A Dependency Investigation (#10) – September, 2026
About This Series
The Magician’s Hands is a series of dependency investigations. Each report examines a single case in which a structural dependency, between a state and an infrastructure owner, a farmer and a seed company, a continent and an energy supplier, was created, normalised, leveraged, and converted into power. The cases span domains and decades. The grammar beneath them does not change.
The series takes its name from a simple observation: the most consequential things happening in the world are rarely the things that take centre stage. While we watch the visible hands, something else is being built in the structural layer underneath. These reports are an attempt to make that layer legible.
The founding article, “”, sets out the full grammar. Each investigation that follows applies it to a specific case.
Scene-setter
Every Dutch consumer, retailer, and public institution now depends on a small cluster of private rails and processors for the basic ability to transact. Behind the terminal that reads a debit card, behind the app that generates a QR code at a market stall, behind the government service that assumes a bank account and a smartphone, sits iDEAL, the domestic transfer rail; Visa and Mastercard, the card networks; ING, ABN AMRO, and Rabobank, the core banking processors; and Worldline, the processing layer that quietly clears transactions few consumers have ever heard of. The publicly visible story is one of convenience and modernisation: cash is inconvenient, card and app payments are fast, hygienic, and safe, and the Netherlands has simply moved further and faster along a path every economy is already on.
This investigation looks beneath that story, at the structural layer where two separate forces, one commercial and one governmental, have been building the same dependency in parallel for two decades without ever needing to coordinate. Private payment-rail consolidation pursued efficiency and reach. Public cashless-society policy pursued convenience, traceability, and sustainability. Neither set out to create a single point of failure. Together, reinforcing each other at every step, they did.
The thesis this report develops across five moves: the Netherlands did not lose cash to a single decision or a single actor. It lost cash to two independently rational architectures that happened to converge, each normalising the other’s advance, until outages at any one of a handful of institutions can now stop the country from being able to buy bread, and until the mechanisms proposed to correct the concentration, the digital euro chief among them, are themselves still years from functioning as anything more than an acknowledgment of the problem they are meant to solve.
Move 1: Creating the Dependency
This is a case of insertion, layered and repeated, rather than a single creation or a single capture. Cash was never seized; it was displaced, transaction by transaction, across roughly two decades, as new intermediaries inserted themselves into an existing flow of value between buyer and seller. The sequence is well known in outline: chip-and-signature displaced the chequebook and much cash retail in the early 2000s; chip-and-PIN displaced chip-and-signature; contactless tap-to-pay displaced PIN entry for small transactions; and from the mid-2010s onward, app-based payment, led by domestic products such as ABN AMRO’s Tikkie, displaced even the terminal itself for a growing share of peer-to-peer and small-merchant transactions. Each step had a genuine hook. Chip-and-PIN was safer than signature. Contactless was faster than PIN. Tikkie turned splitting a restaurant bill or paying a market stallholder into a single tap on a phone already in hand. Nobody adopted any of these because they were told to abandon cash; they adopted them because each one, individually, was a rational improvement on the step before it.
What no individual adopter could see was the cumulative structural consequence: every step narrowed the rails through which money could move to a smaller set of intermediaries. iDEAL, launched as a direct bank-to-bank transfer standard, became the default for Dutch e-commerce, commonly cited as handling somewhere in the range of 60 to 75 percent of online transactions*, because it was cheaper and more trusted than the alternatives it out-competed. Card issuance and acceptance consolidated around Visa and Mastercard as Maestro and V-Pay wound down. Processing consolidated around a small number of firms, Worldline foremost, a concentration whose systemic character was illustrated in November 2025 when a Worldline technical failure disrupted Bancontact payments across Belgium in a single incident, a cross-border demonstration of how one processor’s failure can propagate across an entire national card-payment system at once; a comparable pattern of shorter, more localised disruptions has recurred at individual Dutch banks and processors through late 2025 and 2026, without a single public register yet compiling their frequency*.
The public side inserted itself along the same seam. Municipal decisions and transit-operator policy, moving toward card-and-app-only acceptance through the 2010s, did not create a new payment rail; they simply withdrew the cash option from an existing one, closing off the fallback at precisely the moment the private rails were becoming systemically load-bearing. The key decision points were rarely dramatic. They were procurement choices, terminal upgrades, and app rollouts, each one individually defensible, each one collectively removing an exit.
Move 2: Normalising the Dependency
Narrative normalisation did most of the work. The dominant story, carried by banks, retailers, and government communications alike, cast cashless payment as safer, more hygienic, more efficient, and more modern, a story that accelerated further during the pandemic when contactless payment was explicitly promoted as a hygiene measure. Cash, by contrast, was narratively recast as costly to handle, vulnerable to theft, and administratively burdensome, a framing that made the case for its retirement sound like plain good sense rather than the closing of a systemic fallback. This framing sat uneasily against De Nederlandsche Bank’s own actual position, which is less hostile to cash than the popular narrative suggests: DNB has backed a 2024 parliamentary mandate requiring retailers to accept cash for transactions up to three thousand euros, explicitly to protect the roughly 2.6 million Dutch adults who struggle with digital payment, and has pushed back against retail-sector proposals to narrow that obligation. That DNB’s own defence of cash reads, in public discourse, as regulatory caution rather than as a counter-narrative with real force, is itself a measure of how thoroughly the convenience story had already taken hold by the time the mandate arrived.
Operational normalisation ran alongside it and proved more durable, because it required no story at all. Card and app payment did not need to be argued for once it became the default workflow embedded in point-of-sale hardware, e-commerce checkout flows, and salary and benefit disbursement. iDEAL’s near-ubiquity in Dutch online checkout, and the retrofitting of Mollie, Stripe, and comparable providers to route through it rather than around it, meant that even businesses seeking payment flexibility ended up reinforcing the same rail. Tikkie’s growth illustrates the mechanism precisely: on King’s Day 2026 the app processed a confirmed record of 756,316 payments, roughly one and a half times its normal daily volume, not because anyone campaigned for its adoption but because network effects made it more useful with each new user and more socially awkward to opt out of with each one that joined. The dependency stopped being a relationship and became infrastructure, simply how a bill gets split or a market stall gets paid, well before most users had reason to ask what sat underneath it.
Move 3: Leveraging the Dependency
Direct leverage in this case is faint; no dependency creator has needed to threaten withdrawal of service to extract concessions, because none has needed to. What the case shows instead is structural and option-space leverage operating with unusual clarity. Structural leverage appears in fee architecture: interchange and scheme fees, running roughly 1.0 to 3.0 percent plus a fixed component on debit cards and higher again on credit cards*, are set by the card networks and processors, not negotiated meaningfully by individual merchants, and by the industry’s own account represent a material restraint on payments-market competition, with the heaviest proportional burden falling on low-ticket, high-volume merchants, bakeries, cafés, and market stalls, where blended fees can consume a significant share of thin margins [unverified]. Retailers can see the fee; they cannot easily contest the rail that sets it.
Option-space leverage is the more consequential form here, because it has reshaped choices before any decision point is reached. A Dutch retailer today does not deliberate over whether to accept cash; the operational and cost logic of digital rails has already made that decision feel closed for the majority of businesses, even though the underlying arithmetic, as the independent-shop cash-preference data in this case shows, does not straightforwardly favour digital for every business type. A transit operator planning ticketing infrastructure does not weigh cash acceptance as a live option; the assumption of app-and-card-only has already been built into the request for proposals before any vendor is chosen. Consumers, in turn, no longer plan around the possibility of a payment outage; carrying no cash is the default, not a decision made after weighing the risk, which is precisely why a nationwide payment disruption in late October 2025, which delayed salary and benefit transfers across multiple banks, and a further ING outage in January 2026 that delayed Dienst Toeslagen payments, landed as sudden crises rather than as foreseeable consequences of concentrated infrastructure that DNB and AFM had, weeks before the first of these incidents, jointly warned was building*. The dependency creators did not need to apply pressure at any of these points. The narrowed option space applied it for them.
Move 4: Obscuring the Mechanism
The temporal displacement in this case is long and diffuse rather than sharp. The decisions that mattered, chip-and-PIN adoption, iDEAL’s rise to near-default status, municipal cash-acceptance withdrawal, were made incrementally across roughly two decades by different institutions acting independently, with no single moment at which the cumulative concentration was debated and approved. By the time the consequence became visible, as a pattern of national payment disruptions rather than as a single dramatic event, the officials and boards who had made the underlying procurement and policy choices were largely gone, replaced by successors managing an infrastructure they had inherited rather than built. No one was structurally positioned to be held accountable for a concentration that no single decision had created.
Narrative capture kept a specific question unaskable for most of this period: what happens when the rails fail, not individually but simultaneously. The convenience narrative was built around the ordinary case, the transaction that succeeds, and had little vocabulary for the tail case, the outage that takes multiple institutions down at once because they share processing infrastructure. It is only recently that this question has surfaced with institutional weight: on 20 October 2025, De Nederlandsche Bank and the Authority for the Financial Markets jointly warned, in an explicit public report, that the financial sector’s growing dependence on a small number of non-European cloud, software, and AI providers had become a systemic risk, one that geopolitical tensions could allow a hostile state actor to weaponise through sanctions or targeted disruption. Days later, the country experienced a nationwide payment disruption. The warning and the incident were not directly connected in the public record, but their proximity illustrates precisely how late in the process this question was finally being asked out loud. DNB had, in fact, already begun treating the tail case as real rather than theoretical: in May 2025 its Maatschappelijk Overleg Betalingsverkeer advised households to keep seventy euros in cash per adult and thirty euros per child at home, an amount Nibud calculated as sufficient to cover three days of essential spending, water, food, medicine, and transport, in the event of a payment-system disruption caused by geopolitical tension or a cyber incident. That such advice was necessary at all is itself the clearest admission available that the ordinary case and the tail case had drifted dangerously far apart.
Structural opacity compounds this. The dependency spans private card networks, private domestic transfer infrastructure, three separate core banking institutions, at least one shared processor, the cloud and software infrastructure those institutions increasingly run on, and public procurement decisions made independently by dozens of municipalities and transit authorities. No single regulator, institution, or journalist sits in a position to see the full picture at once; De Nederlandsche Bank and the AFM can assess financial-system and technology concentration, but cash-acceptance policy is set municipality by municipality and, since 2024, contested at the national parliamentary level, while payment-provider fee structures remain commercially confidential. The information required to see the dependency in full is technically distributed across many hands. The institutional architecture to assemble it into one legible picture largely does not exist, which is itself the clearest evidence that the country’s cashless transition has outpaced its own capacity to govern the risk it has created.
Move 5: Converting the Value
The most straightforward conversion in this case is financial: transaction volume that once generated no rent for any intermediary, cash changing hands directly between buyer and seller, now generates a fee at nearly every point of exchange, collected by card networks, processors, and payment-service providers whose margins depend on the volume of transactions that no longer touch cash. This is not a hidden extraction; it is priced and visible on every merchant statement. What is less visible is the second conversion, from financial dependency into policy default. Because digital rails are now assumed rather than chosen, public policy, from transit ticketing to benefit disbursement, is increasingly designed around their availability, embedding private infrastructure into public function without that infrastructure being subject to the accountability a public utility would carry.
A third conversion is normative: convenience has been converted into an implicit judgment about who counts as a fully participating consumer. The digital-exclusion consequences of this case are precise rather than diffuse: ING has confirmed that customers who bank exclusively through its physical Scanner device, rather than its app, will lose the ability to pay online once Wero replaces iDEAL, a group ING itself puts at under one percent of total iDEAL payment volume, but one drawn disproportionately from a wider population of roughly 2.5 million Dutch adults who struggle with digital banking. Rabobank and ABN AMRO report having kept their own hardware-token users compatible with Wero, which sharpens rather than dilutes the point: this exclusion was not a necessary feature of the transition but a specific institutional choice at one bank, made inside a migration that every other actor in the case has treated as self-evidently the right direction of travel. The long-term consequence still unfolding is a resilience deficit: a national payment system that functions well in the ordinary case and fails, simultaneously and nationally, in the tail case, with the population least able to absorb the failure, those without smartphones, digital literacy, or ready cash reserves, bearing a disproportionate share of the risk that the system’s designers, pursuing efficiency and convenience in good faith, did not intend to create and are only now beginning to name.
Analytical Notes
This case features a genuine collision of two dependency architectures rather than a single one. The first is private payment-rail and processor consolidation, driven by efficiency and network economics: iDEAL, the card networks, the core banking processors, and Worldline. The second is public cashless-society policy, driven by convenience, traceability, and sustainability framing: municipal and transit-operator decisions to withdraw cash acceptance. The two did not need to coordinate to reinforce one another; each made the other’s advance look more inevitable, because a citizen who could no longer pay cash on the tram had one more reason to treat the app as a necessity, and a retailer discouraged from handling cash by processing costs had one more reason to support card-only transit and municipal policy. The digital euro, proposed by the European Central Bank explicitly as a response to this concentration, is best understood as a third architecture attempting to insert itself as a counter-dependency: ECB board member Piero Cipollone has argued the case in stark terms, stating that two thirds of euro-area card transactions are governed by the business rules of non-European companies, and framing the digital euro as the construction of a fully European payment infrastructure to reduce that strategic vulnerability. As of this writing it remains a policy and pilot project rather than a functioning rail available to consumers, which means the collision currently has two live participants and one aspirant, arriving on a multi-year timeline while the dependency it is meant to check is already load-bearing today.
This case also contains a documented instance of resistance, which is analytically significant because most Magician’s Hands cases show dependency advancing uncontested. Independent shops, particularly small-ticket, high-volume merchants such as bakeries and market stalls, have been shown to prefer cash on straightforward cost grounds. Betaalvereniging Nederland’s own figures put the average cost to a merchant at 0.17 euros per debit card transaction against 0.61 euros for cash handling, a gap that has roughly tripled since 2017, when cash cost 0.29 euros against a stable 0.17 euros for card; roughly sixty percent of that cost is internal labour, and cash transactions carry a markedly higher labour share than contactless card payments. Yet this resistance is more contained than it first appears: DNB’s own monitoring found that 95.2 percent of Dutch retail locations still accepted cash in 2024, with pin-only acceptance rising only modestly, from 4.5 to 4.8 percent, concentrated disproportionately in large cities. The sharper conflict is not over whether shops accept cash but over how much they are legally required to accept: a 2024 parliamentary mandate requires cash acceptance up to three thousand euros per transaction, a threshold DNB defends as protection for the roughly 2.6 million Dutch adults who struggle with digital payment, and which retail organisations are fighting to cut to three hundred euros, citing hospitality-sector cash-handling costs of roughly 1.36 euros per transaction. This resistance has succeeded at the level of individual merchant cost calculation, where cash remains rational for many small businesses, and has partially succeeded at the level of law, where the acceptance mandate now exists but is actively contested. It has not succeeded at the level of consumer behaviour or transit and municipal infrastructure, where cash-acceptance withdrawal continues on its own logic regardless of merchant preference. The point of failure for resistance in this case sits precisely at that boundary, between commercial and now partly legal protection for cash on the merchant side, and public infrastructure policy on the consumer-facing side, which does not route through the same mechanisms at all.
A further structural note, offered as cross-reference rather than developed in full here: the same concentration logic this case describes at the payment-rail level recurs one layer down, in the cloud, software, and AI infrastructure that increasingly underpins core banking systems themselves. De Nederlandsche Bank and the Authority for the Financial Markets jointly warned, on 20 October 2025, that the financial sector’s reliance on a small number of non-European cloud, software, and AI-model providers now constitutes a systemic risk that geopolitical tensions could allow a hostile actor to exploit, a dependency already examined in detail in this series’ Cloud and Data Infrastructure investigation. The payment case and the cloud case are not the same dependency, but a failure in the latter is now a plausible trigger for failure in the former, a compounding relationship worth naming even where it is not this report’s central argument.
Closing
What this case adds to the grammar established elsewhere in this series is a demonstration that dependency does not require a single dependency creator to be dangerous. It can be built by two architectures pursuing entirely separate goals, neither of which is malicious, neither of which is even aware of reinforcing the other, converging on the same narrowed infrastructure because efficiency and convenience happen to point in the same direction. The Netherlands did not choose to make itself unable to function without a handful of private rails during an outage. It made that choice one terminal upgrade, one municipal procurement decision, one convenience feature at a time, and each of those choices, examined individually, was defensible.
The proposed correction, a public digital rail explicitly designed to dilute private concentration, is itself instructive: it is being built by the same logic of centralisation it is meant to counteract, on a timeline measured in years, while the dependency it addresses is already load-bearing today. Whether the digital euro arrives in time to function as a genuine counter-dependency, or arrives as a third concentrated rail alongside the two it was meant to check, is not yet decided.
The question worth carrying forward is not whether the Netherlands should return to cash, which is neither likely nor, for most transactions, obviously desirable. It is whether a society can deliberately preserve a slower, less efficient, less convenient fallback specifically because its inefficiency is what makes it resilient, and whether any institution is currently positioned to make that argument stick against the accumulated weight of two decades of rational, individually sound decisions all pointing the other way.
Comments are closed