Mastercard has admitted 22 startups to the inaugural cohort of Start Path’s agentic commerce track, per Forkast. The program is the card network’s first dedicated accelerator effort aimed squarely at payments made by autonomous software agents rather than humans with cards. On its face this is a routine corporate accelerator announcement. In context, it is one more data point in a fast-forming pattern: every major payment incumbent — Visa, Mastercard, Stripe, PayPal — has now committed real institutional weight to agent commerce within the span of a single quarter.
What Mastercard Actually Announced
Start Path is Mastercard’s startup engagement program, historically focused on fintech and crypto companies at growth stage. The new agentic commerce track admits 22 startups, which Forkast frames as a signal that Mastercard believes agent-driven payments infrastructure is ready for institutional validation rather than another round of sandbox experiments. The details matter less than the structure: Mastercard is not building a single agent-payment product, it is seeding a portfolio of them and positioning itself as the connective tissue. That mirrors the strategy visible across its other recent moves — joining coalitions, backing standards work, and now cultivating the startup layer that would plug into whatever standards emerge.
Why the Card Networks Keep Showing Up
We noted last week that Visa, Mastercard and Fiserv joined the roughly 25-member Agentic Payments Alliance convened by Rain to standardize agent payments. India is simultaneously pushing agent payments on UPI as an open public standard, in explicit tension with the corporate schemes. The Start Path cohort is the commercialization layer on top of that standards posture. The card networks’ core asset — consumer trust and universal merchant acceptance — does not automatically transfer to agents, which have no card to swipe and no fraud-chargeback intuition. What the networks can offer is merchant reach and settlement certainty. What they cannot easily offer is what stablecoin-native protocols like x402 already demonstrate: machine-native payment over open rails, where an agent pays another agent directly without an interchange layer in the middle.
The Stablecoin Stakes
The open question for stablecoin.hot readers is settlement currency and settlement rail. Data from agent activity on both Base and the XRP Ledger shows autonomous agents default to stablecoins when they pay each other — dollar-pegged assets, not volatile native tokens, and not card-linked credentials. Mastercard’s bet assumes it can keep agent transactions inside its network economics, likely by wrapping agent payments in card-like authorization flows that merchants already accept. The stablecoin-native counter-position assumes agents will settle peer-to-peer in USDC or equivalents, bypassing interchange entirely. Both can win in different segments — B2B merchant-facing agent purchases favor the networks, machine-to-machine microtransactions favor open rails. The 22 startups in this cohort will effectively reveal which assumption has more commercial traction.
What to Watch
Watch which cohort members already settle in stablecoins versus those building card-wrapped agent flows — that composition will tell you how Mastercard expects the money to move. Watch whether any cohort startups engage with open standards efforts like x402 or India’s UPI framework, or whether Start Path’s mentorship steers them toward proprietary Mastercard APIs. And watch for the follow-on metric that actually matters: whether any of these startups process meaningful agent-payment volume within the program’s typical six-month window, or whether the cohort ends as a branding exercise. Mastercard has now spent its credibility on the thesis; the startups have to spend the transaction volume.