Cross-border settlement rails — the dollar-bypass stack vs the dollar-extension rail

Built 2026-06-14. Sources: BIS Innovation Hub/HKMA/Reuters (mBridge), CSIS/Statrys/FXC Intelligence/Wikipedia (CIPS), Modern Diplomacy/watcher.guru (BRICS), the BIS 2025 Triennial FX Survey. Full URLs at the bottom.

Two competing futures for cross-border money. A dollar-BYPASS stack — China's CIPS, the multi-CBDC Project mBridge (live 2025, bypassing SWIFT), the e-CNY, and BRICS Pay — built partly to insulate trade from US sanctions. And a dollar-EXTENSION rail — USD stablecoins + SWIFT — that deepens dollar usage and the Treasury bid. Evidence-graded, with an explicit reality-check: de-dollarization is real at the margins, but the dollar's transactional dominance is intact. Macro counterpart to spec-asia-crypto-payments.

SWIFT — the incumbent, and the reason challengers exist

SWIFT connects 11,500+ institutions and remains the dominant messaging standard. Cutting Russian banks off SWIFT (2022) demonstrated the system's coercive power — the explicit motive for adversaries (and some partners) to build alternatives. Notably, SWIFT signed an MOU to work alongside CIPS — co-opting rather than purely competing.

CIPS — China's fast-growing alternative (big flow, small reach)

China's Cross-Border Interbank Payment System processed ~RMB 175T (~$24.5T) in 2024 (~38% CAGR since 2016); by end-2025, 193 direct + 1,573 indirect participants across 124 countries. On 16 Apr 2025 it briefly surpassed SWIFT's single-day throughput (~$1.76T) — symbolic, though SWIFT's institution count still dwarfs it. CIPS lets China and partners (notably Russia) settle outside dollar/SWIFT exposure — the practical core of transaction-layer de-dollarization. Fact.

Project mBridge — the live multi-CBDC bypass (now Gulf-and-China-run)

• Members: China (PBoC), Hong Kong (HKMA), Thailand, UAE (CBUAE); Saudi Arabia (SAMA) joined Jun 2024 as a full participant; observers across SE Asia, the Middle East, and Africa.
• Live & bypassing SWIFT: mBridge went live in 2025 as the first commercial cross-border CBDC settlement system, near-instant, bypassing SWIFT; it has processed ~RMB 387B (~$55B), ~95% in digital yuan.
• BIS step-back: the BIS (which incubated mBridge at its HK hub) reportedly stepped back (late 2024), leaving China and Gulf states to steward a dollar-bypass rail — a notable governance shift.
• Gulf cross-link: UAE + Saudi membership ties mBridge to the project's fin-gulf-sovereign-ai-capital thread — the same states deploying AI capital are building the bypass settlement layer.

BRICS de-dollarization — real in trade, aspirational in currency

BRICS Pay was demoed late 2024 (early pilot). Russia reports ~90% of intra-bloc trade now settles in national currencies (yuan-heavy) — genuine, sanctions-forced de-dollarization at the trade layer. A "BRICS currency," however, remains rhetoric; central-bank gold accumulation is the more concrete reserve-diversification move. Fact / contested.

Reality check — don't overclaim either way

• Transactional dominance intact: the BIS 2025 Triennial Survey found the USD on 89.2% of all FX transactions (Apr 2025), up from 88.4% in 2022 — no displacement.
• Reserve share eroding: the dollar's reserve share fell from ~73% (2001) to ~54% (2025) — slow diversification, not collapse.
• Where it's real: sanctioned bilateral trade, reserve diversification, oil-pricing experiments — margins, not the core. Capital controls, RMB non-convertibility, trust, and network effects keep the dollar dominant.

Forced vs chosen flows (added 2026-06-16, #60)

The strategic key the two-rails frame needs: which bypass flows are sanctions-FORCED (committed, irreversible) vs strategically-CHOSEN (optionality/hedge, reversible). Flows are fact; the classification is labeled analysis.

• The asymmetry: the bypass stack's durable demand is the forced flows (small but committed); its growth potential is the chosen flows (large but contingent — they persist only while building optionality stays cheap). That's why the dollar at 89.2% of FX coexists with a real bypass build-out: most of it is reversible hedging, not committed exit.
• The feedback risk: every sanctions action that forces a flow also incentivizes chosen hedging by everyone watching — sanctions overuse converts reversible hedges into permanent exits (chosen → forced). The US dollar-defense via USD stablecoins targets exactly the chosen column (keep the dollar convenient so optionality-seekers don't migrate). The contest is won or lost on the chosen flows — and lost fastest by sanctioning so broadly that "chosen" becomes "forced."

The two-rails contest (the strategic frame)

• Dollar-extension (private): USD stablecoins (USDC/RLUSD/USDT) push dollar usage and Treasury demand (macro-stablecoin-treasury-rail); the GENIUS Act institutionalizes it — a market-led defense of dollar primacy.
• Dollar-bypass (state): CIPS + mBridge + e-CNY + BRICS Pay settle outside the dollar/SWIFT — sovereignty and sanctions-resilience, but carrying the same state-control/programmability concerns the project flags for CBDCs.
• Why it matters: if USD stablecoins win, the fiscal-trap Treasury bid (macro-us-fiscal-trap) gains a new buyer base; if the bypass stack scales, the US loses a coercive lever and a funding channel. Both are dual-use control rails — for different states.

What is NOT asserted

• The dollar is not being displaced — transactional dominance is intact (89.2% of FX); de-dollarization is real only at the margins.
• mBridge/CIPS have not reached SWIFT-scale reach.
• Sanctions-resilience is a documented motive, not an endorsement; bypass rails carry the same state-control concerns as Western CBDCs.
• A "BRICS currency" is aspirational, not an existing instrument.
• Overlay edges are excluded from the SCC / Z3 / TLA+ proofs.

Sources: BIS — Project mBridge, Modern Diplomacy — mBridge from BRICS experiment to global dialogue, CSIS — Sanctions, SWIFT, and CIPS, FXC Intelligence — CIPS growth, Wikipedia — Cross-Border Interbank Payment System, Chicago Policy Review — BRICS & the dollar.

← Research index · structured data: spec-cross-border-settlement-rails.json · spec-cross-border-settlement-rails.md