Stablecoin card spending hit a record $789 million in September 2026, according to CryptoNews, pushing annualized volume toward $30 billion — up from $18 billion at the start of the year. Three distinct infrastructure models now compete to capture this flow: Mastercard's vertical integration via ...
"We're not building a crypto product. We're building payments infrastructure that happens to use stablecoins." — Simon Khalaf, CEO, Marqeta, October 16, 2026
Stablecoin card spending hit a record $789 million in September 2026, according to CryptoNews, pushing annualized volume toward $30 billion — up from $18 billion at the start of the year. Three distinct infrastructure models now compete to capture this flow: Mastercard's vertical integration via its $1.8 billion BVNK acquisition; the Coinbase-Marqeta horizontal partnership announced October 16; and Gnosis Pay's self-custodial on-chain settlement. Each model embeds different assumptions about where economic value accrues in the stablecoin-to-merchant payment chain.
The stakes are material. Visa reported its stablecoin settlement program reached a $20 billion annualized run rate in September 2026, a 15x year-over-year increase, across 160-plus card programs on nine blockchains. Mastercard's BVNK unit processes $30 billion annually across 130 countries. Yet stablecoin cards still represent less than 1% of global card spend, according to Spark Money research. The infrastructure race is not over market share today — it is over who controls the rails when that percentage moves from one to ten.
Crypto card spending crossed $18 billion on an annualized basis in January 2026, according to CoinDesk, nearly matching peer-to-peer stablecoin transfers at $19 billion. Monthly volumes climbed roughly 230% year over year through May, reaching a cumulative $7.8 billion for the first five months of the year. By September, monthly spend hit $789 million — a single-month record.
The broader stablecoin payments market provides context. Stablecoins settled $7.5 trillion on-chain in March 2026, according to KuCoin Research, but the vast majority of that volume is bot activity, trading, and internal routing. Allium's report estimated real stablecoin payments — actual economic transfers between distinct parties — at $401 billion to $527 billion from January through August, up 42% to 63% year over year. B2B stablecoin payments grew 733% year-on-year to approximately $226 billion annually, representing roughly 60% of all real payment activity.
Within this landscape, card-based spend is a narrow but rapidly growing wedge. The distinction matters: card payments require fiat conversion at the point of sale, meaning every dollar of card spend generates fee revenue across the custody-conversion-processing chain. The question is which model captures the most of that fee stack.
Mastercard closed its acquisition of BVNK on August 3, 2026, for $1.8 billion ($1.5 billion base plus $300 million earnout), according to Mastercard's investor relations filing. The deal made Mastercard the first major publicly listed card network to purchase its way into stablecoin infrastructure rather than partner into it.
BVNK's platform processes $30 billion annually across more than 130 countries, supporting stablecoin movements on Ethereum, Solana, and Tron. Post-acquisition, Mastercard announced plans to embed BVNK's capabilities across three vectors: 24/7 stablecoin settlement for processors and acquirers; stablecoin checkout via Mastercard's payment gateway; and settlement in USDC, PYUSD, and RLUSD with intraday and weekend settlement cycles.
Economic logic: Mastercard's model vertically integrates the stablecoin layer into the existing network. The card network controls both the traditional interchange and the on-chain conversion, compressing the value chain into a single entity. Mastercard captures issuer processing fees, network fees, and now stablecoin conversion margin. The trade-off is $1.8 billion of upfront capital deployed against an infrastructure that currently handles a fraction of Mastercard's $9-plus trillion annual network volume.
The Forrester assessment noted that by internalizing BVNK, Mastercard is wagering that proprietary stablecoin infrastructure is a core competitive moat rather than a peripheral feature. The earnout structure suggests both parties expect significant volume growth — BVNK's management team must hit performance targets to unlock the remaining $300 million.
On October 16, 2026, Marqeta announced an expanded partnership with Coinbase to enable stablecoin-backed card programs for enterprise clients. The arrangement divides the payment stack into three components: Coinbase custodies stablecoin balances and handles fiat conversion; Coinbase's Base blockchain provides the on-chain settlement layer; and Marqeta issues cards and manages processing through conventional card networks.
The announcement was notably light on specifics. According to The Block's reporting, Marqeta did not name a stablecoin, card network, issuing bank, or supported country for the first customer program. The company described the expected Base settlement as "fast and low cost" but provided no fee schedule or settlement-time target. It is also unclear whether transactions will settle individually on Base or whether programs can batch obligations before an on-chain transfer.
Economic logic: This model distributes value across specialized players. Marqeta earns its processing margin (the company reported $120 billion in total processing volume in Q2 2026, up 32% year over year, with net revenue of $176 million). Coinbase earns custody and conversion fees — its stablecoin revenue was $292 million in Q2 2026, though that figure is primarily interest income on USDC balances rather than conversion margin. Base earns gas fees, which currently run between $0.001 and $0.05 per transaction.
The economic appeal for Coinbase is twofold: it generates direct revenue from custody and conversion, and every card transaction that settles on Base increases Base's network utilization — a metric that feeds into Coinbase's broader L2 strategy. Marqeta gains access to a stablecoin capability without building any on-chain infrastructure. The risk is coordination cost: any dispute over conversion rates, settlement timing, or liability requires negotiation between independent corporate entities.
Coinbase shares fell 3% on the announcement day, according to Benzinga, suggesting the market viewed the deal as incrementally positive but not transformative.
Gnosis Pay represents the structurally distinct third model. Operating across 32 European countries, it is the only stablecoin card at scale where the cardholder retains self-custody of funds until the moment of payment. The card spends EURe (Monerium's EU-licensed euro stablecoin), USDC, or GBPe via a Visa debit card.
According to Paymentscan data, Gnosis Pay processed $211.4 million from January through October 2026, with monthly volumes ranging from $2.15 million (October) to $10.18 million (May). The platform holds 2.6% market share among stablecoin cards. These are small numbers compared with Rain's $3 billion-plus or KAST's $5 billion annualized volume.
Economic logic: Gnosis Pay eliminates the custodial intermediary entirely. Value accrues to the Gnosis Chain (gas fees and GNO token staking rewards — the card offers up to 5% GNO cashback) and to Monerium (as the licensed stablecoin issuer). There is no Coinbase taking a custody cut, no BVNK taking a conversion margin. The trade-off is friction: self-custody requires the user to manage keys, understand on-chain mechanics, and accept that the EEA's regulatory clarity is what makes this model viable — a condition not yet replicated in the United States.
The model implicitly argues that custody is an extractive intermediary layer, not a value-adding service. Whether the market agrees will depend on whether self-custodial UX can match the seamlessness of custodial alternatives.
Each model distributes fee revenue differently across the payment chain. A simplified framework for a $100 stablecoin card purchase:
Mastercard-BVNK (Vertical):
Coinbase-Marqeta (Horizontal):
Gnosis Pay (Self-Custodial):
The vertical model concentrates margin. The horizontal model distributes it. The self-custodial model minimizes it. In the economic-value framework, the question becomes: which intermediary layers are genuinely adding value versus extracting rent?
Beneath these consumer-facing models sits a B2B infrastructure layer dominated by Rain. The company raised $250 million at a $1.95 billion valuation in January 2026, according to CoinDesk. Rain powers stablecoin card programs for partners including Western Union, Nuvei, and KAST, processing $3-plus billion annually across 200-plus partners. Active cards increased 30x over the prior year.
KAST, Rain's most visible consumer brand, reported 1.4 million users and $5 billion in annualized transaction volume as of October 8, with 15-20% month-over-month growth. The company raised $80 million at a $600 million valuation in March 2026 and expects revenue to reach a $100 million annual run rate by year-end.
Rain's position mirrors the relationship between Stripe and its merchant clients: it provides infrastructure that multiple front-end brands can build upon. Forbes characterized Rain's fundraise as igniting a "payment stack war." The war is not between stablecoin and fiat — it is between parties competing to own each layer of a new, blockchain-native payment stack.
Visa has taken a chain-agnostic approach. As of mid-2026, Visa's stablecoin settlement pilot supports nine blockchains: Avalanche, Ethereum, Solana, Stellar, Arc, Base, Canton, Polygon, and Tempo. The program's $20 billion annualized run rate is spread across 160-plus card programs.
Visa carries more than 90% of on-chain crypto card volume despite both Visa and Mastercard supporting 130-plus programs collectively. Visa's strategy avoids the $1.8 billion acquisition cost Mastercard incurred; it also avoids locking into a proprietary conversion layer. The trade-off is dependency on third-party infrastructure providers for the stablecoin-to-fiat conversion — the exact layer Mastercard decided to own outright.
The addition of Base to Visa's supported chains is notable in the context of the Coinbase-Marqeta announcement. If Marqeta's stablecoin card programs run on Visa rails and settle on Base, Visa gets volume growth without building stablecoin infrastructure, Coinbase gets settlement activity, and Marqeta gets processing revenue. The three-party arrangement distributes risk along with value.
The stablecoin card market is not a single race. It is three parallel experiments in payment stack architecture, each with different assumptions about where value accrues and who should capture it.
Mastercard is betting that vertical control of the stablecoin layer — from issuance through conversion through settlement — will prove as durable a competitive advantage as its existing network effects. The $1.8 billion price tag on BVNK reflects a conviction that this infrastructure will process multiples of its current $30 billion within a few years.
Coinbase and Marqeta are betting that specialization and composability will outperform vertical integration — that an open, modular stack where each party does what it does best will attract more card programs and more volume than a walled garden. The sparseness of their October 16 announcement suggests the partnership is still early-stage.
Gnosis Pay is betting that users will eventually prefer to eliminate the custodial intermediary entirely, accepting higher friction in exchange for lower fees and direct asset control. At 2.6% market share, this remains a niche thesis, but one that could scale if self-custodial UX improves and regulatory frameworks expand beyond the EEA.
The data does not yet declare a winner. What it does show is that the battle over stablecoin payments has moved from whether cards would support stablecoins to who owns the infrastructure stack beneath them. The economics of each layer — custody, conversion, settlement, processing — will determine which model scales.