in a different way. Reserve income is a levered bet on the policy rate. Cut rates and the spread compresses โ call it 30 to 50 percent of gross revenue in a fast easing cycle, elastic on duration. Now stack a bank-sized fixed compliance cost on top of a shrinking spread. The small issuer doesn't get regulated out. It gets arithmetically squeezed out.
And notice what GENIUS's yield prohibition actually protects. Not consumers. Deposit franchise value. An interest-bearing stablecoin is a shadow deposit, and shadow deposits reprice the funding base of every commercial bank in the BPI's membership. That is the economic motive underneath the consumer-protection rhetoric, and the letters never say it out loud.
I spent a week in early 2024 parsing thousands of pages of SEC filings to isolate a 0.03% redemption-fee disparity between two spot Bitcoin ETF structures. The lesson carried over intact: the exploitable edge is never in the headline, it is in the operational clause nobody reads. Here the operational clause is reserve custody and settlement routing. If stablecoin reserves must sit at insured depositories or clear through CHIPS and Fedwire, the bank does not merely earn a custody fee โ it acquires structural dependency on stablecoin float, and the float becomes a bank funding source. That is a far larger prize than the payment-fee fight everyone is watching.
Pattern emerging from chaos: this is a structural reallocation, not a total-demand shock. Stablecoin demand is driven by payment efficiency and dollarization abroad; a US entry rule does not delete that demand, it relocates the issuer. Offshore large issuers absorb limited direct impact and lose US-compliant distribution. Licensed US issuers โ the ones already holding trust charters and audited reserves โ come out neutral to positive, because a higher wall behind them is a gift. Bank-issued deposit tokens come out clearly ahead. New entrants get nothing. If the final rule lands meaningfully above the GENIUS floor, expect a two-tier structure: incorporation migrates, US access is served indirectly through licensed partners.
The angle nobody is pricing: TCH is not a lobbying shop, it is a clearing utility. BPI advocates a policy worldview; TCH defends a revenue line. Treating them as one homogeneous "bank lobby" flattens the most interesting split in the filing record. On this issue their interests align. On the treatment of bank-issued stablecoins they may not, and BPI's membership is not internally unified either โ deposit-heavy commercial banks and trading-heavy institutions want opposite things from stablecoins, and the public comment letter hides that tension behind unanimity.
Second blind spot: capture is not permanent, and the escape hatch is procedural, not moral. If the final rule imposes requirements beyond GENIUS's statutory delegation, affected parties have an Administrative Procedure Act claim โ arbitrary-and-capricious, or major-questions overreach. One citation in the coverage, 12 CFR Part 15, I could not independently verify against the current CFR, and I flag that because a wrong section number in a rulemaking story is usually a signal the detail was reconstructed rather than read. The comment period is closing, and that was the only cheap intervention window. After that it is litigation and Congressional Review Act arithmetic. Fork in the road ahead.
Watch three things between now and January 2027. The final rule text against the GENIUS floor โ measure the gap in basis points of compliance spend per dollar of reserves, not in adjectives. The reserve custody language โ if it routes through insured depositories or CHIPS/Fedwire settlement, the banks do not just collect custody fees, they acquire a balance-sheet dependency on stablecoin supply. And the rate path โ because the moat only bites when the spread narrows. Liquidity evaporation detected, and it is not happening on-chain.