"SWIFT replacement" has been the ambient background noise of fintech conferences, investor decks, and central bank innovation labs since at least 2016. The proposition is always presented as obvious: SWIFT is slow, expensive, opaque, and was built for a world that no longer exists. Blockchain will replace it. The specific blockchain changes with the news cycle. What never changes is the framing. I have spent the past month reading primary sources rather than press releases, and the honest answer is that none of the three most serious candidates has done anything close to replacing SWIFT, and understanding why requires looking carefully at what each of them has actually built.

SWIFT processed an average of 53.3 million messages per day in 2024, across more than 11,500 connected financial institutions in 224 countries and territories, with 100% network availability for the year [1]. On 20 December 2024, a single day's traffic reached 59.5 million messages, a record [1]. These are not the numbers of an institution under existential competitive pressure. They are the numbers of an institution that handles more volume than it ever has, growing at 12.1% year-on-year, which its annual review describes as the highest growth rate in fifteen years. The story of blockchain alternatives to SWIFT needs to be read against that backdrop, not instead of it.

Network 1: Kinexys (JPMorgan)

JPMorgan's blockchain payment network has had three names in seven years: JPM Coin (February 2019), then Onyx by J.P. Morgan, then Kinexys, rebranded at the Singapore FinTech Festival in November 2024 [2]. The rebrand coincided with an announcement of on-chain FX settlement for dollar and euro pairs. The volume figure JPMorgan publishes on its Kinexys product pages is $2 billion-plus per day and $1.5 trillion-plus in cumulative notional value since inception [2]. Those numbers are real. The context around them matters as much as the numbers themselves.

Kinexys runs on a permissioned blockchain originally built on Quorum, JPMorgan's fork of Ethereum, which was open-sourced and subsequently transferred to ConsenSys via acquisition in 2020 [3]. The network is accessible only to JPMorgan's institutional clients. It settles transactions between accounts held within JPMorgan's own custody infrastructure. The $2 billion per day that moves through Kinexys does not cross the boundary between JPMorgan and any other bank: it moves between different JPMorgan clients within a single institution's ledger. This is architecturally significant. What Kinexys has built is not an interbank settlement network. It is an intrabank ledger with programmable money features. Siemens AG, the first external corporate client to use the programmable payments functionality, activated that feature in November 2023 [4]. The programmable element is genuinely interesting: it allows treasury teams to set conditions under which payments execute automatically, eliminating the manual approval steps that slow down corporate treasury operations. But that is a cash management product, not a cross-border settlement network.

JPMorgan has now extended Kinexys to operate on public blockchain infrastructure, specifically Base, an Ethereum Layer 2 network, which it describes as its first product offered on public blockchain [2]. This is a meaningful technical evolution. It signals that JPMorgan's engineers have concluded that the performance and security characteristics of public Layer 2 infrastructure are now adequate for institutional transaction flows. What it does not change is the fundamental architecture question: Kinexys remains a JPMorgan-native network, and the transactions that move through it represent intra-institutional settlement. That is valuable. It is not SWIFT.

Network 2: mBridge (BIS Innovation Hub)

mBridge is the most technically ambitious of the three and the most geopolitically complicated. It is a purpose-built, EVM-compatible blockchain developed by the Bank for International Settlements Innovation Hub in collaboration with the People's Bank of China, the Hong Kong Monetary Authority, the Bank of Thailand, and the Central Bank of the UAE, with Saudi Arabia's SAMA joining as a full participant in 2024 [5]. The 2022 pilot, which involved 20 commercial banks executing 164 cross-border payment and FX transactions totalling over $22 million over six weeks, was the first real-value cross-border CBDC pilot at meaningful institutional scale [6].

By November 2025, Atlantic Council data reported that mBridge had processed 4,047 transactions totalling $55.49 billion in cumulative value, a roughly 2,500-fold increase from the 2022 pilot [7]. That growth figure is accurate, and it is impressive. The composition of that volume is not. China's e-CNY accounted for approximately 95.3% of mBridge settlement volume by the same measure [7]. The network that was designed as a multilateral platform for cross-border CBDC interoperability has, in practice, functioned primarily as an internationalisation mechanism for China's central bank digital currency. That observation is not a criticism of the technical architecture, which is genuinely sophisticated. It is a description of the political economy of the network's usage, which reflects the fact that China has a deployed, production-scale CBDC and the other participating central banks do not, at anything close to comparable scale.

The BIS announced in October 2024 that it was handing the mBridge project over to the participating central banks [8]. BIS General Manager Agustín Carstens stated that this was not because the project had failed, but because "it is at a level where the partners can carry it on by themselves." The departure followed significant attention in international media connecting mBridge to BRICS discussions about alternatives to dollar-denominated settlement infrastructure, attention that the BIS found uncomfortable. The project continues under the central bank participants' direct governance, but the institutional imprimatur of BIS Innovation Hub sponsorship is no longer there. Whether that matters depends on how much of mBridge's credibility as a neutral multilateral platform derived from its BIS affiliation, which is a question I cannot answer with confidence from the primary record alone.

Network 3: Ripple and the XRP Ledger

Ripple is the hardest of the three to assess from primary sources, because it is a private company with limited independent disclosure obligations. The SEC case that shadowed the company for four years concluded in August 2025, when both Ripple and the SEC withdrew their appeals, leaving in place the $125 million civil penalty from the August 2024 final judgment [9]. The core question that occupied the litigation (whether XRP was a security) was answered, at least in the SDNY, with a split: institutional sales were deemed securities transactions; exchange sales were not. The legal uncertainty that had kept many regulated financial institutions at arms-length from Ripple's networks is formally resolved.

Ripple's own published figures for its On-Demand Liquidity product, which uses XRP as a bridge currency to settle cross-border transactions without pre-funded nostro accounts, cite over $15 billion in cross-border volume via ODL in 2024 and more than 300 financial institution partners on RippleNet globally [10]. These figures are not independently audited, and Ripple does not publish transaction-level data in a format that allows external verification. The XRP Ledger itself is public and its transaction data is verifiable on-chain: Q4 2024 saw approximately 167 million transactions on the ledger, across the full range of use cases including payments, token issuance, and decentralised exchange activity [11]. The settlement finality on the XRP Ledger is 3 to 5 seconds, confirmed in the official protocol documentation [12]. That speed is real and verifiable. What proportion of the ledger's transaction volume represents genuine cross-border institutional payment flows, as opposed to speculative and retail activity, cannot be read from the on-chain data without additional attribution work that Ripple has not published.

What SWIFT Has Actually Done

While the blockchain alternatives have been building, SWIFT has not been stationary. The SWIFT Global Payments Innovation initiative, launched in January 2017, has transformed the settlement experience on the existing network in ways that directly address the most cited criticisms of traditional SWIFT. By October 2024, 90% of cross-border payment instructions were reaching the destination bank within one hour, bank-to-bank, surpassing the G20's target of 75% within one hour by 2027, and achieving that level in 38 of 40 top receiving country corridors [13]. Approximately 60% of GPI payments are credited to the end beneficiary's account within 30 minutes [13]. This is a material change from the pre-GPI baseline of two to five business days, driven not by a technology replacement but by process re-engineering within the existing messaging infrastructure.

The ISO 20022 migration, which ended the coexistence period on 22 November 2025, is the other significant evolution [14]. The new standard carries richer structured data than the legacy MT format it replaces: more detailed payment purpose codes, full legal entity identifiers for ordering and beneficiary parties, and structured remittance information that enables automated reconciliation on the receiving end. By 24 November 2025, 97% of payment instructions on the SWIFT network were transmitted in ISO 20022 MX format [14]. The practical implication is that the data quality of a SWIFT-routed payment is now substantially better than it was three years ago, which matters because a significant fraction of the friction in cross-border payments has never been about settlement speed: it has been about the downstream reconciliation and compliance processing that happens after the funds arrive.

The Actual Competition

Looking at all three networks alongside SWIFT's own evolution produces a clearer picture than the "race to replace" framing suggests. Kinexys is solving an intra-institutional cash management problem for JPMorgan's largest clients. It is genuinely useful for that purpose and the volume figures show real institutional adoption. It does not compete with SWIFT on cross-bank, cross-border settlement, because it cannot: it is architecturally confined to the JPMorgan client universe. mBridge is attempting to solve the genuinely hard problem of cross-currency sovereign settlement without a common correspondent infrastructure, and the technical approach is sophisticated. The political economy of the network has, so far, made it function primarily as a distribution channel for China's CBDC ambitions rather than as a neutral multilateral platform. Ripple is offering a bridge currency model that eliminates the cost of pre-funded nostro accounts in specific bilateral corridors, which is real value in corridors where those costs are highest. The transparency problem with Ripple's published statistics makes it difficult to assess the actual scale of institutional adoption from sources other than Ripple itself.

None of the three has replicated what SWIFT does at its core: provide a standardised, trusted, universally accessible messaging layer across 224 countries and territories that any bank in any jurisdiction can connect to and rely on. The 53.3 million messages that moved through SWIFT every day in 2024 represent not just transaction volume but network effects that have accumulated over 52 years of institutional use. Breaking that network requires not just a better technology, but a coordination mechanism capable of persuading 11,500 institutions across 224 jurisdictions to simultaneously migrate to a new standard. No blockchain network has solved that problem, because it is not a technical problem. It is a governance and adoption problem, and those are considerably harder than re-engineering a settlement ledger.