October 18, 2026 · BriMindInvest Research Team · 22 min read
Every time a card is tapped, a phone is waved, or a checkout button is clicked, a fraction of the transaction value flows to a small group of companies that neither lend money nor take credit risk. They simply sit in the middle of money movement and collect a toll — on trillions of dollars a year. The payment networks are the closest thing public markets offer to a perpetual royalty on GDP. This guide covers how the payment stack works, the one number that determines profitability, and eight companies across three layers of the stack.
When you tap your card at a coffee shop, the transaction looks instantaneous. Underneath, it passes through four distinct layers, each operated by a different kind of company, each taking a slice. Understanding this stack is the prerequisite for understanding payment stocks, because a company's position in the stack determines its economics entirely.
The critical insight is that the network layer and the acquiring layer are fundamentally different businesses. The network is an asset-light, near-monopoly standard-setter with 60%+ operating margins. The acquirer is a competitive, operationally complex intermediary with 20–30% margins. They are both called “payments companies” but they share almost nothing structurally.
Visa and Mastercard do not lend. They do not hold consumer deposits. They do not fund rewards. They do not absorb credit losses when a cardholder defaults. Every risk in the system — credit risk, fraud risk, funding cost — is borne by someone else. The networks set the rules, authorise in real time, guarantee settlement, and collect a toll. That toll scales with the nominal value of commerce, which means it grows with both real GDP and inflation — a natural hedge that few other business models offer.
The barriers to entry are not regulatory (anyone can theoretically build a payment network) but structural. A payment network is only valuable because both sides — issuers and acquirers — have adopted it. That two-sided network effect means a new entrant must simultaneously convince thousands of banks to issue cards on its network and millions of merchants to accept them. No one has successfully done this at scale in decades. The last attempt, Discover, succeeded only by also becoming its own issuer.
This is a strong claim. Here is the case for it.
Banks lend money and take credit risk. Insurers take underwriting risk. Asset managers take AUM-outflow risk. Payments companies — at least the network layer — take none of these risks. They are asset-light, capital-light, and their revenue scales directly with the volume of money moving through the economy.
The combination of these traits produces financial profiles that look more like software companies than financial companies: 65%+ operating margins for the networks, 30%+ free cash flow margins, high returns on equity with minimal leverage. The reason payments stocks trade at technology multiples is that the economics are technology economics — high fixed costs, near-zero marginal costs, and network effects that entrench the incumbent.
The acquirer and processor layer does not share all of these advantages. It competes on price, carries operational complexity, and must spend continuously on technology and sales. It is still a good business — recurring, growing, and benefiting from the same secular tailwinds — but it is a meaningfully different kind of good business, and the market prices it accordingly.
Every payments company has one metric that determines whether the business is getting better or worse: the take rate. It is the percentage of total payment volume that the company retains as revenue. It is the simplest possible measure of pricing power.
In this example, the merchant pays $2.34 in total fees on a $100 purchase. That $2.34 splits roughly: $1.80 to the issuing bank (interchange), $0.14 to the network (Visa or Mastercard), and $0.40 to the acquirer/processor. The network's take rate on this transaction is 0.14%. The acquirer's take rate is 0.40%.
A payments company can grow volume 15% and see revenue grow 8% if its take rate is declining — because the marginal dollar of volume is lower-yield than the existing base. This happens structurally when:
A declining take rate is not always bad. If a company is winning large enterprise accounts at a lower rate but adding enough volume to grow total revenue, the business is getting bigger and more entrenched even as the unit economics compress. Adyen's take rate has declined for years, and its revenue has compounded at 25%+ because volume growth overwhelmed the rate compression. The question is whether the declining rate reflects competitive pressure (bad) or a deliberate shift toward higher-volume, stickier accounts (often good).
We describe what each business actually is, where its structural advantage lies, and what the live debate about it is. Valuations and financial metrics change quarterly and a stale number is worse than no number — each name links to its analysis page for current figures.
These two companies sit at the top of the payment stack. They do not lend, do not process, and do not touch the money. They set the rules and collect a toll on every transaction that touches their rails. Together they handle over 80% of global card network volume outside China.
The case: Does not lend, does not take credit risk — pure toll booth on $15T+ of annual payment volume. Cross-border transactions carry 2-3x the revenue yield of domestic ones, meaning international travel and e-commerce growth disproportionately benefit the top line.
What to watch: DOJ antitrust suit on debit routing; regulatory risk is the only structural threat. If merchants win the right to route debit transactions away from Visa's network, the domestic debit franchise erodes — though credit and cross-border are unaffected.
The case: Same toll-booth model as Visa with a slightly higher international revenue mix, which gives it more leverage to cross-border recovery. Value-added services — fraud analytics, consulting, cyber and intelligence — now represent roughly 35% of revenue and grow faster than the core network.
What to watch: Same regulatory overhang as Visa. Additionally, stablecoin settlement and real-time payments infrastructure (FedNow in the US, UPI-like systems globally) could theoretically bypass the card rails entirely for certain transaction types.
These companies operate outside or alongside the traditional card stack, building direct relationships with both consumers and merchants. They represent the bet that payment economics can be captured by owning the customer relationship rather than the rails.
The case: 430M+ active accounts and Venmo's embedded position with younger demographics provide distribution that is expensive to replicate. Branded checkout — where the consumer chooses PayPal at the point of sale — carries a meaningfully higher take rate than unbranded processing.
What to watch: Take-rate compression from unbranded processing is the central tension. The shift toward lower-margin Braintree gateway processing dilutes blended margins even as total payment volume grows. Active account growth has stalled; revenue per account is the metric that matters now.
The case: Two-sided ecosystem — merchants on Square, consumers on Cash App — with the potential to close the loop. Cash App's lending and banking products (direct deposit, Cash App Borrow) are high-margin and growing. The seller ecosystem benefits from software lock-in through payroll, invoicing and appointments layered on top of payments.
What to watch: Bitcoin revenue is high-volume, near-zero-margin noise in the P&L. It inflates reported revenue dramatically while contributing almost nothing to gross profit. Strip it out before valuing the business. Management's Bitcoin conviction introduces capital allocation risk that has nothing to do with payments.
These companies sit between the merchant and the network. They do the operational work of connecting businesses to the payment system — providing terminals, processing transactions, settling funds, and increasingly embedding software into the payment relationship to create stickiness.
The case: Clover POS is the fastest-growing SMB payment terminal in the US, with over 900K merchant locations. Core banking technology serves thousands of banks and credit unions, creating switching costs measured in years and tens of millions of migration dollars.
What to watch: Bank tech is a slow-growth, sticky revenue stream — it defends the installed base but does not drive multiple expansion. The market rewards Clover growth; the question is whether Clover can sustain its pace against Toast, Square, and other vertical-specific competitors.
The case: Integrated payment-plus-software model in vertical markets — restaurants, healthcare, retail — where the software creates stickiness and the payments generate volume. The issuer solutions business provides card management to banks, adding diversification.
What to watch: Leverage is higher than peers from the TSYS merger. Debt service constrains capital return and acquisition flexibility. The market gives GPN a lower multiple than Fiserv largely because of balance sheet quality, not business quality.
The case: Built a single global processing stack from scratch rather than stitching together acquisitions, which means one codebase, one data model, and a genuine technology advantage in multi-market settlement. Enterprise clients include Uber, Spotify, Microsoft and eBay.
What to watch: Amsterdam-listed, not US — so there is currency risk and a different regulatory environment. Take rate is structurally declining as volume scales and the client mix shifts toward very large enterprise accounts that negotiate lower pricing.
The case: Vertical integration in restaurants — POS, payroll, online ordering, loyalty, payments, and kitchen display all in one platform. When a restaurant runs its entire operation on Toast, switching costs are enormous. The SaaS layer generates recurring revenue that is independent of transaction volume.
What to watch: The restaurant industry has high churn: roughly 60% of restaurants fail within five years, and closures directly remove locations from Toast's installed base. Net revenue retention depends on upselling additional modules to surviving restaurants, which is working but must continue indefinitely.
These eight are included because they are the most liquid and widely analysed names across the three layers of the payment stack — which makes them a sensible place to learn the sector. Inclusion here is not a view that any of them is attractively priced today. Run each through the framework above and through current numbers before acting.
The bull case for payments stocks is widely understood. The bear case is less often articulated with precision. There are three structural threats worth taking seriously, even if none is imminent.
The US launched FedNow in July 2023 — a real-time, bank-to-bank payment rail that settles in seconds and charges a fraction of a cent per transaction. In markets where similar systems are mature (UPI in India, Pix in Brazil), they have displaced card payments for a meaningful share of person-to-person and bill-pay volume. If merchants adopt real-time pay-by-bank at checkout, the card networks lose the transaction entirely. The counter-argument: real-time rails lack the chargeback protections, rewards infrastructure, and consumer habit that sustain card dominance, and adoption at the point of sale has been slow even in markets with mature infrastructure.
Dollar-denominated stablecoins like USDC and USDT already move hundreds of billions per month on-chain. For cross-border B2B payments — where the card networks charge the highest yields — stablecoin settlement is faster, cheaper, and 24/7. The threat is most acute in the cross-border corridor, which is disproportionately profitable for Visa and Mastercard. But the networks are not standing still: Visa has integrated USDC settlement on Solana, and both networks are positioning to be the infrastructure layer for stablecoin movement rather than its victim.
The DOJ's antitrust suit against Visa in debit, the EU's interchange fee caps, and periodic congressional scrutiny of the Durbin Amendment all target the same economic structure: a two-sided network with pricing power that merchants cannot practically refuse. Regulation that reduces interchange or forces multi-rail routing compresses network economics directly. The risk is not existential — the networks survived Durbin and the EU caps — but each incremental restriction shaves basis points from the toll-booth model.
The honest assessment is that none of these threats has yet produced a measurable impact on Visa or Mastercard's financials. Cross-border revenue growth has accelerated, not decelerated. Domestic card volume continues to take share from cash. The threats are real in theory and invisible in the numbers so far — which is exactly the profile that makes them easy to dismiss right up until the inflection point.
For the processor and acquirer layer, the threat calculus is different. These companies compete in a genuinely competitive market where a new entrant with better technology (Stripe, Adyen) has demonstrably taken share from incumbents. The structural moat at the acquirer level is much narrower than at the network level.
The payments sector is unusual in that it spans a wide valuation range despite every company benefiting from the same secular tailwind. The spread tells you what the market is pricing in — and what it is not.
A common objection to Visa at 30× earnings is that the multiple is too high. This misses the point of what you are buying. Visa's earnings growth is driven by three compounding forces that require no capital investment, no new product development, and no management genius:
Together, these produce low-teens earnings growth with minimal reinvestment, funded by a business that generates more cash than it can deploy — so excess cash is returned through buybacks and dividends. A 30× multiple on a 12–14% compounder with no balance sheet risk and a 30-year runway is expensive only if you believe the runway is shorter than the market thinks.
Block trades at roughly 20× earnings despite operating in the same sector. The discount reflects three real concerns: Bitcoin revenue obfuscation makes the P&L hard to read, management's capital allocation toward cryptocurrency introduces non-payments risk, and the company has not yet demonstrated that it can close the loop between Square (merchants) and Cash App (consumers) in a way that produces a durable margin advantage. The market is saying: good business, uncertain management, prove it.
Block reports Bitcoin revenue at the gross level, which inflates total revenue by 50% or more. Bitcoin gross profit margin is roughly 2%. If you use a revenue multiple, Block looks cheap. If you use a gross-profit multiple — which is the correct lens — it trades much closer to peers. Every sell-side model worth reading uses gross profit, not revenue, as the starting point.
A $100 cross-border credit card transaction generates roughly 10x the revenue of a $100 domestic debit transaction for the network. A $100 branded PayPal checkout generates 3-5x the revenue of the same transaction processed through Braintree unbranded. Volume growth without mix context is meaningless — always ask what kind of volume is growing.
Visa and Mastercard do not touch the money. They authorise transactions, set rules, and collect a toll. Fiserv, GPN, and Adyen actually move the money between banks. The network is a near-monopoly toll booth with 60%+ operating margins. The processor is a competitive, operationally intensive business with 20-30% margins. Putting them in the same bucket produces nonsense comparisons.
Nearly every fintech payment product — Apple Pay, Google Pay, Cash App, Venmo, Buy Now Pay Later — rides on existing card rails. When you tap your phone at a terminal, a Visa or Mastercard transaction fires underneath. The networks are not being disrupted by digital wallets; they are being distributed by them. The disruption risk is to acquirers and issuers, not to the networks themselves.
Revenue growth in payments is meaningless without margin context. A company can grow revenue 20% by processing more low-margin unbranded transactions while gross profit grows 8%. The market figures this out within two quarters. Track gross profit growth, take-rate trends, and operating leverage — not top-line volume.
The payments sector lends itself naturally to a barbell approach: a core position in the structural winners, with smaller positions in higher-growth, higher-risk names where the optionality justifies a different risk profile.
Visa and Mastercard are as close to “buy and forget” as public equities get. They compound at GDP-plus with no credit risk, no balance sheet risk, and a multi-decade runway from cash-to-card conversion globally. The position size decision is primarily about valuation entry point — at 35× earnings, you are paying for perfection and the margin of safety is thin; at 25× (which happens during broad market selloffs), the risk/reward is genuinely attractive for a 10-year hold.
The rest of the sector requires more active judgment because each company has a distinct thesis with distinct risks.
If your target allocation to payments is 10% of a portfolio, a reasonable split might be 6–7% in networks (your core compounders), and 3–4% spread across two or three of the higher-growth names where you have highest conviction. The networks anchor the position; the satellites provide optionality without jeopardising the whole allocation if one thesis fails.
If income is part of the objective, the networks are durable dividend growers — Visa has raised its dividend every year since its IPO, and the payout ratio remains below 25%, leaving substantial room for continued increases. Our dividend growth guide covers how to assess payout durability more generally.
Every company in this guide has a full analysis page with current valuation, profitability, balance sheet and historical financials — so you can apply the take-rate and margin framework above to live numbers instead of stale ones.
Payments is the one corner of financial services where the best companies take no credit risk, carry no meaningful balance sheet, and earn a recurring toll on the nominal value of global commerce. That toll scales with GDP, with inflation, and with the multi-decade shift from cash to electronic payments. It is a structurally exceptional business model.
But not all payments stocks are the same. The network layer (Visa, Mastercard) is a near-monopoly toll booth with 60%+ margins and a 30-year runway. The processor layer (Fiserv, GPN, Adyen) is a competitive, operationally complex intermediary that must continuously earn its position. And the digital layer (PayPal, Block, Toast) is a collection of platform bets with wide outcome distributions.
Track take rate, not just volume. Use gross profit, not revenue, for companies with low-margin pass-through lines. Compare networks to networks and acquirers to acquirers — never across layers. And remember that the biggest risk to the sector is not technological disruption but regulatory action, because the networks' toll-booth economics are ultimately a political question about how much rent a two-sided monopoly should be allowed to extract.
The most common mistake is buying “a payments stock” without understanding which layer of the stack you are buying. The network, the processor, and the digital wallet are three different businesses with three different risk profiles and three different appropriate valuations.
This article is educational and is not investment advice. Company descriptions reflect business models and structural characteristics as of October 2026; financial metrics change quarterly and should be checked on each company's analysis page before making any decision. The authors may hold positions in one or more of the securities discussed. Past performance does not predict future results.
Get AI prediction signals, unlimited stock comparisons, portfolio analytics, and personalized watchlists — free for 14 days, no credit card required.
14-day free trial · No credit card required · Cancel anytime
Not ready to sign up? Get one free email a week instead.