AI will change merchant acquiring

The merchant acquiring sector is at a structural inflection point. Increasing transaction complexity, relentless fintech competition, and tightening regulatory scrutiny are eroding margins and exposing weaknesses in legacy operating models.

An article by David Birch, international keynote speaker, author, advisor, commentator on and investor in digital financial services. 

The annual Capgemini World Payments Report made for useful reading, as always. I found some of the comments around merchants particularly interesting. The report talks about the “service gaps” in key areas such as reliability and onboarding that explain why merchants are shifting to “PayTechs” for support.

I talked to Kartik Ramakrishnan, the CEO of the Financial Services SBU at Capgemini about this and he told me that with the right infrastructure and interfaces, with embedded workflows, banks can stay agile, scale and deliver seamless experiences across channels to meet the demands of merchants. And now there is the AI factor that could mean significant change, something that big players such as banks might be very well-placed to take advantage of.

I think we already have some vision for what AI can do for payments. Better risk models and so on. We already see new protocols coming into use so that agents can make payments on our behalf, such as Google’s Universal Commerce Protocol (UCP) and Agent Payment Protocol (AP2). These address “traditional” uses case: an agent buying your pants at Target.

What payments can do for AI is in many ways a more interesting field for speculation. The Agent Transaction eXchange Protocol (ATXP), founded by Stripe alumni, is food for thought because the transition to non-human users of payment systems has implications on business models. ATXP has positioned itself as the Stripe for bots. It focuses on creating a lightweight protocol stack for agent-to-agent transactions down to the sub-dollar level, anticipating the evolution of entirely new payment systems.

Existing systems are designed for human-initiated transactions mediated through compliance and user consent screens. They assume latency, explicit confirmation and high per-transaction overheads. Agent-to-agent payments may have very different demands: agents need to act within milliseconds and may pay fractions of a cent for data, computation, or API access and these transactions are simply not viable on existing rails.

Stablecoin Perspective

This all rather points towards stablecoins being used for trustful, autonomous exchange between nonhuman actors. Indeed, Michael Novogratz, the founder and chief executive officer of Galaxy Digital, recently told the Goldman Sachs Asia Leaders Conference in Hong Kong, that in the near future “the biggest user of stablecoins is going to be AI”.

(Last year I co-wrote a paper on agent-to-agent payments with Debbie Gamble, an expert on payments infrastructure from a long career in the UK and North America. In that paper we identified the potential for a new payments infrastructure to both re-energise past propositions such as micropayments and create new ones such as supply-chain currencies. If you are interested, the paper is called Agentic commerce and payments: Exploring the implications of robots paying robots and it was published in the Journal of Payment Strategy & Systems, vol.19, no. 1, p.72-84, Spring 2025.)

Business

We are seeing the birth of an AI-driven economy that will be dominated by machine actors, with agents paying other agents directly for the services they require and stablecoins have now become a genuine contender to satisfy these new payment needs. How will the merchant acquiring business respond?

Right now that business is growing at a decent rate. I looked at a few different projections and the broad consensus is that it will grow from something in the region of $25 billion today to something like $50 billion a decade from now. It’s no surprise to see it growing this way, what with worldwide activity in digital wallets and contactless and so on. Steady growth, nothing to see here, move on.

But while the global merchant acquiring industry has become structurally more attractive over the last decade, what is particularly interesting to me is how AI is changing it from a commodity business all about volumes and margins (as it was when I first got into retail payments) into a strategically important sector.

Why? Well as Radi El Haj of RS2 (a global payments processor and technology provider that powers some of the world’s leading banks and payment facilitators) told me, “acquirers are sitting on mountains of data and AI might finally give them the ability to use it”. Indeed.

AI and machine learning have been deployed across a variety of functions on the acquiring side for some time in fraud prevention, chargeback reduction and so on but AI is transitioning from risk management to mission‑critical infrastructure rather than another tool. AI-driven analytics turn transaction data into merchant insights that allow acquirers sell higher‑margin value-added services instead of competing only on basis points. AI is also used to support routing and orchestration to provide more sophisticated optimisation for merchants.

What is coming next, of course, is the era of agentic commerce and this means merchants will need acquiring infrastructure that can safely handle machine‑to‑machine transactions, tokenisation and automated decisioning. As AI agents intermediate demand, some merchants will lose control of customer journeys and data, so they will rely more on acquirers and payment platforms for visibility, attribution, and optimisation across fragmented channels.

I think that this means that AI makes acquiring strategically important, but only for players that evolve from processors to AI‑enabled platforms. The “boring” part (commodity transaction processing with thin margins) remains, but value is now concentrated in insights, orchestration and new agentic models being explored.

Strategically, that means acquiring is once again a battleground for banks, big PSPs, and fintechs: they are buying AI capabilities, building orchestration layers, and repositioning as commerce or “operating system” platforms rather than utilities. For incumbents still focused on scale processing without AI or software depth, acquiring will stay low‑margin and non‑strategic; for those that lean into AI‑driven services and agentic commerce, it is becoming one of the most attractive parts of payments again.

Kartik made the point that the top slots are occupied by bank-backed acquirers who control large enterprise merchants but that in the SME segment, volumes have shifted towards the PayTechs. Hence there is a real opportunity for the banks to win back some of this business in they are able to use AI effectively.

As set out in Capgemini’s report, banks have a clear opportunity to reclaim their position in merchant servicing by leveraging digital capabilities built to support and secure business strategy. They have the opportunity to scale through best-in-class delivery of value-added services because they can capitalise on trust and exploit a marketplace edge in data but the paytechs such as Adyen, Checkout and Stripe are “formidable rivals” to those incumbents that have in many cases been assembled over the years through acquisitions and still depend on platforms held together with gaffer tape.

Conclusions

AI has the potential to reshape every aspect of the acquiring value chain from merchant onboarding to fraud management, pricing, portfolio optimisation, and customer engagement. By integrating AI as an enterprise-wide operating capability, acquirers can move beyond incremental efficiency gains to capture new sources of growth, deepen merchant relationships and create sustainable differentiation.

The strategic imperative is clear: acquirers must function as AI-powered, data-driven service platforms that orchestrate intelligence, automation, and personalisation at scale.