The New Interchange

Stablecoins will not kill the card networks. They are relocating the payment that keeps the networks alive.

Context. On 15 July 2026, Stripe and Advent International offered $60.50 per share in cash for PayPal - a $53.4bn valuation, a 28% premium to the undisturbed close of $47.37, funded by roughly $17bn of equity and about $50bn of committed bank financing from J.P. Morgan and Morgan Stanley.¹ PayPal’s board regards the offer as inadequate; no formal response has been made.² PayPal’s market value peaked near $360bn in 2021 and touched roughly $36bn earlier this year.³ Two weeks before the bid, on 30 June, a consortium of more than 140 companies - Visa, Mastercard, American Express, Stripe, BlackRock, Coinbase and Google among them - announced Open USD (OUSD), a jointly governed dollar stablecoin whose reserve yield flows to its distribution partners rather than to a single issuer.⁴ The two events are being reported separately. They are one story.

1. What a card network sells

Visa and Mastercard will carry roughly $13.5trn of payment volume this year.⁵ The durable assets behind that volume are three:

• A two-sided coincidence: near-universal acceptance and near-universal issuance, each worthless without the other.
• A steering payment: interchange and associated card fees - $198bn for US merchants in 2025, on Nilson data⁶ - collected from merchants and passed to issuing banks, which recycle a large share into the rewards that direct consumer spend onto the rails.
• A trust layer: chargebacks, fraud liability rules and consumer credit. Reversibility, on someone else’s money.

The rails themselves are the least defensible part. Cheaper pipes have existed for a decade and have not mattered, because pipes were never the product.

2. Interchange is a steering payment

A US merchant does not pay ~2% on a credit transaction for message routing; the network’s own fee is a fraction of that. Most of the take is interchange, paid to the card issuer. Part of it compensates the issuer for credit risk and fraud liability. The remainder funds rewards - and rewards exist to steer the consumer’s next transaction onto the same rail. Interchange is, functionally, a payment from merchants to whoever controls consumer choice at the moment of checkout.

This framing explains forty years of failed challengers. Closed-loop alternatives have pitched merchants lower fees while offering consumers nothing: MCX’s CurrentC - backed by Walmart, Target and Best Buy - shut in 2016 without a national launch.⁷ Cheaper rails lose to subsidised habit.

3. Yield-share is interchange reinvented

OUSD’s design: partners mint and redeem at zero cost, with no volume caps, and nearly all interest earned on the reserve assets is distributed to consortium members after a management fee.⁴ The steering payment already exists at scale in the single-issuer model: Circle’s 10-K records $1.4bn of distribution costs to Coinbase in 2025 - roughly 51% of its revenue and reserve income - for carrying USDC, under an agreement that pays Coinbase all reserve income on USDC held on its platform and half of the income generated elsewhere.⁸ OUSD does not invent the payment; it mutualises it.

The legal perimeter is contested but the direction is set. The GENIUS Act prohibits stablecoin issuers from paying yield to holders;⁹ it does not, on its face, prohibit a consortium from paying its distributors. The OCC’s proposed implementing rules would extend the prohibition to affiliates and related third parties that pass yield through to holders - while stating that independently offered merchant discounts for paying in stablecoin fall outside the ban.¹⁰ Value may flow to the parties who control choice; what is being regulated is whether it reaches the holder as yield or reaches the checkout as price.

That is interchange’s job description. Three things change; none of them changes the function.

The payer changes. Interchange is merchant-funded, levied per transaction, and visible on every merchant statement. The yield pool is funded by the balance-holder’s forgone interest - the market rate not earned on float - and appears on no statement. The visibility difference has regulatory history: interchange was capped (Durbin in the US, the Interchange Fee Regulation in the EU) because merchants could see it, price it and lobby against it. A holder-funded steering pool has no equivalent constituency. The payment survives and goes dark.

The tax base changes, from flow to stock. Interchange is levied per transaction; reserve yield accrues on balances held. A yield-funded network rewards whoever persuades users to hold balances - one reading of why a consumer wallet with 400m+ funded accounts¹¹ commands a $53bn bid, and why float, not per-transaction fees, becomes the prize.

Today the scale is an order of magnitude apart. On a stablecoin base of roughly $325bn,¹² the current 3.6% short rate funds a global pool in the region of $12bn a year - against $198bn of card fees for US merchants alone.¹³ The claim is about trajectory, not present size: at the $2trn base the US Treasury Secretary has suggested,¹³ the pool approaches $80bn at a 4% short rate - and shrinks towards nothing at 1%, while card interchange, levied on flow, is rate-indifferent. The challenger model carries a macro exposure the incumbent model does not.

What does not change is the architecture. Both mechanisms route B2B: interchange flows merchant to acquirer to network to issuer, and the issuer converts it into consumer-facing rewards; yield flows consortium to distributor, which must make the same conversion. Consumers never see interchange; they see points. The difference is industrial, not structural: card issuers hold four decades of conversion machinery - rewards engineering, co-brands, habit formation - and the wallets hold almost none. Closing that gap is what a wallet of that scale is for. One asymmetry to hold throughout: the Act leaves tokenised bank deposits outside the framework and free to pay interest,¹⁴ so the strongest competitor for the float sits beyond the perimeter entirely.

4. The two-track game

Stripe sits on both sides of the board. It is a founding OUSD partner alongside Visa and Mastercard, and the consortium entity is run by Zach Abrams, co-founder of Bridge, the stablecoin infrastructure firm Stripe bought for $1.1bn.^4,15 In parallel it has assembled the bypass: Bridge for on/off ramps; Tempo, its own settlement chain, mainnet live since March;¹⁶ Privy for wallet infrastructure; Link, its accelerated-checkout wallet, past 250m consumer accounts on company figures;¹⁷ and now a bid for PayPal’s 400m-account consumer wallet. Citi’s read of the combination: the first fully vertically integrated private digital-dollar stack - issuance and reserves, settlement rails, and merchant processing under one roof.¹⁸

The incumbents run the same two-track strategy from the other end. Visa settles network obligations in stablecoins across nine blockchains, Tempo among them.¹⁹ Mastercard paid up to $1.8bn for BVNK.²⁰ Both joined OUSD. Swift is expanding a blockchain settlement network with more than 40 institutions.²¹ No participant is defending the rails; every participant is bidding for position in the layer where the steering payment will be set.

5. Where the loop closes

Base rates first. Closed-loop schemes have displaced cards only where cards were never entrenched - Alipay and WeChat Pay in China, UPI in India, Pix in Brazil - markets with thin card penetration, two of the three with state sponsorship. In carded markets, merchant-led alternatives have failed serially.

The realistic contest is therefore not the US point of sale. It is the channels where the habit moat is thin or absent:

• Online checkout, where a wallet button competes on equal visual terms with a card form. The closed-loop case for the bid is explicit: Venmo and PayPal balances spent across Stripe’s merchant base, at lower cost to the merchant, without the card networks in the path.²²
• Cross-border and B2B, where the cost being attacked is correspondent banking, not interchange, and stablecoin settlement already competes on speed and price.
• Payouts and treasury, where recipients hold balances - the stock on which the new interchange accrues.
• The agentic channel, below.

6. The agentic channel

Agents do not collect points. The consumer-habit moat that interchange funds is a human phenomenon; agentic checkout replaces habit with optimisation. An agent authorised to complete a purchase evaluates routing on merchant economics, settlement speed and protocol compatibility, not on airline miles. Whoever writes the checkout protocol the agents speak sets the default rail - one reason PayPal’s agentic-commerce protocol positions are themselves cited as part of the acquisition rationale.²³

The standards contest has produced its first artefact. Tempo’s mainnet went live on 18 March alongside the Machine Payments Protocol, an open, rail-agnostic standard for agent payments co-authored with Stripe - and Visa extended MPP to card payments on its own network at launch, with Lightspark extending it to Bitcoin’s Lightning network.¹⁶ The incumbent adopting the challenger’s protocol on day one is the two-track game in a single announcement.

Three limits on the claim. Agentic commerce volume is currently small; the erosion is real but back-loaded. Compliance for autonomous agents - KYC, AML, liability for an agent’s purchase - is unresolved, and several protocols are likely to coexist before any wins. And agents can be paid to prefer a rail, just as brokers were paid for order flow.

The last of these is not an aberration; it is the system’s lawful outlet. GENIUS forbids paying yield to the holder - the human;⁹ the OCC would extend that ban to anyone passing yield through to holders;¹⁰ neither reaches a platform that keeps the economics, and an agent platform is a platform. As steering value seeks a legal destination on the consumer’s side of checkout, the party it can reach is not the consumer but the consumer’s agent. The rewards programme of the coming decade pays your agent, not you. The question from The Double Agent Economy - who does the agent work for - is no longer a design flaw to be audited. It is where the statute channels the money.

7. Who is exposed

• Third-party issuers first. Circle fell about 17% in the session after the OUSD announcement, closing near $62.63 - a decline compounded by a Russell index removal the same day - leaving the stock roughly 55% below its mid-May high.²⁴ Around 96% of Circle’s revenue is interest on reserves;²⁴ a consortium coin exists to stop paying that margin to an outsider, and the August renegotiation of the Coinbase agreement (section 3) is the near-term catalyst - a negotiation Coinbase can now enter pointing at OUSD-style terms as its comparator.⁸ Tether, the largest issuer, is the apparent counterexample - it dominates while paying distributors little - but USDT’s demand is offshore dollar access rather than steered payments; it competes in a different market.
• The loyalty-industrial complex. Rewards are funded by merchant card fees on flow that closed loops and agents remove - and entire industries are levered to that funding. Delta received $8.2bn from American Express in 2025, roughly 14% of adjusted operating revenue and about 1.4 times its adjusted operating income; it guides to $9bn this year and $10bn long term.³¹ United now requires a co-branded card to earn miles on basic-economy fares.³¹ An airline whose card cheque exceeds its operating income is, economically, a loyalty programme with wings - and agents do not collect miles. If steering value migrates from interchange-funded points to platform yield and merchant discounts, the miles economy loses its funding mechanism a decade before the airlines lose a passenger. For the consumer, the transition is quiet: checkout looks identical, points thin, discounts replace miles, and the system is funded by float they never see earning. The points will not be stolen; they will be starved.
• Pure-play processors and gateways, squeezed between vertically integrated stacks on both sides.
• The networks last. Visa and Mastercard are inside OUSD, hold the trust layer, and are integrating the new rails as settlement options. Two scenarios bound the outcome. Base case: accommodation - the networks defend habit-driven consumer volume, take their share of consortium yield, and the steering payment is repriced and redistributed rather than abolished. Erosion case: the contested channels above are where card volume growth sits, and interchange on flow that migrates is not shared but lost. Migration of even 10-30% of addressable online and cross-border volume over a decade - an illustrative assumption, not a forecast - would pressure network multiples while total payment volumes still grew; it presupposes the stablecoin base scales first (section 3). Which scenario dominates turns on the conversion-machinery gap and the agent-protocol contest.

A sovereignty note: Mastercard is exploring the sale of majority control of Vocalink - which processes over 90% of UK salaries and 98% of state benefits - back to British banks over US-ownership concerns,²⁵ and a US 25% Section 301 tariff on most Brazilian goods, citing Pix’s treatment of the card networks among the grounds, takes effect on 22 July.²⁶ Whatever the private stack does, national-rails politics is moving in the same direction: settlement as a sovereignty question. (Companion argument: Sovereignty Not Included.)

8. The merchant’s position

For the corporate paying the $198bn, the contest is leverage before it is technology. The near-term moves are procedural: route cross-border payables and marketplace payouts through stablecoin settlement, where the saving is correspondent-banking spread rather than interchange; take the consortium side’s acceptance incentives while both camps are bidding for flow; and note that the OCC’s proposed rules place independently offered merchant discounts outside the yield ban¹⁰ - the one lawful gate through which the steering payment can reach the consumer is the merchant’s own pricing. Merchants spent two decades litigating interchange. The next negotiation is over which side of checkout its replacement lands on.

Open questions

• Whether any transaction completes. The board calls $60.50 inadequate; management is mid-turnaround; William Blair analysts are publicly sceptical the new CEO would accept.²⁷ PayPal is working with Goldman Sachs and Evercore on alternatives including a potential sale or breakup, and its 28 July earnings are the near-term read on the turnaround the board is defending.²
• Block’s role. CNBC reported Block contributing to the $17bn equity alongside Stripe and Advent; subsequent Reuters reporting says Block joined the April approach and exited before the offer was submitted.²⁸ The difference is strategic, not procedural: with Block, the combination reaches the physical point of sale through Square and Cash App; without it, the contest is confined to the channels in section 5.³²
• Whether a 140-member consortium can govern a coin. Within days of the announcement, Samsung and Dunamu publicly denied formal roles despite appearing in launch materials - headline membership is not committed liquidity.²⁹ Sell-side defence of Circle reached for MCX and Paze as precedents for consortium failure;²⁴ ARK’s stated critique is cold-start liquidity, absent trading pairs and governance friction.³⁰ OUSD is announced, not live; launch is slated for later in 2026, on Solana first.⁴
• Whether yield-share survives a rate-cutting cycle (section 3).
• The regulatory perimeter. The antitrust outcome turns on market definition - drawn around merchant processing, the Stripe-Braintree overlap reads anti-competitive; drawn around the card duopoly, the combination reads as entry.³² A mooted remedy carves Braintree out to Advent, whose payments portfolio (Nuvei; previously Worldpay, Vantiv) gives divested assets a landing place;² the GENIUS framework takes effect no later than 18 January 2027, with the OCC’s affiliate-yield rule not yet final;^9,10 the treatment of a combined PYUSD/OUSD position under that framework is unaddressed in public reporting.

Elliot Ronald advises boards and investors on strategy in regulated markets through Lion Strategy. To discuss what this means for a specific balance sheet: contact@lionstrategy.com.

Part of the Intelligence Economy series, with The Double Agent Economy and Sovereignty Not Included.

To work through where the steering payment lands for your balance sheet, your loyalty economics or your portfolio, start a conversation.

Talk it through with Elliot ←︎ BACK TO THE SERIES