Settlement itself became a choice.

The stablecoin story this week has mostly been about a new coin. Underneath it, something quieter changed: settlement stopped being a fixed daily batch and became a product with knobs on it. And on the regulatory side, a deadline that had been theoretical for two years finally bit.

Mastercard makes settlement a choice, not a schedule

Mastercard confirmed it will offer intraday, weekend, and holiday card settlement across both fiat rails and regulated stablecoins, letting issuers and acquirers pick how and when a card transaction settles rather than waiting for the next business-day batch [1]. That is a genuine change in kind. Card settlement has run on the same batch cadence for decades, weekends and holidays included, and every reconciliation process downstream was built around that rhythm. Making it 24/7 and stablecoin-optional turns settlement itself into a competitive surface, which is a different problem for an ops team than "we added a new payment method."

Open USD adds the names that make it credible

The 140-plus-partner Open USD stablecoin, unveiled 30 June, kept adding weight this week: banks including BNY, U.S. Bank, Huntington, and Citizens sit alongside fintechs like Chime and Stripe, crypto firms Coinbase and Ripple, and the card networks Visa, Mastercard, and American Express [2]. The structural pitch has not changed: any participant can mint and redeem without fees or volume caps, and the interest earned on reserves gets distributed back to members after costs, rather than kept by a single issuer the way Circle and Tether keep theirs [2].

Nuvei completes an in-agent payment with no hand-off

On 2 July, Nuvei, Visa, Arvato Systems, and fashion brand Kings and Priests completed a live agentic-commerce proof of concept in which a merchant's AI agent initiated a purchase and paid for it inside the agent itself, with no redirect to a separate checkout page [3]. Several European issuers, including Alpha Bank, Piraeus Bank, and Bank Leumi, took part, settling on live Visa rails through a tokenised credential inside Visa Intelligent Commerce, governed by shopper-set guardrails such as spend caps and approved categories [3]. Nuvei is aiming for a protocol-agnostic execution layer, Nuvei Agentic, in the second half of the year, including a Know Your Agent registry and agent risk scoring [3].

The EU stopped giving unlicensed crypto firms time

The Markets in Crypto-Assets transitional period ended on 1 July. Every crypto-asset service provider operating in the bloc now has to hold full MiCA authorisation or stop serving EU customers [1]. That deadline had been on the calendar since 2023 and firms had years to prepare for it, which makes it a useful marker: regulators are done extending grace periods for crypto infrastructure that touches payments. The GENIUS Act's own deadline, final rules due from six US federal agencies by 18 July, is eight days out [1].

Read from the rails

Making settlement a choice sounds like a convenience feature. Run it operationally and it is a new failure surface, because every downstream reconciliation, every intraday liquidity forecast, every treasury sweep was written assuming one settlement time a day. Give issuers the option of intraday, weekend, and stablecoin settlement, and you have multiplied the number of states a transaction can be in when someone asks "did this actually land." That is not a reason not to build it. It is a reason the team shipping it needs a reconciliation model that covers every new settlement path on day one, not added after the first mismatched ledger.

The Nuvei proof of concept is the more interesting story long-term, because it collapses the step that every agentic-commerce protocol so far has kept: the hand-off from agent to a separate checkout. Remove that hand-off and you also remove the moment a human, or a fraud system built around a human clicking "buy," gets a last look. The guardrails Nuvei describes, spend caps and category limits, are exactly the controls a payments operator would reach for first. Whether they hold under real fraud pressure is the thing worth watching once volume moves past a proof of concept with one merchant.

Give an issuer more ways to settle and you have multiplied the number of states a transaction can be in when someone asks whether it landed. That is an operations problem before it is a product feature.