Warning – Big Blog 12 Pages

The last three weeks gave us a full sweep of processor earnings, and the results were a mixed bag (a phrase that undersells what actually happened). What we saw was not the processor sector moving together with a common tide. It was a sector splitting apart. Adyen raised guidance and jumped 16% in a day. Toast added a record number of locations and raised its full year outlook. On the “bad side of town”, we have Fiserv, which cut guidance, missed consensus, and is now down roughly 23% for the year with a brand new CEO. FIS is down about 36%. The gap between the best processors and the worst has never been this wide, and I do not think most investors have a working framework for why.

That framework is what I want to lay out here.

The Scoreboard

Start with what was actually reported.

Company	Most recent quarter	Growth	Notes
Adyen	H1 2026 (Aug 13)	Net revenue €1,302.9m, up 19% (21% constant currency)	Processed volume €803.8bn, up 24%. EBITDA margin 49%. Guidance raised to 21% to 23%. Shares up ~16% on the day
Toast	Q2 2026 (Aug 4)	Revenue $1.91bn, up 23%	Record 9,500 net location adds, ~180,000 locations. ARR $2.4bn, up 25%. GAAP operating margin 26%. Guidance raised
Shift4	Q2 2026	Gross revenue $1.295bn, up 34%	Revenue less network fees $624m, up 51% (11% organic). Volume $61bn, up 22%
Block	Q2 2026	Gross profit $3.17bn, up 25%	Record 27% adjusted operating margin. Square GPV $72.85bn, up 13%
Global Payments	Q2 2026	Adjusted net revenue $3.16bn, up 4% normalized	First full quarter post Worldpay. Adjusted EPS $3.46, up 12%
PayPal	Q2 2026	Revenue $8.68bn, up 5%	TPV $486.4bn. Branded checkout up 2%, unbranded PSP up 13%. Transaction margin dollars up 1%
FIS	Q2 2026	Pro forma organic up 5.3%	Banking up 6.1%, Capital Markets up 3.2%. Full year guide $13.6bn to $13.7bn, below consensus
Fiserv	Q2 2026 (Aug 6)	GAAP revenue $5.29bn, down 4%	Organic revenue down 4%. Full year organic guidance cut to negative 1% to flat. Adjusted EPS guide cut from $8.00 to $8.30 down to $7.20 to $7.40

Two companies raised guidance. One cut it by roughly 10% at the midpoint of EPS. Every one of these businesses moves money from a consumer to a merchant. They sit in the same value chain, they touch the same networks, they answer to the same regulators. So why is the outcome so different?

Merchants Pay For Everything In Payments

Let me restate the principle I keep coming back to, because everything else follows from it. As I wrote when PayPal needed a shakeup, in payments merchants “pay for everything”. Understanding their needs is not just important; it is paramount. Merchant discount rate (MDR) is the largest single revenue event in retail payments, and it is the number the rest of the ecosystem is organized around. MDR covers interchange, scheme fees, fx, acquiring and processing. Those components are distributed among issuers, networks, and acquirers. But the MDR is not the whole picture, and a few things sit outside it.

The first is fraud. Fraud losses are not covered by the merchant discount rate. In card not present the merchant absorbs them directly in the US, and the chargeback fees levied on top of the loss are a separate charge again. Interchange buys a guarantee against counterfeit and lost or stolen fraud in card present, and it buys the merchant far less than most people assume online. Note that the MRC estimates the cost of a dispute at $400/incident.

The second is the merchant’s own cost of acceptance. Returns handling, point of sale infrastructure, gateway redundancy, PCI compliance, treasury and reconciliation, chargeback labor, and the people who run all of it are direct merchant operating costs. They are not fraud, and they are not a redistribution of the MDR. They are what the merchant spends internally to be able to accept a payment at all, and they never appear on an invoice from anyone. That is the reason comparing headline acceptance rates across payment methods tells you very little about what acceptance actually costs.

The third is value added services. Processor VAS and network VAS are revenue streams in their own right, billed alongside the discount rate rather than carved out of it, and they are where the growth now sits. Visa’s value added services reached roughly a third of net revenue this quarter and Mastercard’s value added services and solutions roughly 40%. Processors sell their own stack of fraud screening, risk scoring, decline recovery, and orchestration on the same basis. Hold that thought, because who gets to own that VAS revenue line is the argument this whole piece is building toward.

Now the puzzle. The processor is the single entity closest to the merchant. It holds the contract, it holds the settlement account, it holds the operational relationship. When I asked merchants at MRC which partner they were most satisfied with, the near universal answer was their gateway or processor, because it is the only group that actively works to solve merchant operational issues. Proximity plus trust plus contractual control should equal pricing power.

It does not. Adyen earns an acquiring spread of roughly 28 basis points in the US and 24 basis points in Europe (UBS Research, “Brief thoughts on 2026E growth algo,”). Adyen’s blended take rate across all processed volume was 16.9 bps in 4Q25, and the H1 2026 figures imply ~16.2 basis points. Regulated debit in the US carries around 60 basis points of total MDR, of which the issuer takes about 40 and the network about 7 (Roneal Desai, “RD Stablecoin Memo,” June 2025). The party doing the least differentiated work, the issuer, takes the largest slice. The party closest to the merchant takes a sliver.

The entity nearest the customer is capturing the smallest share of the customer’s spend. That is not how value chains normally work, and the reason is worth understanding.

Processing Became Standardized

The core answer is that card processing has been standardized into something very close to a commodity, and the merchants who matter most engineered that outcome deliberately.

Large retailers no longer connect to a single processor. They connect to several, and they route between them based on cost, authorization rate, and uptime. Shopify alone connects to roughly 500 payment providers globally, and a former Shopify payments partnerships manager described the multi sourcing strategy as expected to deliver “flat to lower processing costs” (UBS Research, “Adyen’s opportunity within the Shopify ecosystem,” 20 July 2026). That phrase should worry every processor CFO. Multi sourcing is not a procurement tactic. It is a structural decision to prevent any single provider from ever accumulating leverage.

The clearest proof is what happened to debit. Durbin set the interchange cap for covered issuers at $0.21+ 5bps for non-exempt. That was supposed to be the floor. Instead, the largest merchants used routing rights, converted signature debit into PINless PIN, took the incremental risk onto their own books, and drove their effective debit cost well below the Durbin cap, into the $0.12 to $0.08 per item range. Durbin became the price only for merchants without the sophistication to route around it.

Note what that means for the processor. The merchant captured the savings. The processor executed the routing, absorbed the integration cost, and captured very little of the benefit. Competence became table stakes rather than a source of margin.

The Terminal Gambit

Processors understood the commoditization risk and responded with hardware. Clover at Fiserv, Square at Block, and Shift4’s ecosystem all exist to create stickiness at the physical point of sale and to open a channel for cross sell into higher margin services. The logic is sound. A merchant who runs the business on your terminal cannot switch processors on a Tuesday afternoon. And once you are in the back office you can sell payroll, capital, analytics, loyalty, and gift.

The execution has been uneven. Clover grew revenue 13% in the quarter excluding anticipation and nonrecurring items, with gross payment volume up 9%, and value added services now represent 25% of Clover revenue, up from 24%. Those are respectable numbers sitting inside a company whose overall organic revenue declined 4%. Block posted a record 27% adjusted operating margin with Square gross profit up 13%. Shift4 grew revenue less network fees 51%, though only 11% organically.

Ten years ago I wrote that the last mile was the most ripe for disruption. This is exactly what happened. The problem for the terminal strategy is that decoupling cuts both ways. If the merchant can choose the software independently of the processor, then the software is the asset and the processing is the accessory.

Payment Is The Best Event To Monetize

Payment sits at the heart of so many fintech business models for one simple reason. It is the best event to monetize. It is the moment where value is confirmed, where intent becomes obligation, where data is richest and where the customer has already decided to spend. Getting into the payment flow gives you an opportunity to price.

But notice the phrasing. Getting into the flow gives you the opportunity to price. It does not tell you what to charge for. The processors that are struggling treated the payment event as the product. The processors that are winning treat the payment event as the entry point to something the merchant values more.

eCommerce Is Where The Margin Lives

Card not present is the highest margin category in acquiring, and it is the core of growth for Stripe, Adyen, and PayPal’s Braintree. Complexity is the reason: fraud screening, authorization optimization, tokenization, network token management, cross border acceptance, local payment methods, and subscription logic. Complexity resists commoditization in a way that a card swipe does not.

Stripe published its annual letter in February. Businesses on Stripe processed $1.9 trillion in 2025, up 34% from 2024, which the company noted was “equivalent to 1.6% of global GDP, even as the economy bifurcated.” Stripe also stated it “remained robustly profitable, allowing us to continue investing heavily in product development (with more than 350 product updates last year) as well as acquisitions.” Its programmable financial services now power more than five million businesses, including 90% of the Dow Jones Industrial Average and 80% of the Nasdaq 100. A February tender valued the company at $159 billion.

Adyen’s H1 tells a similar story from the enterprise side. Processed volume up 24%, net revenue up 21% w/ EBITDA margin at 49%, and guidance raised even while the company absorbed two acquisitions (Talon.One in loyalty and Orb in billing). Co CEO Pieter van der Does framed it precisely: “By expanding our role well beyond payments, we execute our long term strategy and solve deeper structural complexity for our merchants, which fundamentally strengthens our customer relationships.”

Read that quote again. It is the whole thesis. Beyond payments. Structural complexity. Not faster authorization, not cheaper transactions.

Braintree is the counterexample. PayPal’s unbranded PSP volume grew 13% while branded checkout grew 2%, which means the low margin business is growing six times faster than the high margin business. Braintree does roughly 45% of PayPal’s total volume and contributes about 14% of transaction margin dollars excluding interest income. Volume without margin is not a business, it is a treadmill.

The JPMorgan Model

JPMorgan merchant services is one of the largest players in eCommerce acquiring, with Amazon, Google, and Apple among the names sending volume. But JPM’s play is fundamentally different. It is a low cost, high reliability, pass through provider. Deliberately dumb eCom pipe.

That is a perfectly rational strategy for a bank that already earns on the issuing side, on deposits, and on treasury services. Acquiring becomes a defensive product to keep a corporate relationship intact rather than a profit center in its own right. The consequence for everyone else is that the largest and most attractive merchants in the market are anchored by a competitor with structurally lower return requirements. Enterprise acquiring revenue in the US is forecast to grow from $10.3 billion in 2025 to $12.3 billion in 2029 (UBS Research, “Analysis of Combined Company Mix,” 14 April 2026), roughly 4% a year, the slowest of the four segments UBS tracks. The biggest merchants generate the least profit growth.

Vertical Specialization: The Toast Answer

Toast is the clearest proof that the winning move is to stop being a processor. Toast added 9,500 net locations in the quarter, an all time record, to reach approximately 180,000 locations. Revenue grew 23% to $1.91 billion, ARR grew 25% to $2.4 billion, and GAAP operating margin reached 26% against $80 million of operating income a year earlier. Toast raised full year guidance twice over on recurring gross profit and EBITDA.

The core payments net take rate is about 50 basis points (UBS Research). That is a good rate, roughly double Adyen’s US acquiring spread, and it exists because Toast is not selling processing. Toast sells the restaurant operating system: point of sale, kitchen display, payroll, scheduling, inventory, online ordering, loyalty, marketing, and Toast Capital when the owner needs to finance that new pizza oven. Payments is bundled inside an ERP and CRM system that the restaurant cannot run without. Autonomous notes the new vertical expansion is approaching roughly $200 million of ARR, nearly doubling, and that Toast IQ Grow reached $10 million of ARR faster than any product in company history.

I flagged this pattern in Strategic Bets in Retail Payments: platforms like Shopify, Toast, and Amazon now control 30% to 40% of consumer to business commerce in the US and EU, and their value lies not in moving money but in owning the context of spend. Toast is that thesis with a P&L attached.

UBS quantifies the shift. US merchant acquiring revenue running through software platforms grew from $3.2 billion in 2016 to $13.3 billion in 2025 and is forecast to reach $22.2 billion by 2029. Over the same period, traditional SMB direct distribution, the classic ISO and processor channel, peaked around $8.2 billion in 2023 and is forecast to fall to $5.2 billion by 2029 (UBS Research, “Analysis of Combined Company Mix,” 14 April 2026). One channel triples. The other shrinks by a third. UBS concludes that “software led distribution is increasingly becoming the channel required to participate in the ~70%+ of industry revenues sourced via underlying SMBs.”

This is the same conclusion I reached ten years ago from the merchant side. As I said then, the long tail is where the margin lives in payments, particularly in acquiring. Profitability in payments is in the long tail, in Stripe and Toast, and not in the top 20. Fiserv, FIS, and Global Payments own an enormous share of volume and a shrinking share of profit growth.

Fiserv: The Worst Position In The Sector

Which brings me to the company in the most difficult spot. The quarter was ugly. Revenue down 4%, organic revenue down 4%, adjusted EPS of $1.84 against consensus near $1.91, and a full year organic guide cut to negative 1% to flat. Adjusted EPS guidance came down from a range of $8.00 to $8.30 to a range of $7.20 to $7.40. That is not a rounding adjustment, that is a reset. The stock is down roughly 23% this year and roughly 60% over twelve months. Michael Lyons resigned as CEO in June, president Dhivya Suryadevara departed in July, and Takis Georgakopoulos has taken over. Jana Partners is applying activist pressure.

Autonomous Research downgraded Fiserv from Outperform to Underperform in November 2025 and cut the price target from $165 to $60. Kenneth Suchoski’s thesis is worth restating because it explains the trajectory rather than the quarter: trough EBIT margin guidance of 33% against a pre First Data peak of 32.8%, meaning years of integration and leverage produced no margin improvement at all; churn risk across Clover, the merchant back book, and core banking; a pattern of overcharging clients; and product development spend that fell from 8% to 9% of revenue down to 6% to 7% after the First Data acquisition (Kenneth Suchoski, Autonomous Research, November 2025).

That last number is the one that matters. A processor facing commoditization cut investment in the only thing that reverses commoditization. Meanwhile Fiserv’s legacy relationship led distribution sits squarely in the swimlane UBS forecasts to shrink by a third.

Then there is Star. I wrote in July about the retailer and industry feedback on a possible Star sale and separately about the Zelle at POS hypothetical. My read then was 80% confidence it would not happen, that the political timing was the worst possible, and that Fiserv had purposefully leaked the discussions as a new CEO looked to divest and focus. The sell side agrees on the odds. Autonomous calls a bank consortium acquisition “low probability,” values Star and Accel at $3.5 billion to $5.0 billion against a Fiserv enterprise value near $55 billion, and UBS estimates the impact on Visa at under 2% of net revenue even in the scenario where Chase acquires it.

So the asset is worth roughly 7% of the company, the probability is low, and the strategic logic depends on regulators tolerating an obvious effort to end run Durbin. Georgakopoulos now says the debit networks are “part of the holistic review” and that “if we are not best in class, then we will look at what are the alternatives.” That is a reasonable thing for a new CEO to say. It is not a growth strategy.

What I do not see anywhere in the Fiserv story is a plan to reverse the share trajectory. Divesting assets improves the multiple arithmetic. It does not make a merchant choose you.

The Future: Authentication Crunch Is Coming

Everything above describes the sector as it exists today. Now consider what happens next, because I think the processor group is set for another transformational crunch and very few people are pricing it.

One of my favorite quotes in this industry comes from the late Ross Anderson: if you solve for authentication in payments, everything else is just accounting.

Sit with that. A very large share of what a processor sells today exists because we cannot reliably establish who is on the other side of a transaction. Fraud screening, risk scoring, chargeback management, device fingerprinting, velocity rules, decline recovery, compelling evidence workflows, 3DS orchestration. This is a significant portion of the value added services stack that processors use to defend their spread.

Authentication is improving fast, and critically, it is improving inside the networks and the platforms rather than inside the processors. Visa reports tokens now exceed 40% of switched transactions. Passkeys are becoming the default on both major mobile platforms. Network level authentication and identity services are compounding.

The fraud disparity I saw at MRC makes the point sharper. Large merchants like Amazon and Walmart run around 7 basis points of fraud while mid tier merchants run around 50. Fraud attacks the weakest link. As authentication improves systemically, the weak links get stronger, the addressable fraud pool shrinks, and a meaningful slice of processor value added services revenue shrinks with it.

So how do you differentiate as a processor when authentication is held within the network? The answer is uncomfortable and simple. You have to do a lot more than card processing. Toast does. Adyen does. Stripe does. Fiserv, at its current rate of product investment, does not.

What This Says About Stripe And PayPal

This dynamic is exactly why I read the Stripe and Advent offer for PayPal the way I do. The facts as they stand: on 15 July, Stripe and Advent offered $60.50 per share, valuing PayPal above $53 billion. PayPal’s board rejected it on 20 July as undervaluing the company. As of this week the parties are reportedly still talking, and PayPal trades below the bid, which tells you what the market thinks of completion odds.

I wrote up the deal and then gave my opinion on it, so I will not repeat the full argument. The short version: of course Stripe wants a giant “on us” network. Combining Stripe’s merchant side with PayPal and Venmo’s consumer side and settling internally is one of the most defensible structures in financial services. It is why American Express exists.

But getting the consumer side and getting off cards are two very different challenges. PayPal’s volume is roughly 85% dependent on cards. You cannot rewrite the rules of a three party network when the overwhelming majority of your volume settles through four party rails whose rules you do not control. And the consumer asset you are buying grew active accounts 0.3% year over year while eCommerce compounds near 8%.

The authentication point sharpens this. If better authentication genuinely commoditizes the risk and fraud layer, then the thing Stripe is buying (scale, volume, a consumer brand) is worth less than the thing Stripe already has (product velocity, developer distribution, and complexity that merchants will pay for).

Issuers Want The Processors Out (won’t happen)

There is a parallel campaign worth naming. Issuers look at the roughly 40 basis points that the acquiring and processing layer takes and see it as part of the too many mouths to feed problem in payments. The logical issuer response is to build better authentication and use it to push the processing layer out of the value chain, replacing processor value added services with issuer value added services.

It will not work, and the reason is merchant behavior.

Merchants have no interest in dealing directly with any issuer, and none at all in taking on issuer value added services to replace processor services. I asked merchants directly about an issuer built eCommerce wallet and the consensus response was blunt: this is a typical tone deaf response from issuers that have refused to engage in solving the problems we have today, why would we want to educate consumers to use something from a group that never listens to us. Most added that they would depend on their processor to make the recommendation. That last clause is the entire ballgame.

More fundamentally, the largest merchants want to keep the risk and manage the fraud themselves. Amazon and Walmart run 7 basis points of fraud because they invested billions in doing it better than anyone would do it for them. They are not going to outsource their best cost advantage to an issuer consortium. Only the high fraud merchants genuinely want someone else’s solution, and they are the least attractive customers to serve.

So the issuer campaign against processor economics runs into the same wall every issuer led initiative has hit for fifteen years. The merchant is the one paying, and the merchant is not asking for this.

A new Network? Europe, Google, And The eIDAS Question (no way)

In Europe a different attack is forming. Google is hoping that eIDAS digital identity will let non card schemes compete at parity with cards, with Google absorbing much of the processing and orchestration role in eCommerce and agentic commerce. They have done versions of this in India with UPI and in Brazil with Pix, and as I have argued, this battle is Google’s to lose. Google is delivering digital ID passes in Google Wallet across Spain, Italy, France, Ireland, and Estonia this summer, alongside a Direct Checkout feature. Every EU member state must offer a digital identity wallet by the end of 2026.

I am skeptical, for two reasons.

First, credentials are not trust. I laid this out in Why eIDAS Will Fail in Banking: banks do not verify identity, they absorb liability, and eIDAS assumes trust is transferable while banking law says it is not. A cryptographically perfect credential does not answer the supervisor’s question about why this customer was classified this way. The same structural problem applies to payment guarantees.

Second, and more practically, European debit cards are already a very efficient network. Regulated interchange is 20 basis points for debit and 30 for credit. Merchants are not in pain. Banks earn that interchange and have no incentive to fund an alternative that pays them less. UBS makes this point directly, noting that European banks “would be unlikely to opt for a network alternative that reduces interchange revenue, in addition to the card networks being high functioning and relatively low cost to merchants today” (UBS Research, “Addressing investor questions around European payments efforts,” 27 July 2026).

Wero is the most credible attempt yet because it did not try to build a card scheme. It bought iDEAL and Payconiq, started with peer to peer to solve the cold start problem, and is now moving into eCommerce with more than 50 million users. It deserves respect. But UBS sizes the realistic damage carefully: if Wero and the digital euro together captured 10% of the genuinely contestable European revenue pool at some distant future point, which would count as success, the impact on Visa and Mastercard net revenue would be roughly 75 to 150 basis points. The near term effect is pricing pressure at negotiation time, not volume loss. That is a headwind, not a displacement.

Europe has tried this repeatedly. EAPS, PayFair, Monnet, and EPI 1.0 all failed on the same point: banks lacked a compelling economic incentive to fund an alternative to a system that already works. As I wrote in Europe’s Siege, commercial constructs are the true engines of the payment ecosystem, and Europe keeps trying to build them with legislation.

On Stablecoins, Briefly

I will keep this short because I have made the argument at length. Stablecoins are not free, and they are not a replacement for cards in consumer payments. They are another rail, and they will do well where cards do not play: micropayments, business to business, cross border, disbursements, and uncarded markets.

The arithmetic is the argument. Take a $65 card not present debit transaction. Today the total merchant discount is around 58 basis points. Run the same transaction on a stablecoin backed card and the merchant cost rises to roughly 61 basis points, because you add gas fees and an offramp cost without removing the issuer. Run it as a direct stablecoin wallet payment and the headline drops to 40 basis points, but once you add back the fraud you reintroduce by leaving the card authentication and chargeback framework, the true merchant cost is around 52 basis points (Roneal Desai, “RD Stablecoin Memo,” June 2025). Net saving: roughly 7 basis points, in exchange for giving up card benefits, consumer protections, and taking on right tail fraud risk.

No merchant CFO takes that trade. I paid $0.53 in Ethereum gas on a $2.50 test transaction, a 21% effective rate, worse than any interchange in the world. Yes, Tempo and similar infrastructure will fix the gas problem. But solving gas fees gets you to a 7 basis point advantage that evaporates the moment fraud shows up.

For processors specifically, stablecoins are not the disruption to fear. Authentication is.

On Visa And Mastercard, Less Briefly

While the processor group fragments, the networks keep compounding, and this quarter reinforced why they remain my largest personal holdings. Visa reported fiscal Q3 net revenue of $11.6 billion, up 14%, with organic constant currency growth of 12% and value added services up 34% to roughly $3.8 billion, now approximately a third of net revenue. Mastercard reported net revenue of $9.3 billion, up 14%, with value added services and solutions at roughly 40% of revenue growing high teens and contributing about 60% of total revenue growth. Both raised guidance.

Autonomous titled its July analysis “VAStly Durable,” and the framing is right: “investors are underestimating the durability of VAS growth, with Visa and Mastercard still organically growing VAS at ~22% and ~18% respectively, even after adjusting for major sporting events” (Kenneth Suchoski, Autonomous Research, 15 July 2026).

Here is why this matters for the processor question. The networks are capturing the authentication layer, the tokenization layer, the risk scoring layer, and the identity layer, and monetizing all of it as value added services. Those are precisely the capabilities that processors have used to defend their spread against commoditization. Autonomous also expects both networks to be near term beneficiaries of AI, as banks raise tokenization rates to defend against AI driven fraud and expand cybersecurity budgets.

Visa is now pushing further into issuer processing and core banking through Pismo, having entered 19 new markets since the acquisition, targeting a US total addressable market of roughly $20 billion to $25 billion by 2030. The networks are moving into the processing layer while the processing layer struggles to move anywhere.

And the merchant litigation settlement removed a 21 year overhang while preserving the fundamental economics. The interchange reduction is real, the Honor All Cards repeal is real, and neither breaks the model.

The Hurdles: The Next Five Years

Where does this leave the processing industry? Six problems, roughly in order of severity.

Authentication compresses the risk premium. As tokens, passkeys, and network authentication mature, the fraud and risk services layer shrinks. Processors that priced complexity they did not create will find the complexity gone.

Software led distribution keeps taking share. The SMB direct channel falls from $7.6 billion to $5.2 billion by 2029 while software platforms climb to $22.2 billion. Any processor whose distribution is a sales force rather than a software platform is fighting the tide.

Multi sourcing caps pricing permanently. Shopify’s 500 provider architecture is the template. Enterprise merchants will never again allow one processor enough concentration to hold pricing power.

The largest merchants are the least profitable. Enterprise acquiring grows 4% a year, and JPMorgan’s deliberately low margin, pass through model anchors pricing at the top of the market.

Consolidation math is running out. Fiserv and First Data, Global Payments and Worldpay, FIS and Worldpay and then not Worldpay. Scale was supposed to deliver margin. Fiserv’s trough EBIT margin guidance of 33% against a pre First Data peak of 32.8% is the verdict on that thesis.

AI cuts both ways. Autonomous argues persuasively that merchant acquiring faces less AI disruption risk than software generally, since hardware, distribution, and scale are real advantages. I agree for the acquiring function. I am less relaxed about the services layer stacked on top, which is exactly where the margin is.

What To Watch

A few markers that will tell you whether this framework is right.

• Fiserv’s product development spend recovers toward 8% or 9% of revenue. If Georgakopoulos is serious, that line moves before the revenue line does.
• Toast’s non restaurant vertical ARR. If the roughly $200 million run rate keeps doubling, vertical specialization is a repeatable playbook rather than a restaurant specific accident.
• Adyen’s EBITDA margin through the Talon.One and Orb integration. Management guided to roughly one point of dilution against a 2028 target above 55%. Holding that line while acquiring into loyalty and billing would prove the beyond payments thesis.
• PayPal’s branded checkout growth rate. It has held at 2% for two quarters, which is stabilization. Three percent would be a genuine turn. One percent, and the board’s rejection of $60.50 becomes hard to defend.
• Value added services growth at Visa and Mastercard for the first sign of deceleration. That is the single best proxy for whether the authentication and services layer is compounding or saturating.
• Star Sale. I remain around 80% confident it does not happen.

The Verdict

The gap between the best and worst processors is not a valuation anomaly waiting to close. It is the market correctly pricing two different businesses that happen to share an industry label.

One group sells card processing and defends its spread with services that better authentication will erode. That group is Fiserv, FIS, and to a degree Global Payments, and its problem is structural rather than cyclical.

The other group uses the payment as an entry point and sells the merchant something harder to replace: a restaurant operating system, a global complexity abstraction layer, a programmable financial stack. That group is Toast, Adyen, and Stripe, and its take rate survives because it is not really a payment take rate.

For investors the evaluation question is not what is this processor’s volume growth or how does the multiple compare to peers. It is a simpler one. If authentication were solved tomorrow and card processing became pure accounting, what would this company still sell, and would the merchant still pay for it?

Ask that of Toast and the answer is obvious. Ask it of Adyen and Stripe and the answer is most of the business. Ask it of Fiserv and the answer is why the stock is where it is.

Merchants pay for everything in payments. They have spent twenty years learning to pay less for the part that moves the money. The processors who understood that they were never really selling movement are the ones raising guidance this month.