gettyimages.com
Every year, companies moving roughly $75 billion in goods between South Korea and Japan must route those payments through a detour that has nothing to do with either country: convert won to US dollars, thread the funds through a chain of correspondent banks, then convert those dollars back to yen at the other end. A new joint venture announced today is the most technically specific attempt yet to make that detour permanent history.
SBI Digital Practice (SBIDP), the wholly owned Canton Network subsidiary of Japan’s SBI Group, and NodeInfra, a South Korean blockchain software firm with established Canton validator credentials, signed a memorandum of understanding on August 7, 2026, to co-develop the Musubi Project — a stablecoin-based cross-border payment and settlement network linking Japan and Korea. The name “musubi” (結び) is the Japanese word for connection or binding. The infrastructure runs on the Canton Network, the privacy-preserving enterprise blockchain built by Digital Asset, and its smart contracts are written in Daml, the language Canton uses to encode financial rights and obligations at the protocol level.
The significance of the project is not that two firms signed a memorandum of understanding, which is a starting point, not a finish line. The significance is the specific architecture they chose, and what that architecture does to a fifty-year-old problem in international finance.
What Correspondent Banking Actually Costs
To understand why the Musubi Project’s mechanism matters, it helps to be precise about what it replaces.
When a Korean importer wires payment to a Japanese supplier today, the won leg and the yen leg do not travel together. The won leaves Seoul; it moves through one or more correspondent banks — intermediaries that hold accounts on behalf of other banks — before reaching a bank that can convert it and pay out in Tokyo. The two legs settle sequentially, not simultaneously, and that gap between “we sent” and “they received” is the structural source of what financial markets call Herstatt risk, as explained in the ECB Economic Bulletin’s April 2026 analysis of cross-border payment inefficiencies.
The term comes from the 1974 collapse of Bankhaus Herstatt, a German bank that received Deutsche marks from counterparties in FX trades before German regulators shut it down mid-business-day. The expected US dollar payments never arrived in New York. Counterparties had delivered their legs and received nothing, as documented in CLS Group’s Shaping FX series. The episode named an entire category of exposure that the correspondent banking system has never fully closed. According to the June 2026 BIS Quarterly Review of the 2025 Triennial Survey, only 36% of average daily FX settlement — roughly $5 trillion of approximately $14 trillion daily — settled via full payment-versus-payment protection in April 2025. A further $7.6 trillion settled using methods that mitigate but do not eliminate the risk, with more than $1.4 trillion exposed to full Herstatt-style settlement risk on any given day.
Continuous Linked Settlement (CLS), launched in 2002, materially reduced the aggregate risk pool. But CLS works through multilateral netting — it aggregates obligations across members and settles net differences simultaneously. It is not atomic at the individual transaction level; it processes in batches. And its settlement membership threshold effectively excludes the mid-tier institutional players that make up much of the Japan-Korea corporate payment corridor.
How Canton’s Two-Phase Commit Closes the Gap
The Musubi Project’s architecture attacks the problem differently. Canton Network uses a two-phase commit protocol coordinated by a component called the Global Synchronizer, whose role is defined at the protocol level. The mechanics work as follows:
A Daml smart contract specifies all the preconditions for a transaction: which parties are involved, what they owe each other, and what must be true for the transaction to complete. The Global Synchronizer then coordinates execution across all relevant Participant Nodes — the servers operated by the transacting institutions. Only when every Participant Node has independently validated its portion does the synchronizer commit the transaction to all ledgers simultaneously. If any validation fails at any node, the entire transaction reverts. Neither party ends up in a partial state.
Applied to the Musubi Project’s PvP architecture: the won payment and the yen payment are not two transactions scheduled in sequence. They are a single cryptographic event. The yen leg cannot settle unless the won leg settles at the same instant, and vice versa. There is no window during which one party has delivered and the other has not. The structural gap that produced Herstatt risk does not merely shrink — it ceases to exist.
ClearToken, an FCA-authorized digital financial market infrastructure provider, made the same architectural choice when it deployed CT Pay on Canton in March 2026, describing PvP on Canton as “the stablecoin equivalent of CLS, eliminating Herstatt risk.” The Musubi Project extends that model to the Japan-Korea bilateral corridor specifically.
Why Daml, and What Sub-Transaction Privacy Means for Banks
The smart contract language the Musubi Project uses, Daml, is not interchangeable with the general-purpose contract languages used on public blockchains like Ethereum. The distinction matters for institutional adoption.
Most smart contracts define execution logic but manage data visibility at the application layer — the software that sits on top of the blockchain. In Daml, contract visibility is protocol-enforced at the contract level — not managed by an application layer sitting on top of the blockchain. A Daml contract specifies which parties are stakeholders and what each stakeholder can see. No other network participant can read the content of a transaction in which they have no stake. This means a bank on the Musubi network executing a won-to-yen payment will not expose its trading position to other institutions on the same network — a requirement that is effectively non-negotiable for any institution operating under bank confidentiality standards.
This is what Canton’s designers call sub-transaction privacy, and it is the primary reason Canton has attracted more than 600 financial institutions including Goldman Sachs, BNP Paribas, Franklin Templeton, Broadridge, and Euroclear, with assets on the network valued at over $6 trillion. The network can be shared infrastructure without being transparent infrastructure.
What the Musubi Project Will Build, and Who Does What
The two firms have divided responsibilities along their existing strengths.
SBIDP, led by representative director Shimotsu Ryo, will develop the integration layer connecting the Musubi network to existing Japanese financial systems, and will leverage SBI Group’s established banking, securities, and asset management networks to recruit Japanese institutional participants. SBI Group became a Super Validator on Canton — one of the nodes involved in transaction approval and management — as part of its SBIDP formation announcement on July 28, 2026.
NodeInfra will lead development of the core payment protocol, the Daml smart contracts, and the developer tooling. The Korean firm will also handle participant onboarding for Korean financial institutions and custodians, and will support engagement with Korean regulators. Yang Young-seok, CEO of NodeInfra, said the firm intends to apply its Daml smart contract development expertise and digital asset infrastructure experience to delivering “a safe and trustworthy cross-border payment infrastructure.”
Beyond PvP atomic settlement, the project will implement distributed peer netting, which aggregates offsetting positions across participants before settling only net amounts on-chain. This reduces the gross volume of on-chain transactions required and lowers liquidity requirements for participating institutions. The network will also feature member-governed autonomous governance, giving participating financial institutions a direct voice in protocol evolution.
The Regulatory Gap the Project Must Navigate
The Musubi Project’s initial work will use test tokens — a synthetic yen test token and a synthetic won test token — rather than commercially issued regulated stablecoins. That is not a technical limitation. It is a legal one.
Japan’s stablecoin regulatory framework, activated by the Financial Services Agency on June 1, 2026, creates a clear compliance pathway for stablecoins issued by domestic trust banks. SBI Group’s JPYSC — a yen-backed stablecoin issued by SBI Shinsei Trust Bank under the Payment Services Act as a Type III Electronic Payment Instrument — carries no transaction cap, meaning it can handle institutional-scale payments without ceiling. JPYSC is the instrument SBIDP plans to connect to the Musubi network on the Japanese side.
The Korean side is a different situation. South Korea does not yet have a legal framework authorizing commercial issuance of Korean won-pegged stablecoins. The Digital Asset Basic Act (DABA), which would establish licensing, reserve, and redemption requirements for KRW stablecoins, has stalled in the National Assembly over a dispute between the Bank of Korea and the Financial Services Commission on issuance eligibility. South Korea’s ruling Democratic Party committed in late July 2026 to reintroduce the DABA in September 2026, as reported in TechTimes’ Kbank-HashKey MOU coverage, but until it passes, the Musubi Project’s KRW leg can only operate in test-token mode.
Shimotsu Ryo framed the project explicitly within SBI Group’s broader on-chain ambitions: JPYSC and the group’s financial networks are the institutional entry points that Musubi will connect to. The migration from test tokens to regulated stablecoins is a regulatory timeline question, not a technical one.
Why This Corridor, and Who Else Is Trying
The Japan-Korea bilateral corridor is one of the most active trade relationships in Asia and one of the most underdeveloped in terms of direct digital settlement infrastructure. South Korea exports roughly $29 billion to Japan annually while importing approximately $46 billion, for total bilateral goods trade of around $75 billion — all of it currently routed through USD-denominated correspondent channels.
Multiple parties identified the opportunity earlier this year. In February 2026, SBI Ripple Asia and DSRV partnered on joint research into blockchain payment rails between the two countries, using the XRP Ledger as the potential infrastructure. The Musubi Project uses Canton Network instead, prioritizing the institutional privacy architecture and the atomic PvP settlement protocol. The two efforts are distinct and potentially complementary — the XRP Ledger approach targets a different part of the settlement stack.
Asia-originated stablecoin payments have become the dominant global flow in this asset class, accounting for an estimated $245 billion and approximately 60% of total global stablecoin payment volume, according to McKinsey and Artemis Analytics. A direct won-yen settlement rail operating at institutional scale would, if it reaches sufficient liquidity and participant depth, capture a portion of that flow without converting through dollars — which is what “cutting out the dollar” means in practice for a business treasury team.
The project will not be the only Canton-based cross-border settlement infrastructure. ClearToken’s CT Pay operates on Canton for stablecoin FX. The US-UK Transatlantic Taskforce on Markets of the Future has committed 54 global financial firms to live deployment on Canton for repo markets. What makes the Musubi Project distinct is its bilateral currency-pair specificity: rather than building a generic settlement network, SBIDP and NodeInfra are building a dedicated rail for a specific corridor where the dollar routing cost is measurable, the trade volume is large, and both regulatory regimes are advancing toward stablecoin authorization.
Whether that specificity is a strength or a constraint will depend on how quickly DABA passes in Seoul, how many Korean financial institutions SBIDP and NodeInfra recruit, and whether the liquidity they assemble is sufficient to create a genuine alternative to the correspondent banking chain. The road ahead involves engineer exchanges, joint development programs, and a regulatory clock that neither company controls.
Frequently Asked QuestionsHow does Canton’s atomic settlement actually eliminate Herstatt risk rather than just reducing it?
The distinction is architectural. Continuous Linked Settlement, the existing PvP system used by the global FX market since 2002, reduces Herstatt risk through multilateral netting — it aggregates obligations across member institutions and settles net differences simultaneously. That approach still processes transactions in batches and has a membership threshold that excludes many mid-tier players. Canton’s two-phase commit protocol makes each individual transaction a single cryptographic event: the won payment and the yen payment are bound together so that both complete at exactly the same moment or neither completes at all. There is no settlement window. The gap that produces Herstatt risk does not narrow — it does not exist.
Why do Japan-Korea business payments currently go through US dollars at all?
The correspondent banking system connects banks without direct relationships through chains of intermediary institutions, and the US dollar is the world’s reserve currency — it is the language most correspondent banks speak. A Korean bank that wants to pay a Japanese bank converts won to dollars because dollars are the common medium that both countries’ correspondent chains support. Bilateral currency swap arrangements and direct settlement rails reduce this dependency but have not eliminated it for the high-volume corporate payment tier. The Musubi Project would create a dedicated direct rail for won-yen, bypassing the dollar conversion entirely, as detailed in the ECB’s April 2026 analysis of cross-border payment reform.
When will the Musubi Project use real stablecoins rather than test tokens?
On the Japanese side, the regulatory infrastructure already exists: JPYSC, SBI Shinsei Trust Bank’s yen stablecoin launched June 24, 2026, has no transaction cap under Japan’s Payment Services Act and is the designated on-chain yen instrument for the Musubi Project, as covered in TechTimes’ reporting on the JPYSC launch. On the Korean side, the Digital Asset Basic Act must pass the National Assembly before commercial KRW stablecoin issuance is authorized. South Korea’s ruling party targeted September 2026 for reintroduction of the bill; passage and implementation will take additional time after that. The Musubi Project timeline is, in practice, a function of Korea’s legislative calendar. For fuller context on where that legislative timeline stands, see TechTimes’ earlier reporting.
Could the Canton architecture used here work for other currency corridors that lack CLS coverage?
That is the larger implication the Musubi Project’s architecture opens. CLS covers approximately one-third of global FX settlement and is effectively accessible only to large settlement members. The majority of non-G10 bilateral currency corridors — including many high-volume corridors in Asia, Latin America, and the Middle East — have no equivalent atomic PvP infrastructure. If the Musubi Project demonstrates that a bilaterally governed Canton rail can attract sufficient institutional participation to provide competitive liquidity on a specific currency pair, the architecture is replicable. The Japan-Korea corridor is the proof of concept. Whether the proof holds will determine whether the model scales, a question the BIS has examined in its cross-border payment technology research.