Mastercard acquires BVNK for up to $1.8 billion after joining the agentic payments protocol
On March 18, 2026, the payments blockchain Tempo launched its mainnet and the Machine Payments Protocol — an open standard that lets AI agents request, authorize and settle payments entirely on their own. Among the launch partners were not only Anthropic, OpenAI, Shopify and Standard Chartered, but also the two players the protocol initially threatens most: Visa and Mastercard. Since then, Mastercard has completed its acquisition of the stablecoin company BVNK for up to $1.8 billion, Visa has rolled out its first public Crypto Labs product, and the same group of companies has launched a new stablecoin that sent Tempo's stablecoin supply up roughly 161 percent in a single week. The question is not whether the card networks are participating in the agentic economy — they are, actively. The question is whether they are defending their position or giving it up.
What actually happened
Tempo is a payments-focused layer-1 blockchain incubated by Stripe and the crypto venture firm Paradigm. According to Fortune reporting, the company raised $500 million in a Series A in October 2025 at a $5 billion valuation, with Thrive Capital, Greenoaks and others as investors. On March 18, 2026, the mainnet went live, and Stripe and the Tempo team simultaneously launched the Machine Payments Protocol (MPP), which they co-wrote (Startup Fortune).
The launch coalition was broad: Anthropic, OpenAI, DoorDash, Shopify, Revolut, Nubank, Ramp, Standard Chartered — and, which is the telling detail in this story, Visa and Mastercard. At launch, more than 100 services appeared in Tempo's MPP directory, from model providers and developer infrastructure to compute platforms and data services, with listings from OpenAI, Anthropic and Google.
But participation is only half the story. In the months after launch, both card networks built parallel, card-based alternatives — and this pattern is the starting point for this article's analysis.
How the protocol works — and why it threatens
MPP is built around an economic problem in the agentic economy: AI agents make many micro-payments — an API call here, a data purchase there — and traditional card processing is simply not built for that. Payments of a few cents sit poorly with a model in which every card transaction carries its own cost.
Tempo solves this by bundling small transactions into aggregated blockchain settlements instead of running one card transaction per API call. Fees average under $0.001, and the chain has no native token (Cryptobriefing). The absence of a token removes some of the speculative dynamics that often accompany new networks. The agent requests, authorizes and settles the payment itself.
This is the core of the potential threat to the card networks: if bulk agent-to-agent payments happen on a blockchain with fees under one cent, the foundation for per-transaction card fees erodes precisely on the payments that may come to dominate machine-to-machine commerce.
The hedge: the card networks' own track
What makes the picture more nuanced is what Visa and Mastercard did in addition to becoming MPP partners. According to Startup Fortune, both extended the protocol so that it also supports card-based payments — that is, the rails they themselves control. But they also built their own, card-native alternatives:
Visa CLI. Through Visa Crypto Labs, Visa rolled out a command-line tool that lets AI agents execute card payments straight from a terminal — no API keys, no pre-funded account required. It is Visa Crypto Labs' first public product, and is currently in closed beta with access available on request via GitHub authentication.
Mastercard's Agent Pay for Machines and BVNK. Mastercard launched Agent Pay for Machines in June 2026, as part of a protocol stack that, according to Forkast News, now also includes Stripe's MPP (March 2026) and Google's Universal Commerce Protocol (January 2026) (Forkast via Yahoo Finance). The company also completed the acquisition of the stablecoin infrastructure company BVNK for up to $1.8 billion — described as the largest stablecoin acquisition to date.
It is worth emphasizing: reading this as "hedging" — join the standard that threatens you, extend it to your own rails, and simultaneously build alternative card-native tracks — is analysis, not reported fact. Neither company has, in the available source material, described its strategy in such terms. But the pattern is documented: they are inside MPP and building what could function without it.
OUSD: the same coalition, the same logic — now in stablecoins
On September 30, 2026, the dollar-pegged stablecoin OUSD went live, developed by Open Standard with Coinbase, Mastercard, Shopify, Stripe and Visa as its five founding partners (Forbes via Yahoo Finance).
The construction is thought through. OUSD is issued by Bridge, the stablecoin infrastructure company Stripe bought for $1.1 billion in 2024. Reserves are held at BlackRock, Lead Bank and BNY, and Bridge will, according to its own statements, publish monthly reserve attestations. The five founders have committed more than $1 billion to seed initial liquidity, and each received an equal initial equity stake in Open Standard. According to the launch post, partners earn rewards proportional to the supply and activity they drive on their platforms. The coin runs natively on Base, Ethereum, Solana — and Tempo.
The effect on Tempo was immediate. According to DefiLlama data, Tempo's total stablecoin supply climbed by roughly $450 million, or 161 percent, in one week — to around $729 million (Cryptobriefing).
The OUSD coalition shows a similar "join and counter" logic to MPP: the same card networks, the same Stripe, the same Shopify — this time around the value layer of agentic payments themselves, not just the protocol.
The numbers on the capital flows
Several key figures sketch the opportunities and the scale — with caveats:
- $33 trillion in stablecoin transfer volume in 2025, up 72 percent year over year. Note: this figure comes from secondary reporting, and the underlying source is not stated.
- Visa's stablecoin settlement annualized at $20 billion as of September 2026, a 15-fold increase year over year, according to Forkast.
- Stripe's stablecoin card volume of $1.2 billion in September 2026, same source.
These flows are no longer marginal. But they say something about the infrastructure being built — not about whether agentic payments have actually reached mainstream commerce.
What remains unresolved
Several things should be kept open:
The sources are secondary. Everything above rests on reporting from Fortune, Forbes, Cryptobriefing and Forkast — not on direct statements or documentation from Stripe, Tempo, Visa, Mastercard or Open Standard. Launch partner lists, valuations and product claims are therefore only as solid as the underlying journalism.
OpenAI's and Visa's Tempo relationships are not final. Cryptobriefing notes that OpenAI's and Visa's partnerships with Tempo are "not yet finalized" (Cryptobriefing). They should not be presented as signed deals.
The partner counts conflict. The figures for how many companies are in the consortiums — over 100, over 140, over 200 — vary between sources. The difference may be due to different points in time or different definitions of "partner," but it is not resolved by the available material.
Commerce is still missing. Forkast points out that the protocol stack itself is nearly complete — MPP, Universal Commerce Protocol, Agent Pay for Machines — but that merchant adoption is not. There are payment rails for agents; there are far fewer stores that accept them.
What is left standing
What is established fact is that Visa and Mastercard are now present on both sides of a technological frontier their cards do not control: they are MPP partners and building card-native alternatives; they are OUSD founders and buying up stablecoin infrastructure for record sums. What is not established is what the strategy will actually lead to — whether the card networks manage to move agentic payments onto rails they own, or whether Tempo's economics (under $0.001 per transaction, aggregated settlements) gradually eat away the per-transaction model on the payments that matter most.
For now, the card duopoly has chosen not to stand and watch. Whether that is enough will be decided by commerce — and it remains the big, unfinished piece.