Stablecoin Settlement Architecture on Card Networks: How Visa and Mastercard Do It

Systems analysis only. This article describes payment architecture and public regulatory events; it is not investment, legal or financial advice.

Card payments feel instant, but the money behind them is not. When you tap a card, an authorization message crosses the network in about a second, yet the funds move between banks days later through net settlement over wires and central bank rails that sleep on weekends. Visa and Mastercard are now changing the last leg of that pipeline without touching the first. The result is a stablecoin settlement architecture in which issuers and acquirers settle network obligations with dollar tokens on public blockchains, while cardholders and merchants see nothing different.

This matters now because the numbers have stopped being pilots. Visa reported a $3.5 billion annualized stablecoin settlement run rate in late 2025 and more than $20 billion by September 2026, and Mastercard closed its BVNK acquisition in August 2026. Meanwhile the CLARITY Act, the US market-structure bill, failed a Senate cloture vote on September 15, 2026.

What this covers: where settlement sits in the card lifecycle, how each network wires stablecoins into treasury and reconciliation, a reference architecture with failure modes, the regulatory constraints that shape design, and a practical checklist.

Context and Background

A card transaction has three distinct phases that engineers often blur. Authorization is a real-time message exchange in which the acquirer asks the issuer whether the cardholder can pay, and the network routes it. Clearing is the batch exchange of final transaction records, typically once or several times daily. Settlement is the actual movement of money, where the network nets every participant’s obligations into a single position and members pay or receive that position through a settlement bank or central bank account.

Only the third phase touches money, and it is the slow one. Traditional settlement runs on banking-day calendars: Visa’s own description of its US stablecoin launch notes that stablecoin settlement supports seven-day windows, in contrast to the five-business-day cycle of conventional rails (Visa investor relations, December 16, 2025). A purchase on Friday evening generates an obligation that may not be funded until Monday, and a bank holiday stretches it further. Every hour of that gap is float that someone has to finance, and a settlement-risk exposure that someone has to cap.

The stablecoin idea is narrow: replace the final transfer of the net position, and only that, with a token that moves 24/7 and reaches finality in seconds to minutes depending on chain. It is not a new card scheme, and it does not change interchange, chargebacks or dispute rules. Readers who want the broader landscape of tokens, ramps and custody should start with our overview of stablecoin payment infrastructure in 2026, and those interested in the bank-led alternative can compare it with the bank stablecoin consortium architecture.

The public timeline is short. Visa ran its first USDC settlement pilot in 2021 and by December 2025 had launched US-domestic USDC settlement with Cross River Bank and Lead Bank on Solana. In April 2026 it reported a $7 billion annualized run rate, up 50% quarter over quarter, and added Arc, Base, Canton, Polygon and Tempo to its chain list, bringing the total to nine (Visa press release, April 29, 2026). Mastercard, for its part, announced on June 3, 2026 plans to expand settlement to include stablecoins with intraday, weekend and holiday options (Mastercard press release).

One caution on scale. A $20 billion annualized run rate sounds large, but Visa itself is described as settling roughly $15 trillion in payments annually (Fortune, July 16, 2026). Stablecoin settlement is therefore well under one percent of total network volume. The architecture matters because the growth rate is steep and because the design decisions being made now, about custody, netting and reconciliation, will be expensive to reverse, not because the current share is large.

Why settlement, and not authorization, is the target

Authorization has a latency budget of a second or two and an availability target above four nines. A public blockchain cannot honor either, and no network wants to put block times inside the authorization path. Settlement has the opposite profile: it is periodic, batch-oriented, tolerant of minutes of latency, and dominated by liquidity and counterparty-risk concerns rather than throughput. That mismatch is why every major network started here.

There is a second reason. Settlement is a bilateral treasury problem between regulated institutions that already have contracts, KYC files and legal recourse. Putting tokens under that arrangement does not require consumers to hold wallets or merchants to accept crypto. The on-chain leg is an implementation detail hidden behind existing member agreements, which is also why the approach scales politically inside banks.

Who is participating and in what role

The public participants fall into four groups. Issuers of the token, chiefly Circle for USDC, Paxos for PYUSD, USDG and USDP, and Ripple for RLUSD, create and redeem the asset. Settlement banks such as Cross River, Lead Bank and CBW Bank connect the token to fiat accounts. Infrastructure and acquiring firms such as Nuvei, ARQ and, after August 2026, BVNK inside Mastercard handle wallets, conversion and orchestration. Chains provide the ledger.

Each role maps to a distinct risk: issuer redemption risk, bank operational risk, orchestrator custody risk and chain finality risk. A useful way to read any announcement is to ask which of those four risks it moves, and which it leaves where it was.

The Reference Architecture: Stablecoin Settlement Inside the Card Lifecycle

Direct answer: In a card-network stablecoin settlement architecture, authorization and clearing remain unchanged. The network nets each member’s obligations as usual, then instructs or receives a stablecoin transfer to a network-controlled wallet instead of a wire. Treasury systems reconcile the on-chain confirmation to the ledger, and banks handle conversion between tokens and fiat at the edges.

Figure 1: The card lifecycle with a stablecoin fork at the final leg. Everything left of the net settlement position is unchanged; only the transfer method differs.

Figure 1 shows the structural point: the position calculation is the interface. Whatever produces a number for “issuer X owes the network Y dollars on this date” is untouched, and the choice of rail is a routing decision downstream of that number. That is why both networks describe stablecoin settlement as an option participants elect, not a replacement. Mastercard’s June announcement frames it as “giving partners more choice in how and when transactions are settled,” including settling in fiat or regulated stablecoins.

Layer 1: the unchanged card core

The network’s authorization switch, clearing engine and dispute systems keep operating on ISO 8583-style message flows and proprietary batch formats. Nothing in the token leg requires those systems to know that a chain exists. The only integration surface is the settlement-position output, which today feeds a bank-instruction generator producing wire or ACH-like payment orders.

This design keeps blast radius small. If a chain halts or a token is paused by its issuer, the network can still net positions and fall back to fiat, because the fiat rail is still connected. The fallback path is not a courtesy, it is what allows regulators and risk committees to approve the change at all.

Layer 2: settlement orchestration

Between the position and the transfer sits the new component: an orchestration layer that decides which asset on which chain satisfies each obligation. Inputs include the member’s elected asset, the network’s supported list, wallet address allow-lists, compliance screening results, and cutoff times. Outputs are signed transfer instructions or expected-inbound records.

Visa’s chain list makes the need obvious. With Avalanche, Ethereum, Solana and Stellar supported earlier and Arc, Base, Canton, Polygon and Tempo added in April 2026, a single “settle in USDC” instruction has nine possible destinations, each with different fees, finality behavior and custody arrangements. Mastercard’s list is similar: Arbitrum, Base, Canton, Ethereum, Polygon, Solana, Tempo and XRPL, across USDC, PYUSD, USDG, USDP, RLUSD and SoFiUSD.

An orchestrator therefore behaves like a small routing engine. It should be deterministic, auditable and configured per member, because a settlement instruction that lands on the wrong chain is not recoverable by a chargeback. This is the same pattern discussed in our piece on payment orchestration platform architecture, applied to treasury rather than checkout.

Layer 3: custody and the on/off-ramp

Tokens must be held somewhere. In practice that means institutional custody wallets or MPC-based key management controlled by the network or its settlement agent, with segregated addresses per member or per corridor. The other half of the layer is conversion liquidity: when a member funds in dollars but settles in tokens, or vice versa, someone must mint, redeem or trade at the boundary.

This is where acquisitions like Mastercard’s purchase of BVNK make architectural sense. Mastercard announced the $1.8 billion deal in March 2026 and completed it on August 3, 2026, saying it intends to integrate BVNK’s on-chain infrastructure to support settlement and treasury flows, cross-border B2B payments and payouts (Mastercard press release). The press release does not disclose volumes or licensing detail, and I will not infer them; the widely reported purchase price comes from the announcement coverage rather than the completion notice.

Visa took a different route for the on/off-ramp. In August 2026 it announced a partnership with Zerohash to let eligible Visa Direct clients prefund accounts and send stablecoin payouts, and CoinDesk reported that Visa then issued a request for proposal seeking a settlement and over-the-counter partner holding exchange licenses across the US, Canada, the UK and Singapore (CoinDesk, August 18, 2026). The contrast is instructive: Mastercard chose to own the ramp, Visa is assembling it from partners. Neither is obviously superior, and I would treat the RFP details as reported rather than confirmed by Visa.

Layer 4: ledger and reconciliation

The final layer is the one most often under-designed. Bank settlement produces statements on a schedule, camt.053-style end-of-day files that a reconciliation engine matches to expected positions. On-chain settlement produces an event stream: transaction hashes, block confirmations, and reorg risk on chains that permit it.

The reconciliation system must therefore treat the chain as an untrusted, asynchronous source and apply a finality policy: a transfer is not “settled” in the general ledger until it passes a per-chain confirmation rule. Getting that rule wrong in either direction is costly. Too lax and a reorged transfer creates a false credit; too strict and funds sit in limbo during the exact weekend windows the design was meant to fix.

Figure 2: The four layers of a card-network stablecoin settlement stack, from the unchanged card core down to ledger reconciliation.

Figure 2 is the structural view; the sequence view in the next section shows the same stack in time.

What Visa and Mastercard each emphasize

Reading the public materials side by side, Visa’s story is volume and plumbing. Its pilot began as USDC settlement between issuers or acquirers and Visa, has grown to nine chains, and is now paired with the Visa Stablecoin Platform launched July 16, 2026, which lets banks and fintechs mint, manage and move stablecoins through Visa while integrating with existing treasury and settlement workflows. Fortune reported it supports OUSD, USDC and USDG and reaches a network of about 15,000 financial institutions and more than 200 million merchants.

Mastercard’s story is optionality and ownership. Its June announcement is explicitly a set of plans subject to regulatory approval, with Cross River, Lead Bank, CBW Bank, ARQ and Nuvei expected among the first partners across the US and Latin America. It emphasizes multiple issuers on day one, rather than a single primary token, and pairs settlement with intraday, holiday and weekend timing. Its BVNK deal adds a wallet and orchestration layer it did not previously own.

For system designers the takeaway is that both networks converge on the same four-layer shape while differing in how much of layer 3 they own. That difference determines who carries custody risk and who carries licensing burden, which the next section examines in detail.

Deeper Analysis: The Daily Settlement Sequence, Liquidity, and Token Choice

A settlement day, step by step

The sequence below shows one settlement cycle for a single issuer and acquirer pair. It is a generic model assembled from the public descriptions of both networks, not a copy of either network’s internal procedure, which neither publishes in detail.

Figure 3: A generic stablecoin settlement cycle. The network nets positions, the debtor sends tokens to a network wallet, and the creditor is paid once finality is confirmed.

Throughout the day, authorizations and clearing files accumulate. At the cutoff, the network nets every member’s debits and credits into one position per currency. The issuer, which usually owes money because cardholders have spent against its credit lines or accounts, receives a settlement statement and sends the stablecoin amount to the network-controlled wallet. Once the chain confirms, the network sends the acquirer’s credit position to the acquirer’s wallet, and the acquirer reconciles.

Two properties deserve emphasis. First, the network sits in the middle in both directions, so it is the counterparty to each member, not the members to each other. That preserves the existing risk model, where the network guarantees settlement. Second, the middle wallet is where concentration risk lives, because it holds member funds in transit. Custody controls there deserve more scrutiny than anything on the card side.

The working-capital story, with an illustrative calculation

The strongest economic argument is not speed, it is float. Consider an illustrative acquirer processing $2 million of purchases per day, every day of the year including weekends. Under a five-business-day settlement calendar, the acquirer typically funds merchants ahead of network payment, so the gap between paying merchants and receiving network funds spans the weekend. If merchants are paid daily but network settlement lands only on business days, then over a Friday to Monday span the acquirer may be out of pocket for three days of volume, roughly $6 million. This is an illustrative figure, not a measured one.

At a 5% annual cost of funds, carrying $6 million for the weekend costs about $820 per weekend, or roughly $43,000 a year if every weekend looked the same. That is small against $730 million of annual volume, but it scales linearly and it is pure friction. Cross-border corridors, where pre-funding of nostro accounts ties up far larger balances, are where the economics become material, which is why both networks stress cross-border and weekend cases.

Now flip to the issuer side, which is where the most interesting public data lies. Stablecoin-linked card programs, meaning cards funded from stablecoin balances, must settle obligations to the network daily before their own customers’ payments arrive. Visa’s September 2026 update cited over 160 such programs and observed that some early-stage programs need only a few million dollars and settle every day, which makes conventional bank credit lines impractical (The Block, September 8, 2026).

Settlement-linked credit as an emergent layer

One response has been onchain credit. Credit Coop, as reported around Visa’s announcement, built a stablecoin-denominated revolving credit facility secured by settlement receivables, using Visa’s daily settlement files and smart contracts to automate funding and repayment. Reported outcomes include borrowing-cost reductions of up to 30% as more lenders joined, and Rain, a Visa Principal Member, financing roughly $2 billion since August 2023 with no defaults across thousands of repayments.

I treat those figures as vendor-reported and unaudited. They still illustrate a real architectural pattern: once settlement data is machine-readable and the settlement asset is programmable, the receivable itself becomes collateral that can be lent against in near-real time. Traditional bank settlement produced that data too, but not in a form a smart contract could consume.

This pattern also imports new risks. A facility that repays automatically from settlement flows depends on the file feed being correct, on the oracle or attestation path that brings it onchain, and on the network continuing to publish. Those are single points of failure that a plain bank line of credit did not have.

Choosing the settlement asset and chain

Not all tokens are equal for this purpose. The selection criteria I would apply are these:

Redemption certainty: who redeems, how fast, and under what legal claim. Under the GENIUS Act proposals, issuers must redeem within two business days, extending to seven calendar days during heavy redemption events, according to a law-firm summary of the OCC’s proposed rule (Sullivan & Cromwell). A settlement design should assume redemption is fast but not instantaneous.
Reserve quality and segregation: the OCC proposal calls for one-to-one reserves that are identifiable and segregated, composed of high-quality assets such as Treasury securities and Federal Reserve deposits.
Chain finality and fee predictability: a network that settles nine chains needs a per-chain confirmation policy and a fee model.
Compliance controls: freeze and blacklist capability at the issuer level is a feature for regulators and a risk for treasurers, because a frozen address inside the settlement wallet stalls a cycle.
Liquidity depth at ramps: a token that cannot be converted at size within the cutoff window is not a settlement asset regardless of its market capitalization.

Stablecoins versus tokenized deposits

The decision that architects most often get wrong is treating “stablecoin” and “tokenized deposit” as synonyms. A payment stablecoin is a bearer-style liability of a non-bank or bank-affiliate issuer, backed by reserves. A tokenized deposit is a commercial bank’s deposit liability represented on a ledger, and it inherits deposit insurance rules, bank capital treatment and the bank’s own balance sheet.

Dimension	Payment stablecoin	Tokenized deposit
Issuer	Permitted payment stablecoin issuer	Chartered bank
Backing	Segregated reserves, high-quality assets	Bank’s general balance sheet
Yield to holder	Prohibited by the GENIUS Act on payment stablecoins	Bank may pay interest, subject to product terms
Network reach	Cross-institution, cross-chain by design	Usually within a bank or consortium
Settlement finality	On-chain finality plus issuer redemption	Bank ledger plus interbank settlement
Best fit	Multi-party network settlement across banks	Intra-bank or consortium treasury movement

Networks currently lean on stablecoins because they span institutions without needing every bank to join a common ledger. Consortium approaches, covered in our analysis of how US Bank, Stellar and Qivalis structure bank stablecoin consortia, reflect the opposite bet. The likeliest end state is asset-agnostic orchestration, where the network’s settlement layer accepts whichever regulated liability the member holds, which is exactly the posture Mastercard describes.

New chains built for payments

Several chains on both networks’ lists were designed with payments in mind: Circle’s Arc and Tempo, alongside Canton for institutional privacy. Our analysis of Circle’s Arc mainnet as a stablecoin-native layer 1 covers why stablecoin-denominated gas and deterministic finality matter for treasury use. For a card network, predictable fees and sub-second finality reduce the reconciliation ambiguity described earlier.

The caveat is maturity. A chain that has run for a few months has no multi-year operating record, and a network that lists nine chains is implicitly betting that most of them will not be needed. In practice I expect concentration: a couple of large chains carrying most volume and the rest present for member choice and negotiating leverage.

What the CLARITY Act Stall Means for System Design

Regulation is the constraint most likely to be misread, so precision helps. Two separate US laws are involved, at very different stages.

The GENIUS Act, the stablecoin issuer law enacted in 2025, is in force but still being implemented. Its effective date is the earlier of 18 months after enactment (January 18, 2027) or 120 days after final implementing rules, per the summary cited above. The OCC published proposed implementing rules in early 2026 with comments due May 1, and the FDIC issued its own proposal in April. Its core provisions matter for settlement design: one-to-one segregated reserves, a redemption obligation, and a prohibition on issuers paying interest or yield merely for holding a payment stablecoin.

The CLARITY Act (H.R. 3633, the Digital Asset Market Clarity Act) is a broader market-structure bill. The House passed it on July 17, 2025 by 294 to 134. The Senate Banking Committee reported a substitute on June 1, 2026, but on September 15, 2026 the Senate did not invoke cloture on the motion to proceed, falling short of the 60 votes required (CoinDesk, September 15, 2026). Outlets differ on the exact tally, which was very close to even, so I am not quoting one. Senator Tillis entered a motion to reconsider, so the bill is stalled rather than dead.

Figure 4: The settlement layer sits downstream of issuer regulation. GENIUS governs the assets already; the stalled CLARITY Act mainly affects yield and market structure, not the settlement mechanics.

Why the stall barely touches settlement

The central dispute in the CLARITY negotiations was yield. Banks argued that stablecoin rewards would drain deposits; Citigroup’s Jane Fraser and JPMorgan executives were cited voicing that concern in coverage of the fight (Benzinga). The bill would have barred service providers and affiliates from paying passive, deposit-like returns on payment-stablecoin balances, extending beyond the GENIUS Act’s issuer-level ban.

Network settlement is a pass-through activity. Tokens arrive in a settlement wallet and leave within a cycle; nobody is paid yield for holding them overnight, and a well-designed program should not warehouse balances at all. So the yield fight mostly affects consumer wallets, exchanges and reward programs, not the settlement leg. That is the honest reading, and it matches the fact that Visa’s run rate kept climbing through the stall.

Where the stall does bite

There are three genuine effects. First, uncertainty about market-structure treatment: whether a given token, or the platform that moves it, is a security, commodity or payment instrument affects licensing for orchestrators and ramps. The SEC and CFTC are pursuing rulemaking, and SEC Chairman Paul Atkins was quoted saying guidance “won’t be durable without a law underpinning them.”

Second, yield-adjacent products built on settlement balances, such as sweeping idle prefunding into tokenized Treasuries, sit in a gray zone. The industry workaround described in coverage separates payment functions from lending or converts balances into tokenized securities, which is legal engineering more than architecture, and it can break if rules tighten.

Third, bank participation. Banks are the settlement legs of this architecture, and their willingness depends on capital and deposit treatment. A stalled bill leaves banks relying on interpretive letters and proposed rules, which slows risk approval even where the technology is ready. That is my analysis rather than a reported fact.

Design consequences

The practical consequence is to keep the architecture asset-agnostic and jurisdiction-configurable. Encode token eligibility, chain eligibility and yield-bearing prohibitions as policy data, not code. A rule change should be a configuration release, not a re-platforming, and the audit trail must show which policy version governed each settlement.

Outside the US the map differs, and networks operating globally must carry multiple regimes at once. The Visa RFP reported by CoinDesk, which required exchange licenses in the US, Canada, the UK and Singapore, is an example of that multi-jurisdiction burden landing on a single vendor slot.

Trade-offs, Gotchas, and What Goes Wrong

Issuer concentration. If most volume runs in one token, an issuer freeze, depeg or operational outage stalls settlement everywhere. Mastercard’s multi-issuer list is a hedge; Visa’s platform adding USDG and OUSD alongside USDC is a similar one. Hedging only works if members can actually switch assets at short notice, which requires funded wallets and tested ramps in each.

Finality mismatches. Chains differ in finality time and reorg behavior. A reconciliation rule tuned for a fast-finality chain will misfire on a slower one. Store the confirmation policy per chain and version it, and do not let a global default hide the difference.

Ramp liquidity at the cutoff. The most likely failure is not a hack but a queue: many members converting at the same daily cutoff, with thin off-ramp liquidity. Stagger cutoffs, pre-position inventory, and test with stress volumes rather than average ones.

Weekend operations. Seven-day settlement means seven-day operations. If treasury staff, fraud analysts and reconciliation teams work weekdays, the technology’s headline benefit becomes an unstaffed risk. Someone must own the alert on Sunday at 2 a.m.

Address hygiene and irreversibility. A wrong-address transfer cannot be recalled. Use allow-lists, whitelisting cool-down periods for new addresses, and dual approval above thresholds. This is one reason networks initiate transfers from controlled orchestration rather than letting members push freely.

Compliance freezes. Sanctions screening applies to wallets as it does to wires. A frozen counterparty inside a netted cycle forces a decision about whether to re-net, delay or settle the rest. Define that procedure before it happens.

Vendor-reported metrics. Run-rate figures are annualized from recent volume, not audited annual totals, and are reported by the networks themselves. Treat them as directional. I could not find independent verification of the $20 billion figure beyond trade-press coverage of Visa’s announcement, and secondary sites conflict on small details.

Legal certainty of finality. Technical finality is not legal finality. Whether a confirmed on-chain transfer discharges a payment obligation depends on the network rules and member agreements, and the law here is still developing. Contract language, not block depth, defines settlement.

Practical Recommendations

For a bank, processor or fintech evaluating stablecoin settlement on a card network, start small and keep the fiat path live. The strongest early use cases are weekend and holiday coverage, cross-border corridors with heavy prefunding, and stablecoin-funded card programs whose treasury already lives on chain. For a large domestic issuer settling once a day in dollars, the gain is modest and the operational burden real.

Sequence the work around the risks that hurt, not the features that demo well. Build reconciliation and custody controls first, because those are the layers where errors cost money. Add chain and asset choices only as fast as you can monitor them. Choose the settlement asset by redemption terms and ramp depth, not headline market cap, and keep at least two eligible assets configured.

Finally, decide who owns which risk in writing. Whether a network, a bank or an orchestrator carries custody, conversion and finality risk should be explicit in the contract. Ambiguity there is where disputes will emerge.

Checklist

Keep fiat settlement as a tested fallback path and rehearse switching to it.
Model float savings with your own volume and cost of funds before committing.
Define per-chain finality rules and reconcile against them, not a global default.
Run at least two eligible tokens and two ramps; test switching quarterly.
Use address allow-lists, cool-downs and dual approval for wallet changes.
Staff weekend operations and alerts to match seven-day settlement.
Store token, chain and yield rules as versioned policy configuration.
Confirm legal finality language with counsel; this article is not legal advice.

Frequently Asked Questions

How do Visa and Mastercard settle with stablecoins?

Both networks keep authorization and clearing unchanged, then let participating issuers, acquirers or banks meet their net settlement obligations by sending stablecoins on supported blockchains instead of using only wires and bank rails. Visa’s programme centers on USDC across nine chains; Mastercard has announced plans for USDC, Paxos tokens, RLUSD and others, subject to regulatory approval. Cardholders and merchants see no difference in how they pay or get paid.

Do consumers or merchants need crypto wallets for this?

No. Stablecoin settlement is a back-office change between regulated institutions. A shopper still taps a normal card, and a merchant still receives local currency from its acquirer under existing agreements. The wallets involved belong to banks, processors, the network and its settlement agents. Separate products, such as stablecoin-funded cards or stablecoin payouts, do involve users holding tokens, but they are distinct from network settlement.

How big is stablecoin settlement compared with total card volume?

Small but growing fast. Visa reported a $20 billion annualized stablecoin settlement run rate in September 2026, up more than 15 times year over year, against roughly $15 trillion of annual payments reported by Fortune. That is well under one percent. Run rates are annualized from recent activity and self-reported, so treat them as a trend indicator, not audited totals.

Did the CLARITY Act failure stop stablecoin settlement?

No. The Senate fell short of the 60 votes needed to proceed on September 15, 2026, but the GENIUS Act already governs payment stablecoin issuers and is being implemented. The CLARITY debate centered on yield and market structure, which touch consumer rewards and exchanges more than pass-through network settlement. The bill remains procedurally alive through a motion to reconsider, but timing is uncertain.

What is the difference between tokenized deposits and stablecoins?

A tokenized deposit is a bank’s own deposit liability represented on a ledger, so it carries the bank’s balance-sheet and deposit rules. A payment stablecoin is issued by a permitted issuer against segregated reserves and, under the GENIUS Act, cannot pay holders interest merely for holding it. Stablecoins move across institutions more easily; tokenized deposits often work best inside a bank or a consortium.

Why did Mastercard buy BVNK?

Mastercard said the purchase would connect digital assets with traditional payment rails for cross-border B2B payments, payouts, settlement and treasury flows. Architecturally it gives Mastercard ownership of the wallet, conversion and orchestration layer that Visa is instead sourcing from partners. Reported pricing was $1.8 billion, though the completion notice did not restate the price, and volumes were not disclosed by Mastercard.

Stablecoin Settlement on Card Networks: Visa and Mastercard Architecture