Four distinct blockchain-based architectures are now competing to rewire the $190 trillion annual cross-border payments market. SWIFT's permissioned shared ledger, built on Hyperledger Besu with 40+ bank participants, entered MVP build phase on March 30, 2026, targeting live tokenized deposit set...
"This is not really about disruption. It's about choice." — Tom Zschach, former Chief Innovation Officer, SWIFT (departed April 2, 2026)
Four distinct blockchain-based architectures are now competing to rewire the $190 trillion annual cross-border payments market. SWIFT's permissioned shared ledger, built on Hyperledger Besu with 40+ bank participants, entered MVP build phase on March 30, 2026, targeting live tokenized deposit settlements by mid-year. Stripe's Tempo mainnet launched March 18 with sub-second finality and fees under $0.001, backed by Paradigm and already onboarding Mastercard, Visa, Anthropic, and OpenAI. JPMorgan's Kinexys processes $1 billion daily in blockchain-based payments, growing 10x year-over-year. Meanwhile, Fireblocks and stablecoin-native rails processed $33 trillion in 2025 and are projected to clear $50 trillion in 2026.
The legacy correspondent banking system — where cross-border transfers still cost 2-5% in combined fees and take one to three business days — is being squeezed from all sides. The question is no longer whether blockchain enters mainstream payments infrastructure, but which architecture captures the settlement layer.
The current cross-border payments system relies on correspondent banking relationships where a payment may pass through two to five intermediary banks before reaching its destination. According to the Bank for International Settlements, this system has experienced sustained decline in many corridors over the last decade, with greater concentration contributing to rent-seeking behavior.
Key friction metrics:
The structural issue is not speed alone. It is that each intermediary in the correspondent chain extracts value — compliance costs, FX conversion margins, nostro/vostro account funding requirements — without proportional utility to the end user. This is the economic inefficiency that all four blockchain approaches target, though each attacks a different part of the stack.
SWIFT announced on March 30, 2026, that its blockchain-based shared ledger had completed the design phase and entered MVP build, with real-world transactions targeted before year-end.
Architecture: The ledger uses an EVM-compatible stack built on Hyperledger Besu, an open-source Ethereum client. It is permissioned — no native cryptocurrency, no public chain exposure. SWIFT operates the orchestration layer; banks retain full authority over keys, assets, and settlement through RTGS systems or correspondent relationships.
Participation: Over 40 financial institutions participated in design, up from the 30 announced at Sibos Frankfurt in September 2025. Participants include JPMorgan, HSBC, Deutsche Bank, Bank of America, BNP Paribas, Lloyds Bank, and Wells Fargo. More than 25 institutions have committed to go live by June 2026.
Mechanism: The ledger records, sequences, and validates interbank payment commitments via smart contracts. Banks issue tokenized deposits — digital representations of commercial bank money — and the ledger provides a shared, real-time record of obligations. Settlement still occurs through existing RTGS systems, but the orchestration and commitment layer moves on-chain.
Criticism: Fireblocks CEO Michael Shaulov dismissed SWIFT's approach publicly, noting the cooperative has been running blockchain experiments since 2018. "You can technically browse the internet using a fax machine — but I haven't seen anyone doing it," Shaulov wrote on LinkedIn, per Ledger Insights reporting.
The criticism has a structural basis: SWIFT's model adds a blockchain coordination layer but retains the existing settlement infrastructure. The value proposition is operational efficiency — reduced reconciliation, 24/7 availability, better liquidity visibility — rather than a fundamental redesign of the settlement stack.
Stripe and Paradigm's Tempo launched mainnet on March 18, 2026, taking a fundamentally different approach: build a purpose-built Layer 1 blockchain optimized exclusively for payments.
Architecture: Tempo is built on Paradigm's Reth SDK, the highest-performance EVM client available, using Simplex Consensus via Commonware to achieve approximately 0.5-second deterministic finality. The chain targets 100,000+ TPS with transaction fees under $0.001. Unlike general-purpose chains, Tempo reserves dedicated payment lanes for TIP-20 stablecoin transfers, isolating payment traffic from DeFi congestion.
Compliance integration: Tempo is ISO 20022 compliant — the international financial messaging standard used by banks and SWIFT itself — allowing enterprise integration with existing reconciliation systems. This is a deliberate bridge between crypto-native infrastructure and traditional financial plumbing.
Adoption: The public testnet, which began in December 2025, included Mastercard, UBS, Klarna, and Visa. Mainnet launch partners include Anthropic, OpenAI, DoorDash, Shopify, Nubank, Revolut, and Standard Chartered. The Machine Payments Protocol, co-developed with Stripe, enables autonomous software and AI agent payments without human approval at each step.
Economic model: Tempo targets the full value chain — not just interbank orchestration, but end-to-end settlement. By eliminating intermediary banks and settling in stablecoins natively, Tempo collapses the fee stack. A $10,000 cross-border B2B payment that costs $200-500 on correspondent rails would cost under $0.01 in network fees on Tempo, though FX conversion and compliance costs remain external.
JPMorgan's Kinexys (formerly Onyx/JPM Coin) represents a third model: a single bank building proprietary blockchain infrastructure and opening it to clients.
Volume: Kinexys processes an average of $1 billion per day in blockchain-based payments as of January 2026, according to American Banker. Payment transactions grew 10x year-over-year, and the platform has executed some of the largest repo transactions on any blockchain globally.
Architecture: Kinexys uses deposit tokens (JPM Coin) — tokenized representations of JPMorgan deposits — for real-time settlement. The system operates on a permissioned blockchain and supports multicurrency operations. In January 2026, JPMorgan announced plans to make JPM Coin interoperable across multiple chains to "unlock liquidity," per CoinDesk reporting.
Limitations: Kinexys processes a fraction of JPMorgan's overall $10 trillion daily payment volume. The system is inherently limited to JPMorgan's client base and counterparty network. Cross-chain interoperability — connecting Kinexys to other bank blockchains or public chains — remains in pilot phase, with a Chainlink-Ondo integration test for tokenized asset settlement announced in early 2026.
Strategic position: Kinexys demonstrates that single-bank blockchain payments work at billion-dollar daily scale. The challenge is network effects: a bank-specific ledger cannot become an industry standard without either opening the infrastructure or connecting to a shared layer — which is precisely what SWIFT's shared ledger attempts to provide.
Running parallel to all institutional efforts, stablecoin settlement on public blockchains has reached scale that dwarfs most private initiatives.
Volume: Stablecoins processed $33 trillion in settlement volume in 2025, according to multiple industry sources. Projections for 2026 exceed $50 trillion. B2B stablecoin payments surged from under $100 million monthly in early 2023 to over $6 billion monthly by mid-2025.
Institutional adoption: Visa launched USDC settlement in the United States, settling over the Solana blockchain through partners Cross River Bank and Lead Bank, reaching an annualized run rate of $3.5 billion by November 2025. Mastercard acquired BVNK for up to $1.8 billion in March 2026 to accelerate stablecoin payment infrastructure. Ripple's payment platform crossed $100 billion in cumulative processed volume as of March 2026.
Infrastructure layer: Fireblocks launched its Network for Payments in September 2025 with 40+ institutional participants including Stripe's Bridge unit and Circle, positioning as middleware between stablecoin issuers, exchanges, and traditional finance. This layer competes directly with SWIFT's shared ledger for the orchestration role but uses stablecoins rather than tokenized deposits as the settlement asset.
USDC supply dynamics: Circle's USDC took 64% of stablecoin transaction volume in Q1 2026, flipping Tether's USDT on this metric — a development already covered in separate webthreepedia reporting.
| Dimension | SWIFT Shared Ledger | Stripe Tempo | JPMorgan Kinexys | Stablecoin-Native Rails | |---|---|---|---|---| | Base layer | Hyperledger Besu (EVM) | Reth SDK / Simplex | Proprietary permissioned | Ethereum, Solana, others | | Settlement asset | Tokenized bank deposits | Stablecoins (USDC, others) | JPM Coin (deposit token) | USDC, USDT, others | | Permissioning | Permissioned, bank-only | Permissioned mainnet | Permissioned, JPM clients | Public, permissionless | | Finality | Not disclosed | ~0.5 seconds | Near-instant (private) | 0.4s (Solana) to 12s (Ethereum) | | Target TPS | Not disclosed | 100,000+ | Not disclosed | Varies by chain | | Tx cost | Not disclosed | <$0.001 | Internal to JPM | $0.001-$0.50 | | Participants | 40+ banks | Stripe, Mastercard, Visa, fintechs, AI cos | JPM clients | Open | | Live status | MVP build (live H2 2026) | Mainnet (March 18, 2026) | Live ($1B/day) | Live ($33T in 2025) | | ISO 20022 | Yes (native) | Yes | Yes | No (most) | | Regulatory model | Bank-regulated | Licensed MSB/EMI | Bank-regulated | Varies by jurisdiction |
Each architecture redistributes economic value differently across the payment chain:
SWIFT model: Preserves existing bank economics. Tokenized deposits still require correspondent relationships for final settlement. SWIFT captures orchestration fees. Banks retain FX margins and compliance revenue. The value proposition is cost reduction through operational efficiency (less reconciliation, fewer failed payments), not fee compression.
Tempo model: Compresses the fee stack most aggressively. Network fees below $0.001 eliminate intermediary extraction. Value accrues to Stripe (platform fees), stablecoin issuers (float income), and end users (lower costs). Banks lose intermediary revenue but may gain as Tempo participants.
Kinexys model: Internalizes value within JPMorgan's ecosystem. Clients benefit from faster settlement; JPMorgan retains all fees. Scalability is limited by single-institution network effects.
Stablecoin-native model: Distributes value most broadly. Stablecoin issuers earn on reserves ($5.7 billion in revenue for Circle and Tether combined in 2025). Blockchain validators earn gas fees. Users capture the largest share of savings. Compliance costs are externalized to participants.
The cross-border payments market is undergoing a structural split. Incumbent banks are consolidating around SWIFT's shared ledger and internal blockchain initiatives like Kinexys. Fintechs and technology companies are building on Tempo and stablecoin-native infrastructure. The two camps share EVM compatibility and ISO 20022 compliance as common technical standards, but diverge fundamentally on who controls the settlement layer and where economic value accrues.
The data suggests coexistence rather than winner-take-all. SWIFT's 11,500-institution network and regulatory embeddedness make displacement unlikely in the near term. But stablecoin rails processing $33 trillion annually — with projections above $50 trillion for 2026 — demonstrate that the permissionless alternative has already achieved scale that no permissioned system matches. Tempo's combination of institutional partnerships and crypto-native performance represents a deliberate attempt to bridge both worlds.
The economic value question is clear: correspondent banking's multi-percent fee extraction is unsustainable when sub-cent settlement exists. The transition timeline is the only variable. SWIFT's ledger preserves bank economics longer; stablecoin rails compress them faster. The market will price the difference.