Adyen is the invisible plumbing that takes the money when you tap "Pay" on a website or in a shop, checks with your bank that the money is really there, moves it to the shop's bank account, and keeps a tiny sliver of each payment as its fee. In the jargon, it sits in merchant acquiring (the business of signing up merchants and collecting their payments) and payment processing (moving the messages and money between merchant, networks, and banks).

What makes it "full-stack" is that it owns every layer itself. It is not a card network like Visa or Mastercard (the toll road in the middle that routes payments between banks and sets the fee rules) - Adyen rides that road and pays tolls to it. It is not an issuing bank (the shopper's bank that hands out the card) - Adyen sits on the merchant's side. And it is not just a payment gateway (the checkout front door that captures card details); Adyen owns the gateway, the acquirer role (the licensed institution that collects the money and pays it into the merchant's bank account), and the processing underneath.

The whole business, all EUR 1,394 billion of processed volume in 2025, is nothing more than one authorize-and-settle event - a single shopper pressing "Pay" - repeated billions of times, with a few cents skimmed off each time it clears. Everything upward (volume, take rate, margin, the market's valuation) is just this one atom aggregated and then priced.

Follow a single payment. A shopper presses "Pay" -> the gateway captures the card details (using tokenization, swapping the real card number for a non-reversible stand-in so the merchant never stores sensitive data) -> Adyen, as its own acquirer, sends an authorization request through the card network's scheme switch (the routing engine that maps the card number to the right bank) -> the issuing bank approves or declines within a second or two based on funds and fraud risk.

The share of attempts that get approved is the authorization rate, and this is where Adyen's footprint pays off: processing inside the shopper's own country (local acquiring) instead of cross-border lifts approvals materially. Higher approvals -> more of the merchant's attempted sales actually clear -> more processed volume Adyen earns on. The approved transaction is later captured, batched, cleared and settled - the network nets the day's transactions (cancelling what each bank owes against what it is owed so only the leftover difference moves) and the merchant is paid out one or two business days later (T+1 / T+2 funding), minus fees.

So the chain runs: one tap -> authorization -> approval (gated by the auth rate) -> clearing and settlement -> money in the merchant's account -> a few kept cents of revenue. The single most important capability inside that chain is whatever decides whether the payment can happen at all, which is the fine print.

At the moment of payment the world splits in two. Either the shopper uses a global card (which routes through the Visa/Mastercard duopoly and pays its tolls) or a local payment method (LPM) like Brazil's Pix or India's UPI - an account-to-account (A2A) payment that moves money directly bank-to-bank with no card network in the middle, often over a central-bank-run instant rail. Which world is even available was decided long before this moment, by whether Adyen holds the right acquiring licence (a regulator's permission to collect payments for merchants in that country) and scheme membership (formal admission to that specific rail) there.

This is the heart of the thesis. The raw "move-the-transaction" rail is a commodity: many vendors supply it at near-identical quality, so any price cut drops straight to profit and competitors can match it. Adyen's durable advantage therefore cannot be the rail. It is the per-country regulatory navigation plus acquiring licence plus banking licence (DNB) (a full bank authorization from the Dutch central bank that lets Adyen hold and settle funds itself) plus scheme membership and a local legal entity on the ground.

That stack is slow, capital-intensive, and repeated country by country (an EU e-money licence alone needs EUR 350,000 minimum capital plus full KYC and AML / CTF programs and lead times of a year or more). In the EU, passporting lets one Dutch licence cover the whole European Economic Area, but most emerging markets force the entire process to be repeated - which is exactly why it is hard for a rival to copy quickly. Each new market multiplies addressable volume without needing the take rate to rise. This is the chokepoint that gates scaling units across geographies (the moat), distinct from the card-network chokepoint that gates every single transaction.

On every cleared transaction Adyen charges a gross fee, but it keeps almost none of it. Most is pass-through costs: the interchange fee handed to the shopper's issuing bank (often 1.5 to 3 percent on US credit cards) and the scheme fees handed to Visa and Mastercard (roughly 0.11 to 0.14 percent). Adyen collects these and immediately hands them on - they are the cost that will not shrink no matter how big Adyen gets, because the card network sets them.

What is left after stripping pass-through is net revenue, the true top line. It is made of a small fixed processing fee per transaction (roughly 0.10 to 0.15 euros) plus a settlement fee that embeds Adyen's own markup (the markup floor is around 0.60 percent). Adyen prices on transparent interchange++ pricing, which shows the interchange, the scheme fee, and its own markup as three separate line items, rather than an opaque blended pricing rate.

Divide net revenue by processed volume and you get the net take rate: roughly 16 to 17 basis points (a basis point is one hundredth of a percent), meaning Adyen keeps about 17 euro-cents of profit-eligible revenue per 1,000 euros it moves. Read forward, the identity is simply: processed volume x net take rate = net revenue. This is the single most important number - the cents kept per unit, not the headline volume.

The net take rate is under a slow, multi-year squeeze called take-rate compression. The cause is the enterprise merchant mix: because Adyen uses tiered / volume-based pricing (bigger merchants get cheaper per-transaction rates), the more its volume comes from giant enterprises, the lower the blended cents-per-euro drifts - even as total euros rise. So net revenue can grow far slower than processed volume. The rate ticked from about 17 bps in 2022 down toward roughly 16.2 bps by H2 2024 before a mix-driven uptick to about 17.1 bps in H2 2025.

The same concentration is also a demand-side cliff. Giant merchants multi-home - deliberately connecting to several processors at once via a payment orchestration layer and splitting volume between them - so they can run a competitive RFP (a formal bid process) and re-weight volume with a dial, not a rip-and-replace. That gives them pricing power and means a single client re-routing its volume hits the business all at once. Adyen's switching cost (the cost and risk of leaving) is a real second moat layer, but it is softened because the largest merchants multi-home anyway.

The one structural escape from the compression-and-toll problem is to route volume off cards entirely onto A2A local methods, which carry near-zero interchange. The gross fee there is smaller, but Adyen's markup is a larger share of it, and Adyen is often the only provider that can light the method up at all - turning scarcity into pricing power.

Net revenue then meets the cost base. Because Adyen runs essentially one single platform / single code base worldwide, each extra transaction costs almost nothing to serve - this is operating leverage, where fixed costs spread over rising volume so most new revenue drops straight to profit. As a result most incremental net revenue falls through to EBITDA (earnings before interest, taxes, depreciation and amortization - a rough measure of operating cash profit before financing and non-cash charges). The EBITDA margin is about 53 percent and rising, with a target above 55 percent by 2028. Subtracting interest, tax and depreciation from EBITDA gives net income, the bottom-line profit the equity market ultimately owns.

The public market takes that stream of EBITDA and net income and capitalises it - turning a recurring annual profit into a single lump-sum value by applying a multiple - at a market capitalization of roughly EUR 28 to 29 billion in mid-2026. It values the operating profit on an EV/EBITDA (enterprise value divided by operating cash profit: how many years of profit the market pays for the whole company) of roughly 12 to 15 times, and the bottom-line profit on a P/E ratio (price-to-earnings: euros paid per euro of annual profit) in the low-to-mid 20s.

That is a growth-stock multiple: it bakes in expected continued ~20 percent net-revenue growth and expanding margins, partly discounted by the take-rate-compression overhang. The multiple expands when investors believe the volume runway (new geographies, new local methods) is long and margins keep rising, and compresses fast when they fear compression or commoditization of the rail - which is why the shares re-rated down sharply over the prior year (from a 52-week high near EUR 1,750 to around EUR 920). The multiple times the profit is the end-state: what the market pays today for the right to those future units of Adyen profit.