Tokenized A2A, stablecoin infrastructure, and correspondent banking are not competing alternatives - they occupy distinct layers of the institutional payment stack. This framework, developed by Post Oak Labs, is a fully consolidated reference for institutions evaluating where tokenization fits in their architecture and revenue model.
The most common institutional mistake is treating these three as substitutes. They are not. Stablecoin/CBDC infrastructure is the monetary primitive - what digital money is. Tokenized A2A is the institutional rail - how banks move it at scale. RTGS is the final settlement backbone beneath everything. The confusion costs institutions time, money, and the wrong vendor relationships.
3-5 intermediary hops per payment - 2-5 day settlement - high pre-funding requirements - limited data fields per transaction
Account-to-account tokenized settlement - 1-2 direct fees - near-instant finality - ISO 20022 - B2B cross-border - corporate treasury - interbank settlement - tokenized MMF distribution
Central bank digital currency - private stablecoin rails - reserve management - programmable money - enabling/base layer on which A2A can optionally operate
Pix (Brazil) - SPEI (Mexico) - FedNow (US) - UPI (India) - FPS (UK) - retail/consumer P2P; not the institutional A2A layer
Permissioned DLT, Public L1 blockchains, Central bank RTGS[12]
The monetary primitive - a tokenized representation of fiat value. Answers the question of what the settlement asset is: bearer instrument, programmable on-chain value, CBDC unit, or tokenized deposit. Does not, by itself, solve how regulated institutions move it at institutional scale.
The institutional payment rail - orchestration, routing, and settlement workflow above the asset layer. Answers the question of how regulated banks, central banks, and large corporates transfer value efficiently across borders. Can carry stablecoins, CBDCs, or tokenized deposits as the underlying asset.
The settlement backbone - central bank money on central bank books. Not replaced by A2A; abstracted and used more efficiently. A2A nets positions across participants intraday and end-of-day, reducing gross RTGS settlement volumes while preserving finality in central bank money.
A2A is a payment rail - the mechanism by which value moves between bank accounts. Stablecoin is a monetary instrument - what moves on the rail. An institution that conflates the two will architect for the wrong problem. Where the two models differ on a structural dimension, a contextual note explains whether the difference is a design trade-off or a fit-for-purpose distinction. Neither model has a universal advantage over the other; each is optimized for a specific position in the stack.[1]
| Dimension | Tokenized A2A | Stablecoin Infrastructure |
|---|---|---|
| Role in Stack | Orchestration, routing, and settlement workflow layer - application/network layer | Base asset and settlement primitive layer - asset/ledger layer |
| Position in Stack | Operates above stablecoin/CBDC as institutional B2B settlement rail | Enabling/base layer; programmable money that A2A rails can optionally carry |
| Nature of Instrument | Tokenized deposit - programmable claim on the sender bank's fiat reserves; fiat never leaves the bank balance sheet during the transaction | Privately issued digital bearer instrument representing fiat value; backed by reserves and operates entirely on-chain |
| Core Function | Moves funds between bank accounts (B2B, interbank, treasury) via tokenized claims | Represents tokenized fiat value as an on-chain liability; enables direct P2P transfer |
| Balance-Sheet Treatment | Stays on issuing bank's balance sheet; funds remain inside the banking system and support traditional lending | Off-balance-sheet for banks; reserves held with the stablecoin issuer, removing deposits from the banking system Trade-off[2] |
| Settlement Mechanism | Tokenized claim transfer, then net settlement via RTGS at end-of-cycle; fiat never moves between banks during the transaction itself | Direct on-chain transfer of tokenized value via blockchain consensus; settlement is the transfer itself |
| Issuer Identity | Regulated commercial bank or central bank - known, licensed entity | Nonbanks (Circle, Paxos), bank subsidiaries, or regulated banks (JPMorgan, BNY Mellon) |
| Reserve Backing | 1:1 fiat reserves held at the originating bank; transparent custody under banking supervision | Varies by issuer: cash, Treasuries, crypto, or algorithmic; reserve quality and transparency differ significantly by issuer Context[3] |
| Redemption Rights | Redeemable at par with the originating bank; clear legal claim under banking law | May trade at discount or premium in secondary markets; redemption terms vary by issuer; regulated issuers moving toward standardized at-par redemption obligations Context[3] |
| Network Type | Closed, permissioned ledgers among known, KYC'd institutional participants | Public or open blockchains, or private/permissioned ledgers - depends on the issuer and deployment |
| Access Model | Account-based; restricted to participating banks and corporates | Wallet-based; accessible to anyone with internet access - a deliberate design property for consumer and DeFi use cases Context[4] |
| Regulatory Status | Treated as a bank deposit; eligible for deposit insurance and banking supervision | Not deposit-insured; regulated under specialized frameworks (MiCA in EU, GENIUS Act in US); operates as a non-bank liability Context[3] |
| Pre-funding Requirement | Near-zero for A2A rail participants; just-in-time settlement via netting cycles; frees capital from correspondent pre-funding requirements | Full reserve backing required by design - a structural property providing collateral transparency and supporting at-par redemption, not simply an operational inefficiency Trade-off[2] |
| Data per Transaction | ISO 20022 native - substantially richer structured data than legacy MT messages; banking-native messaging standard[5] | Token metadata and smart contract logic; minimal structured data relative to ISO 20022 - sufficient for on-chain use cases |
| Compliance and Identity | Graduated KYC tiers; shared attestation across permissioned network; compliance native to the banking perimeter | Requires additional wallet screening, third-party compliance checks, and blockchain monitoring to verify on-chain entities |
| Integration with Banks | Designed to integrate with core banking systems via middleware APIs; no core replacement required | Operates largely outside traditional banking; requires wallet/blockchain integration and dedicated fiat on/off-ramps |
| Dependency | Can carry stablecoins, CBDCs, or tokenized deposits as the underlying asset - asset-agnostic by design | Does not require A2A; can be used standalone on public or private ledgers |
| Transaction Volume | Institutional; varies by corridor and deployment - not yet aggregated at an industry-wide level | On-chain stablecoin volume exceeded $8.9 trillion in H1 2025; first month exceeding $1 trillion was September 2025[6] |
| Primary Use Cases | Institutional B2B cross-border, corporate treasury, interbank settlement, tokenized MMF distribution | Cross-border remittances, crypto trading, DeFi, dollarization tool, consumer-facing payment rails |
Stablecoins solve "what is the digital dollar?" - A2A solves "how do banks actually move money using it at scale?" Stablecoins without A2A: liquidity exists, but institutional rails are fragmented. A2A without stablecoins: rails exist, but the settlement asset can still be legacy fiat. Tokenized A2A keeps fiat within the regulated banking system while leveraging tokenization for efficiency. The two models are most powerful in combination, not in competition.
Correspondent banking moves balances across banks. A2A moves claims instantly and settles balances later. RTGS is not replaced by A2A - it is abstracted and used more efficiently. That structural shift drives fewer intermediaries, less capital lock-up, faster settlement, and richer compliance data. Where the two differ, contextual notes explain whether the distinction reflects an architectural trade-off or a practical operational difference. Correspondent banking's key structural strength - global network reach - is called out explicitly.[1]
| Dimension | Tokenized A2A Rails | Correspondent Banking / RTGS |
|---|---|---|
| Topology | Direct "four-corner" model: sender bank - settlement ledger - receiver bank; no intermediary banks on the token transfer path itself | Fragmented multi-hop network; payment passes through 3-5 or more correspondent banks, each adding fees and processing time[7] |
| Intermediaries and Fees | 0-1 logical intermediaries; approximately 0.2% total in direct settlement fees (Post Oak Labs production benchmark in targeted B2B corridors; excludes FX spread captured as originating bank revenue)[8] | 3-5 intermediary hops per payment; layered fees at each (SWIFT plus correspondent plus FX spread plus nostro maintenance); total 2.5-5% plus FX spread[7] |
| Settlement Speed | Seconds to minutes; under 10 seconds demonstrated in multi-CBDC corridors (Project mBridge, PoC conditions)[9] | 2-5 business days for final settlement on legacy non-gpi routing; SWIFT gpi: ~60% of payments credited to end beneficiaries within 30 minutes, nearly 100% within 24 hours (SWIFT, 2024–2025). Delays beyond 24 hours occur primarily in corridors with capital controls, legacy back-office systems, or time-zone mismatches — not in standard gpi routing[9] |
| Liquidity Pre-funding | Near-zero for participants; just-in-time settlement via netting cycles; frees capital currently held in correspondent pre-funding[2] | Heavy reliance on pre-funded nostro/vostro accounts across all active corridors; industry estimates put total trapped globally at approximately $27 trillion[2] |
| Capital Efficiency | High; reduced trapped liquidity; capital freed for productive lending and investment | Low; significant capital lock-up in correspondent balances earning below-market returns while idle — at 5% rates, $1B in a non-interest-bearing nostro costs roughly $50M/year in lost yield (actual opportunity cost varies by treasury management practice; some balances earn partial overnight returns, but typically well below market) Context[2] |
| Operating Hours | Typically 24/7 on DLT/permissioned ledger, subject to network risk policy and bilateral agreements | RTGS: limited to operating hours and maintenance windows; correspondent layers add timezone, cut-off, and batch processing effects |
| FX Handling | On-ledger real-time FX at point of redemption; originating bank captures the spread rather than losing it to correspondents | Opaque spread markups at each correspondent hop; spread captured by intermediaries rather than the originating bank Trade-off[4] |
| Data per Transaction | ISO 20022 - substantially richer structured data than legacy MT; automated reconciliation; approximately 30% fewer exceptions vs. MT-based workflows[5] | Legacy MT messages carry far fewer data fields than ISO 20022; manual reconciliation required; high exception rates; zero real-time payment status visibility[5] |
| Reconciliation | Automated, ledger-native via ISO 20022 rich data; immutable audit trail from token minting | Manual, exception-heavy; post-hoc audit trail reconstruction from multiple systems at multiple institutions |
| Programmability | Full conditional logic: escrow, milestone payments, trade finance automation, atomic settlement | Static instructions only; no conditional logic or programmable execution in legacy MT messaging |
| KYC/AML Overhead | Shared attestation across network participants; graduated KYC tiers; compliance data travels with the payment | Repeated independently at each correspondent hop; each bank runs its own checks on the same counterparties Trade-off[10] |
| Audit Trail | Immutable from token minting; regulatory reporting embedded at the ledger layer from inception | Fragmented; reconstructed from multiple systems at multiple institutions after the fact |
| PCI Scope | Reduced; actual account data never transits the tokenized rail | Full account data transits through all intermediaries in the correspondent chain |
| Settlement Mechanism | Token transfer on-ledger; net settlement via RTGS or bilateral netting at end-of-cycle - RTGS is retained as the final settlement backbone | Payment-by-payment SWIFT instructions; gross RTGS settlement on central bank books for each individual transaction |
| Network Reach | Limited to participating institutions currently; A2A network effects are still building Trade-off[11] | SWIFT connects over 11,000 financial institutions across 200-plus countries; established global reach with deep liquidity pools - a genuine structural strength of the correspondent model[11] |
| Governance and Risk | Closed multilateral platform with shared rules; consistency enforced by protocol; participants are known and permissioned | Bilateral correspondent relationships and SLAs; credit and FX risk held in intermediaries; RTGS operates with a central bank/sovereign backstop |
A2A is not an RTGS replacement - it is an enhancement. RTGS remains the final settlement layer (central bank money); A2A is the transaction layer that uses RTGS for net settlement at end-of-cycle. RTGS is not eliminated - it is abstracted and used far more efficiently. That shift drives fewer intermediaries, less capital lock-up, faster perceived settlement, and richer compliance data. The primary near-term limitation of A2A is network reach: SWIFT's 11,000-plus member institution footprint represents a global coverage that tokenized A2A networks are still building toward, and correspondent banking remains the only viable option for corridors not yet served by A2A participants.[11]
The correct infrastructure choice depends on the scenario, counterparty type, market maturity, and regulatory environment. No single infrastructure is universally superior - each is fit for specific use cases, and the most robust institutional architectures combine all three in the correct dependency order. Correspondent banking retains a role wherever A2A network reach has not yet extended.
Stablecoin/CBDC = enabling layer (what is the digital money?) - A2A = institutional payment rail (how do banks move it at scale?) - RTGS = settlement backbone (how do positions net finally?). The most compelling institutional deployments combine all three in the correct order of dependency, not as competing bids for budget but as complementary layers of a single coherent architecture. Correspondent banking retains a genuine and important role for global reach in corridors where A2A network participants have not yet been established.
Post Oak Labs works with commercial banks, central banks, and large corporates that are ready to design and deploy tokenized A2A infrastructure - not just study it. We have built production systems in the Caribbean and South Asia. We know what works.