Adyen N.V. (Euronext: ADYEN)

1. Executive snapshot

2. What the company does

Large merchants selling in many countries and channels face a fragmented payments stack: a different acquirer per region, separate gateways, separate fraud tools, separate terminal vendors and reconciliation files that do not match. Each extra provider costs conversion, because a legitimate payment declined by an issuer is lost revenue. Adyen's pitch is to replace that stack with one contract and one integration that accepts, processes, acquires and settles the payment end to end (FY2025 AR).

The unit of economics is one euro of processed volume. A shopper pays €100 at a merchant. Adyen routes the authorisation to the issuer through the card scheme, captures and clears the transaction, receives funds from the scheme and pays the merchant net of fees. The merchant fee combines interchange and scheme fees (passed through at cost), Adyen's mark-up, and a fixed processing fee per transaction. In H1 2026 that left Adyen €1,302.9m of net revenue on €803.8bn of volume — 16.2 bps, or €0.16 on the €100 — of which 49% became EBITDA (H1 2026 SL).

Where the revenue comes from

The revenue lines show the mechanics. H1 2026 non-interest revenue of €1,445.3m split into settlement fees €866.8m (the mark-up on pass-through costs), processing fees €321.6m (per-transaction), other services €194.4m (FX, risk, issuing and similar) and terminal sales €62.5m. Costs incurred from financial institutions (€94.7m) and cost of goods sold (€56.4m) are deducted to reach net revenue (H1 2026 SL, note 1). Since 1 January 2023 Adyen books interchange and scheme fees as agent, so IFRS revenue no longer includes them (H2 2023 SL).

Adyen reports three pillars. Digital (e-commerce-only merchants) earned €719.7m in H1 2026, up 13%, on €427.9bn of volume. Unified Commerce (merchants with online and in-store volume) earned €417.7m, up 25%, on €240.9bn. Platforms (marketplaces and vertical software firms embedding Adyen for their own sellers) earned €165.5m, up 37%, on €135.0bn (H1 2026 SL). Platforms monetises at about 12 bps against 17 bps in the other two (calc), but adds embedded products — accounts, cards, capital — for 293K active business customers.

The monetisation logic is volume-tiered. Merchants who send more volume move into lower price tiers, so the take rate falls as the base matures; management says it manages to net revenue, not take rate (H2 2023 SL). The business is therefore transactional in form but relationship-driven in substance: the typical merchant's share of wallet rises from under 20% in years 3–7 to over 40% after year 12 (H1 2026 SL).

The product scope is widening. The Talon.One (loyalty and promotions, €750m) and Orb (usage-based billing, $335m) acquisitions closed 1 July 2026, and Adyen now describes itself as a 'financial operating system for modern commerce' rather than a payments company (H1 2026 SL; June 2026 FAQ). These add subscription-style software revenue whose economics differ from per-transaction fees; their margins are not disclosed.

3. Industry, competitive position & moat

A card transaction's fee is split three ways. Issuers take interchange — capped in the EU at 0.2% for consumer debit and 0.3% for credit (Regulation 2015/751), and averaging 0.73% on US debit in 2023 (Federal Reserve Reg II data). Visa and Mastercard take scheme fees and earn the richest margins in the chain: 60% operating margin at Visa (FY Sep-2025) and 57.6% at Mastercard (FY2025). Acquirers and processors compete for the residual, which is regulated from both sides and set by negotiation with merchants who can compare providers.

The acquiring layer is split between slow-growing consolidators and fast-growing single-stack processors. By volume, Global Payments/Worldpay (~$3.7T pro forma, closed Jan 2026), JPMorgan and Fiserv lead the US. Their merchant segments still report high margins — Global Payments 49.4%, Nexi 53.1% EBITDA — but grow 1–3% organically, and the weakest have broken: Worldline wrote off €4.7bn of goodwill in 2025 and raised about €500m of emergency equity in 2026, while Fiserv's merchant margin fell to 34.5% and it reset guidance (company releases). Stripe ($1.9T volume, +34%), Adyen (€1,394bn, +8%, +21% excluding one customer) and Checkout.com ($300bn, +64%) are taking the growth (company disclosures, FY2025).

The relevant market is global and enterprise. Adyen's take rate of about 17 bps compares with roughly 34 bps for Global Payments/Worldpay's mixed book, 95 bps at Shift4 and about 1.56% at PayPal (calc from FY2025 filings). The gap reflects customer mix rather than inefficiency: enterprise merchants pay the lowest prices in the industry. PayPal's Braintree episode shows the consequence — its unbranded processing volume grew 30% in 2023 while transaction-margin dollars fell, and PayPal then repriced its largest merchants and accepted flat volume (PayPal Q4 2023 update; Q2 2025 call). Enterprise processing is only profitable at scale and on a low cost base.

Barriers that sound high bind less than expected. Acquiring licences and scheme memberships are obtainable — Stripe won a Georgia merchant-acquirer charter in 2025. What binds harder is the combination Adyen holds: a banking licence from De Nederlandsche Bank, US and UK branch licences, direct connections to local rails (principal membership of France's CB and STET, 2026; UAE Category II licence), and an S&P A- credit rating (H1 2026 SL). The licence lets Adyen hold merchant funds, access schemes directly and offer accounts, cards and capital without a partner bank. Rising fixed compliance and fraud-liability costs under PSD3/PSR favour scaled, well-capitalised providers (EU Council, Nov 2025).

What a well-funded competitor would find hardest to copy is the single codebase combined with a decade of enterprise relationships. Legacy acquirers built by acquisition run several platforms — Worldline decommissioned four legacy platforms in 2025 and attributes part of its merchant churn to migrations (Worldline FY2025). Adyen ships a new feature once, across all regions and channels, which is why Unified Commerce customers can run in-store and online on the same data. Its claimed conversion uplift (average +0.9 points via Uplift, H1 2026 SL) is a vendor figure; no independent measurement was found, though Stripe's 2025 letter reports enterprises shifting volume after acceptance gains, so the purchasing criterion is real.

Outside evidence supports the claim that Adyen is taking share from legacy acquirers and contradicts any claim that it is the fastest modern processor: Stripe and Checkout.com grew faster in 2025. Large merchants multi-home, routing volume between providers (FY2025 AR, concentration risk), which caps lock-in. Agentic commerce is being standardised by others — Stripe and OpenAI's Agentic Commerce Protocol, Visa Intelligent Commerce, Mastercard Agent Pay — and runs on network tokens that erode provider-specific card vaults.

The industry evidence suggests winners in acquiring are engineering-led, single-stack, rail-agnostic processors with enough scale to absorb fixed compliance cost, that earn above the base rate through performance and value-added services. Adyen fits that description more closely than any European peer; against Stripe the question is distribution breadth, not architecture.

4. Growth engine

Adyen's growth is overwhelmingly organic and volume-led. Net revenue grew 19% in H1 2026 (21% at constant currency) on volume growth of 24%. Roughly two thirds of the growth came from customers onboarded in 2024 or earlier, net of tiering; the rest from the 2025 cohort ramping, new wins and financial products (H1 2026 SL). Price is a headwind, not a lever: volume grew faster than net revenue because merchants moved into lower tiers.

Acquisitions become material only from H2 2026. Talon.One (about €60m ARR expected by end-2026, growing 30–40%) and Orb close on 1 July 2026. Management expects them to add about 1 percentage point to 2026 net revenue growth and cut 2026 EBITDA margin by about 1 point including one-off costs (June 2026 FAQ; H1 2026 SL). Guidance moved from 20–22% standalone to 21–23% including acquisitions, so standalone organic expectations are unchanged.

Headline volume can mislead in both directions. FY2025 volume grew only 8%, but 21% excluding a single large-volume customer first flagged in Q3 2024 (H2 2025 SL). Adyen has not named the customer. The Platforms pillar grew volume 28% in H2 2025, or 54% excluding eBay (H2 2025 SL), so a large platform customer is also shrinking its footprint.

Reported versus underlying growth

Growth drivers, ranked

5. Margin, cash & capital allocation

Adyen's cost base is mostly people and technology, not transaction costs. Pass-through interchange and scheme fees sit outside net revenue, and the remaining direct costs (financial-institution fees, terminal COGS) are small. H1 2026 operating expenses were €738.0m, of which employee costs €431.1m (FTE 5,020) and other operating costs €229.8m (H1 2026 SL). Because the platform is one codebase, incremental volume needs little incremental cost; margin is therefore set by how fast Adyen hires relative to revenue.

The history shows that choice directly. EBITDA margin rose to 63% in 2021 when volume grew 70% and hiring lagged, then fell to 46% in 2023 after a deliberate two-year hiring push took FTE from 2,180 to 4,196 (H2 2021 and H2 2023 SL). It recovered to 53% in 2025 as headcount growth slowed to under 10% (4,771 FTE). The >55% target for 2028 assumes revenue keeps outgrowing hiring (Investor Day 2025).

Cash conversion is high because the business needs little capital. Capex has run at about 5% of net revenue, mostly data centres; free cash flow (EBITDA less capex and leases) was €1,081.3m in 2025, 87% of EBITDA (H2 2025 SL). Capex rises to about 7% in 2026 as Adyen pulls 2027 data-centre spend forward to lock in hardware supply and prices (H1 2026 SL). Working capital is negative: merchant funds in transit (€8.1bn payables to merchants at 30 June 2026) sit on the balance sheet as cash until paid out.

That float is a second, borrowed profit stream. Interest on central- and commercial-bank balances is booked as finance income below EBITDA: €349.8m in 2024 and €267.6m in 2025, against €29.3m in 2022; in 2020 net finance income was a €50.7m charge, mostly other financial results (xBRL). Net finance income was 28% of pre-tax profit in 2024 and 20% in 2025; management said 2025 net income growth was 'hindered by the interest rate environment' (H2 2025 SL). Net income therefore tracks policy rates as well as the payments business.

Financial spine

Capital allocation

Capital allocation has been almost entirely internal. Adyen has paid no dividends and bought back no shares over the period (xBRL cash-flow statements). Cash went, in order, to people (opex rather than capex), data-centre capex, leases, and — for the first time at scale — acquisitions: Talon.One €750m and Orb $335m, of which €655.2m had been prepaid at 30 June 2026 (H1 2026 SL). Retained earnings grew from €873m (2020) to €4,581m (2025); weighted share count rose about 1% from 2024 to 2025 through share-based pay (FY2025 AR).

Adyen carries no borrowings; its only financing liabilities are leases of €409m (H1 2026 SL). Cash of €12.4bn at 30 June 2026 falls to €4.9bn excluding merchant balances, and to about €4.6bn net of the acquisitions (H1 2026 SL). The interim CFO said the truly available portion is 'a small fraction' after regulatory reserves, operating buffers and the credit rating (H1 2026 call). The behaviour reveals founders who treat the balance sheet as a banking asset and growth funding, not as distributable surplus; the 2026 deals signal a new willingness to buy capability the company once built itself.

6. Cyclicality, constraints & what to monitor

Adyen's revenue follows merchant sales in nominal euros, so a consumer downturn lowers volume growth, but the business has never reported a revenue decline. In 2020 lockdowns cut in-store volume, yet total volume still grew 27% as spend moved online (H2 2020 SL). The 2022–23 slowdown hit differently: North American digital customers 'shifted focus from growth to cost savings' and pushed on price and routing, while Adyen's hiring push compressed margin (H1 2023 SL). The downside mechanism is therefore enterprise merchants using a downturn to renegotiate tiers and move volume, more than falling transaction counts.

Exposure is diversified by region (EMEA 55%, North America 27%, APAC 11%, LatAm 7%) and by vertical, with 300 merchants accounting for 60% of growth (H1 2026 SL). Travel and live events carry merchant-failure risk: if a merchant fails before delivering goods, chargebacks fall on Adyen when payouts cannot absorb them (FY2025 AR, merchant potential liability). Currency matters: FX cut two points from H1 2026 growth.

Position in the cycle. Volume growth (+24% in H1 2026) is back near its 2023–24 range after the 2025 dip caused by one customer, but well below the 49–70% of 2021–22. EBITDA margin at 49–53% sits between the 2023 trough (46%) and the 2021 peak (63%). Take rate, at 16.2 bps, is near its low (15.5 bps in 2024) after a structural decline from 22.5 bps in 2020. Finance income is past its 2024 peak as ECB rates have fallen. Management expects margin to be flat in 2026 standalone and above 55% by 2028.

Durable versus borrowed

What to monitor, and where it is published

7. Risks, unknowns & questions for deeper work

• Multi-homing turns single decisions into large volume swings. Enterprise merchants deliberately split volume across providers (FY2025 AR). One unnamed customer cut FY2025 volume growth from 21% to 8%; eBay's shrinkage halved Platforms' reported volume growth in H2 2025. A few more such moves would slow growth while Adyen keeps hiring for the forecast, hitting growth and margin together.
• Tiered pricing makes growth depend on wallet share, not price. Every successful ramp lowers the take rate. If new products (Uplift, Talon.One, Orb, issuing) fail to offset tiering, net revenue growth converges on market volume growth minus price concessions — the Braintree pattern at PayPal.
• Stripe out-scales Adyen in the fastest-growing segments. Stripe grew volume 34% in 2025 and powers over 5 million businesses directly or through platforms (Stripe 2025 letter), and co-authored the leading agentic checkout protocol with OpenAI. If platform and AI-native merchants standardise on Stripe, Adyen's highest-growth pillar faces a stronger incumbent.
• Account-to-account rails can bypass card economics. Pix handled ~80bn transactions in 2025 and UPI 85% of India's digital payments; the EU's Instant Payments Regulation and Wero push the same model in Europe. Adyen supports local methods, but its take rate on A2A payments is not disclosed, so the effect on revenue per euro is unknown.
• Acquisition and leadership transition. Talon.One and Orb are Adyen's first material acquisitions, bringing different revenue models and cultures into a company whose edge rests on one in-house codebase. They coincide with a CFO transition to an interim CFO (H1 2026 call); the reason for the change is not in the sources.
• Banking-licence obligations. As a DNB-supervised credit institution, compliance failures in AML, sanctions or fraud could restrict onboarding in a region — the licences that protect the model also concentrate regulatory risk in one entity.

What the sources could not answer

• Identity of the single large-volume customer and its remaining volume (NOT FOUND).
• Net revenue share of the largest merchants; Adyen discloses only that concentration is 'within risk appetite' (NOT FOUND).
• Revenue model and margins of Talon.One and Orb beyond Talon.One's ~€60m ARR target (NOT FOUND).
• Adyen's take rate on account-to-account and local payment methods versus cards (NOT FOUND).
• Independent verification of Uplift conversion gains (NOT FOUND — vendor claims only).
• Stripe and Checkout.com net revenue, take rates and margins (not disclosed).
• Reason for the CFO transition and the permanent appointment (NOT FOUND).
• Deployable excess cash after regulatory and rating buffers (management: 'a small fraction' of €4.9bn; unquantified).
• Cause of the 2019 net revenue restatement from €496.7m to €534.3m (inferred: presentation change; not verified).

8. Investor takeaways