The stablecoin settlement stack.

The dominant 2025-26 story — stablecoin volume ($33T) now exceeds Visa + Mastercard combined ($25.5T); the duopoly is in trouble — is everywhere. It fails on two counts: a methodology problem and a target problem. Both matter for reading the rest of this landscape.

Wrong on the volume

The $33T is methodologically inflated — it counts wrapped tokens, intra-protocol routing, and bot traffic as economic activity. Visa's own Onchain Analytics filter cuts it to ~$10.2T LTM. McKinsey's stricter supply-side filter cuts it further to ~$390B annualized. Roughly 1.5% of card volume.

Wrong on the target

Even at $390B, the disruption vector points at SWIFT, correspondent banking, and Western Union — not card networks. The real victims are JPM/Citi FX desks, MoneyGram, and the global wholesale messaging layer.

Volume composition — Artemis, mid-2025 annualized

The corridors that matter

• US ↔ Mexico — Bitso Business $82B 2025 TPV. ~10% of corridor through stablecoins.
• Argentina + Venezuela — stablecoin adoption >40% of the adult population. LatAm volume $324B 2025, +89% YoY.
• Nigeria — $92.1B received on-chain (6th globally).
• Turkey — heavy USDT-on-Tron.
• SE Asia — 60% ($245B) of genuine payment volume.

Pattern. USDT/Tron owns EM retail + grey-zone B2B. USDC owns US-compliant B2B + corporate treasury. The split is regulatory, not technical.

The corridor the rent map skips — Africa

Sub-Saharan Africa received $205B on-chain in the year to Jun 2025 (+52% YoY), and the rails majors spent 2025-26 wiring it: Visa × Yellow Card USDC settlement across ~20 markets (Jun 2025); the Circle Payments Network signing Flutterwave, Yellow Card, and Onafriq — 400M+ mobile wallets (2025); Mastercard × Yellow Card (May 2026) targeting remittances, B2B settlement, and treasury across Ghana, Kenya, Nigeria, South Africa, and the UAE. Regulation moved in step: Kenya’s VASP Act puts stablecoin issuance under the central bank (Nov 2025), Ghana passed its VASP law, Nigeria’s Investments and Securities Act classified digital assets as securities — and Nigeria and South Africa both exited the FATF grey list (Oct 2025).

The structural difference from LatAm: the last mile is mobile money, not cards. M-Pesa, MoMo, OPay, and PalmPay own cash-in/cash-out — so the African stablecoin play settles behind mobile-money front-ends rather than replacing them.

Where the saved interchange goes

A merchant pays 1.5% on Stripe stablecoin checkout vs. 2.9% on cards. The "saved" 1.4% does not reach the merchant. Captured by Stripe (1.5% flat), Coinbase Commerce (~1%), Circle (reserve yield, 50%+ rev-shared to Coinbase), BVNK/Bridge/Conduit (0.1-0.5%), and reserve custodians (Lead Bank, BNY Mellon, Anchorage, Cantor).

Takeaway — fees migrate from interchange to the orchestration, ramp, and custody stack. The rents shift; they don't disappear.

Stop reading "stablecoins" as one industry. Five rent layers, each with distinct economics and incumbents.

The Coinbase observation: 100% of USDC yield on Coinbase-held balances, 50% of all other residual yield. By 2029, Coinbase's USDC revenue ($6B/yr) exceeds Circle's ($3B/yr).

The issuer buys the Distribute layer

Tether ran the same arithmetic in reverse: $200M into Whop (Feb 25 2026, $1.6B valuation) — a creator marketplace where 18.4M users across 144 countries earn ~$3B a year — embedding self-custodial wallets via its Wallet Development Kit, payouts in USDT and USAT. The mechanism outweighs the deal: in Nigeria USDT is ~7% of fiat crypto purchases — buying stablecoins is the wrong onramp; getting paid in them is the right one. A platform that defaults to USDT payouts conscripts its earners into holders, no purchase decision required.

Circle is the issuer. Coinbase is the equity.

The settlement-layer threat — already absorbed

• Visa USDC settlement — Dec 16 2025 via Cross River + Lead Bank, USDC on Solana. $3.5B annualized.
• Mastercard MTN + BVNK — $1.8B BVNK acquisition (Mar 17 2026), largest stablecoin deal on record.

Card networks are the disruptors of themselves.

The merchant-acceptance threat — contained

Consumer-paying-merchant-in-stablecoin is <0.02% of global consumer payments outside EM and crypto-natives. Five years of attempts; no breakthrough.

Verdict. V/MA do not get circumvented in this cycle. They absorb the new rails as a settlement modality. The shorts on V/MA are reading the wrong target.

Live market share — May 2026

USDT + USDC = 83%. Despite GENIUS, MiCA, and 10+ new launches — duopoly intact.

Tether's two-tier hedge

Q1 2026: $191.7B assets vs. $183.5B liabilities. USAT launched Jan 27 2026 via Anchorage, Cantor custody. The thesis: serve compliant US institutional through USAT; keep offshore USDT for everyone else.

The three real threats

• USDG / Global Dollar Network. Paxos consortium — Robinhood, Kraken, Galaxy, Anchorage, Bullish, Mastercard, Worldpay. "Virtually all rewards returned to participants" attacks USDC's structural rev-share weakness.
• Bank deposit tokens. JPM Kinexys live; SocGen USDCV active. Watch any BofA/Citi joint launch — that's the sleeper risk to all non-bank issuers.
• The Trump factor. USD1: $0 → $4.4B in 14 months, ~60% Trump-family; Cantor: ~$600M / 5% Tether stake. Enforcement is asymmetrically suppressed.

Takeaway — duopoly intact through 2026. The threat vector to watch is USDG (consortium yield-share), not Tether's hedge or political tailwinds.

Stripe's four acquisitions in fifteen months

                Open Issuance · Sep 2025
                ├── USDB, CASH, mUSD, USDSL, USDH
                │
    Bridge ($1.1B · Feb 2025)  ←→  Privy (Jun 2025)
    ─ Stablecoin orchestration  ─ 75M+ embedded wallets
    ─ $20B+ TPV (4× YoY)        ─ Account abstraction
                │
                Stripe Core
                $1.9T TPV · $6.9B revenue
                │
                Tempo L1 ($500M / $5B)
                Mainnet live Mar 18, 2026
                ─ Stripe + Paradigm · 0.6s finality

The cannibalization math: at $20B Bridge TPV vs. $1.9T core, 1.5% stablecoin take is higher than 30-50 bps Stripe nets on cards. Per-dollar stablecoins are accretive. The bet: stablecoins become the default for global B2B settlement + emerging-market USD access — a $200T+ flow where cards were never present.

The crypto-native mirror — MoonPay

Stripe is not alone in assembling the stack; MoonPay built the same shape from the crypto side, and with four acquisitions of its own — Helio (~$175M, Jan 2025), Iron ($100M+, Mar 2025), Meso (Sep 2025) and DFlow (~$100M, May 2026). Its Headless on-ramp (May 14 2026) renders Apple Pay inline through embeddable frames, with no MoonPay brand in the flow — the consumer-facing mirror of Bridge, live in 100+ countries. And its April 2026 pact with Woori Bank to distribute a won-backed stablecoin takes the model sovereign: a national currency riding third-party global rails.

The native operators

Standout — Slash: $100M Series C at $1.4B (April 2026), $3B+ annualized stablecoin volume in 12 months via Bridge.

Honest read. Most "stablecoin neobanks" are Mercury-clones with stablecoin sleeves. Real differentiators: yield pass-through, payout-corridor density, card-network principal membership. The carve-out is the EM digital-dollar account, where the moat is demand-side — a weak local currency — not product-side: ARQ and Kast monetize populations that want to hold dollars, not move them.

Where the volume actually lives — EM rails

• Bitso — LATAM heavyweight. $82B TPV 2025. MXNB stablecoin.
• ARQ (ex-DolarApp, ex-Revolut founders) — $70M from Sequoia + Founders Fund (Mar 3 2026). 2M+ users, $10B+ annualized.
• Kast — $80M Series A (May 2026). 1M+ users, ~$5B annualized, 170+ countries.
• Lemon Cash — 5M+ accounts AR+PE.
• Fasset — $51M led by SBI (May 2026). 2M customers, $32B annualized.

The enterprise pattern — abstraction, not adoption

The 2026 corporate model is provider-abstracted: Worldpay × BVNK pays out in 180+ markets without the client holding a token; Circle’s Managed Payments sells the same promise; Thunes moved funding windows from T+2 to T+0 on USDC and cut its weekend float. The enterprise never touches the stablecoin — it buys the settlement property.

Disconfirmation. The corridor math cuts both ways. FSB 2025 monitoring puts B2B cross-border fees at 0.2% against 1.4% FX cost — FX is 87% of the total, and that prize goes to whoever executes FX best, not fastest. Wise runs a 0.53% take rate with 65% instant transfers on plain fiat rails. Jack Zhang’s critique — off-ramping to local fiat can cost more than interbank FX — still holds in G10 corridors. And the IMF reads ~80% of stablecoin transactions as bot-driven arbitrage. Stablecoins win the corridors banking abandoned; they do not yet beat the corridors banking serves well.

Disconfirmation. Tempo launched with four validators, all team-operated. If it stays a permissioned database with EVM compatibility, the "stablecoin-as-economic-internet" framing collapses into "Stripe built a faster ACH."

Every chapter so far taxes the flow of dollars — issuance, ramp, orchestration, custody. But the rail's cheapness does something the rent map misses: it spawns a consumer-app layer that captures end-user spend directly, on settlement that costs a fraction of a cent. Strip out interchange, strip out the licence, and the economics that result are unlike anything in fintech.

The clearest tell is profit-per-head. Tether sets the benchmark at the issuance layer — and the apps above it run the same pattern on even smaller crews. Visa clears its volume through ~30,000 people; this stack does it through dozens.

Hyperliquid is the limit case. Eleven people, no outside capital, a self-built L1 settling in USDC — it turned down a $1B investment at a ~$10B valuation because it had no use for the money. Roughly 99% of its revenue is auto-converted to its token and burned; there is no audited profit line because the team takes no distribution. It moves more value through eleven people than entire rent layers move through thousands.

The prediction markets are the layer going mainstream. Polymarket took up to $2B from ICE — the NYSE's parent — at an ~$8B valuation (Oct 2025), and its founder became the youngest self-made billionaire on the Bloomberg index. Kalshi raised $1B at a $22B valuation led by Coatue (May 2026) on ~$178B of annualized volume and 90%+ of US prediction-market activity. Both settle in USDC. The rail is the plumbing; the wager is the product.

Honest read. This layer is overwhelmingly grey or speculative — gambling, memecoins, perpetual futures, event wagering. The legitimate consumer-payment use case is still <0.02% of spend, as the absorption chapter holds. But that is the point, not the caveat: the app layer pays because it is grey — no licence, no interchange, no acquirer in the loop — and the near-free rail is what makes the arbitrage bankable.

Takeaway — the rent layers tax the dollars that move; the app layer captures the dollars that get spent. On a rail with no interchange and no licence, spend-capture is the richest seat in the stack — and the limit case is eleven people clearing ~$800M a year.

Agent-to-agent payments

• Coinbase x402 — ~$600M annualized (Mar 2026).
• Stripe Agentic Commerce + MPP.
• Catena Labs (Sean Neville) — $18M a16z.
• Skyfire KYAPay, Lightspark Grid + UMA.

Cross-chain — the UX collapse

• CCTP V2 (Circle) — 13+ chains. $2.4B moved Mar 2026 alone.
• USDT0 (Tether / LayerZero) — $70B+ since Jan 2025.

Inter-issuer par — the clearinghouse layer

Cross-chain moves one issuer’s token between chains; it does nothing for moving between issuers. USDC, USDG, USDB and the rest don’t redeem at par against each other — "different stablecoins have varying values across platforms," the 19th-century free-banking problem re-run on-chain, when private banknotes traded at a discount to distance and issuer. The emerging fix is a clearinghouse that nets and settles across issuers at par.

• The Better Money Company (TBMC) — $10M led by a16z crypto (Mar 2026). Building a stablecoin clearinghouse; launch partners span issuers (Paxos, Bridge, M0, Agora, Frax, Brale, MetaMask, Phantom) and builders (Privy, Modern Treasury, LayerZero, BitGo, Mesh). Investor Sean Neville also founded Catena Labs (above) — the agent-payment and par threads share a cap table.

Takeaway — agent payments are the first crypto use case beating cards on merit. The 0% Base USDC onramp vs. 4.5% card onramp is the asymmetry that converts.

1. V/MA absorbed-not-displaced through 2027. Falsifier: either name posts stablecoin-attributable revenue decline >5% in any 4Q window.
2. Coinbase × USDC rev-share central S-1 risk factor. Falsifier: Circle 10-K or new entrant doesn't flag as material risk by 2027.
3. USDG passes $5B mcap by EOY 2026. Falsifier: <$5B Dec 31 2026.
4. USAT crosses $10B by mid-2027. Falsifier: <$10B Jun 30 2027.
5. Mastercard × BVNK closes Q3-Q4 2026. Falsifier: deal blocked or terminated.
6. Major US bank announces deposit token in 2026. Falsifier: none by Dec 31 2026.
7. Agent-payment TPV crosses $5B annualized by EOY 2026. Falsifier: <$5B per public disclosures.
8. Tempo remains 4-validator at end-2026. Falsifier: more than 4 independent validators by Dec 31 2026.
9. A consumer app on stablecoin rails clears $1B annualized revenue with <100 staff through 2026. Falsifier: no Hyperliquid / Pump.fun / prediction-market operator sustains it on public or credibly-reported figures.
10. Kalshi or Polymarket crosses $300B annualized volume by EOY 2026. Falsifier: both <$300B per public disclosures at Dec 31 2026.
11. Whop stablecoin payouts ship and clear $250M+ annualized by EOY 2027. Falsifier: not live, or <$250M per public or credibly-reported figures.
12. An EM digital-dollar neobank (ARQ / Kast class) crosses 5M users by EOY 2027. Falsifier: none does per public figures.
13. A multi-issuer stablecoin clearinghouse settles cross-issuer at par in production by EOY 2027. Falsifier: no TBMC-class clearinghouse live with ≥3 issuers redeeming at par per public or credibly-reported figures at Dec 31 2027.