Money began as a ledger. It is becoming one again.

From about 8000 BC, farming settlements across the Near East, the region we would now draw as Iraq, Syria and their neighbours, counted their goods with small clay tokens, using a cone for a measure of barley, an ovoid for a jar of oil and a cylinder for an animal ⁸. Around 3500 BC those tokens started being sealed inside clay envelopes, with their shapes pressed into the wet surface so that the contents could be read without breaking the seal. Someone eventually noticed that the impressions made the tokens redundant. By roughly 3350 BC at Uruk the impressions had become proto-cuneiform tablets, the oldest known writing on Earth, and nearly all of it accounting.

That last claim is a large one, so it should not rest on one authority. Denise Schmandt-Besserat, who spent a career cataloguing the tokens, traced writing’s descent from them directly ⁸. The independent and more conservative study of the Uruk tablets themselves, Archaic Bookkeeping by Nissen, Damerow and Englund, reaches the same destination by a different road. The earliest written tablets are overwhelmingly administrative records of rations, grain, herds and labour, with literature arriving only centuries later ⁹. And the accounting historian Richard Mattessich read the token-and-envelope system as a genuine accounting system in its own right, functioning before writing and before abstract counting ¹⁰. The honest scope is Mesopotamia, since writing arose separately in China and Mesoamerica for other purposes, but for this essay that scope is the whole point.

In Mesopotamia, writing was invented to keep accounts, which makes the oldest known text on Earth a payments artefact.

What those tablets describe is a functioning payments system with no cash in it. Barley and weighed silver served as units of account, with the shekel running to about 8.3 grams. Temples and palaces took deposits, made loans and moved value between accounts by amending the record. Hammurabi’s code, around 1754 BC, regulated all of it, capping interest at 20% on silver and a third on grain, and making deposits enforceable only with a written contract and witnesses.

Payments regulation is thus not a modern affliction (or benediction depending who you ask) at all, and it is older than the alphabet. Here is the riverbank at its very first appearance, and notice that it cuts both ways, constraining the lender whilst making the deposit worth trusting at all.

Pay close attention to who built this system, because the answer is nobody in particular. The tokens belonged to farmers, the envelopes to temple clerks, and the tablets to scribes refining a convenience into an institution over forty centuries. The first payments system was co-created by its users and its record-keepers to solve a common set of problems, a habit this essay never loses.

Two things from this era never leave this essay. The first is the unit of account, meaning the habit of pricing everything in a common measure, and the second is the institutionally held ledger as the place where money actually is. Everything that follows is a negotiation over how far from that ledger a payment can safely and accurately travel.

Somewhere around 650 to 600 BC, in Lydia, a kingdom in what is now western Turkey, someone struck lumps of electrum to a fixed weight and punched a mark into them. Electrum is a naturally occurring alloy of gold and silver, washed down in the beds of Lydian rivers like the Pactolus, and it was the metal from which the world’s first coins were struck ¹¹. The earliest hoard sits under the temple of Artemis at Ephesus, and slightly later coins carry the name WALWET, attributed to King Alyattes. The clever part was not the metallurgy. Natural electrum varies wildly in gold content, and the early official coins actually ran leaner than river electrum, so what the stamp really said was that the issuer of the mark stands behind this. This is an early example of how valuation had moved from the metal to the mark, which is to say from commodity to institution.

The coin solved a genuinely new problem, which was paying someone with whom you shared no institution. A tablet in Uruk works between parties who already share a temple or a palace, whilst a coin clears and settles instantly and anonymously between parties who may share nothing in common other than the sovereign whose face it bears. It is the first payments-bearer instrument, and it is also the first time in this history that settlement is instant. The trade-off is that the ledger disappears, because a coin remembers nothing.

That trade-off turns out to be the deepest idea in this essay, and it took an economist until 1996 to state it precisely. In a paper titled, perfectly, “Money is Memory”, Narayana Kocherlakota proved that anything money can do, a complete record of past transactions could also do, concluding that “from a technological point of view, money is equivalent to a primitive form of memory” ¹². Money, in other words, is society’s substitute for a ledger it cannot cheaply keep. The coin in your hand is a portable stand-in for a missing record. Read that way, the whole history in this essay becomes one sentence.

When keeping the record is expensive, money hardens into objects, and when keeping the record becomes cheap, money dissolves back into the ledger it always was.

That is precisely why historically serious money movement and transactions never fully adopted the coin. Harken back to ancient Rome. Coin ran in the street, whilst its bankers, the argentarii, moved large sums between accounts by perscriptio, or written book entry, and their ledgers were admissible in court. When Cicero needed to fund his son’s studies in Athens, he did not ship a chest of denarii across the Mediterranean, but arranged a permutatio, a paper transfer between bankers.

You can see thus that account-to-account payments were already the premium product, twenty-one centuries before anyone called it that.

Between roughly 800 and 1400, four civilisations that mostly were not talking to each other converged on the same design. For starters, Tang China’s feiqian, or “flying cash”, was in merchant use by 804 and officially sanctioned by 812, and it let a tea trader deposit coin in the capital and redeem a matched certificate in the provinces.

The Islamic world’s suftaja and hawala moved value across the Abbasid empire on trust between brokers, settled by periodic netting rather than shipment. England’s Exchequer split hazel tally sticks into stock and foil, the grain of the wood serving as an unforgeable checksum, and the halves circulated as transferable instruments. And finally, the Italian city-states perfected the bill of exchange, which bundled remittance, foreign exchange and credit into one piece of paper.

Notice the riverbank bending the river again here, because the Church’s usury doctrine banned lending at interest outright, so the bankers tucked the interest inside the exchange rate where the lawyers could not reach it. The economic historian Raymond de Roover showed that the bill of exchange took its four-century shape precisely to stay on the right side of that prohibition ¹³. Regulation did not stop the credit, but it did decide the container the credit travelled in.

That four civilisations reached the same design without copying each other is itself the tell. Biologists call it convergent evolution, where separate lineages under the same pressure arrive independently at the same solution, in the same way the eye evolved more than once. The pressure here was identical everywhere, manifesting as the cost and the danger of moving metal.

The solution devised was the same too. A written claim that travels in the metal’s place. It is the first time in payments history that one answer appeared in four rooms at once, and it will not be the last, because a thousand years later India and Brazil will build instant account-to-account rails independently and arrive at almost the same machine.

China went furthest. In 1024 the Song state nationalised the private jiaozi notes of Chengdu’s merchant houses and issued the world’s first government paper money, driven, characteristically, by an infrastructure problem. Sichuan ran on iron coin, and nobody wants to settle a large invoice in iron.

Each of these four instruments is a claim on a distant ledger, carried by hand.

The payment message and the settlement have come apart, so the note moves now and the money moves later, if at all, and that separation of message from money becomes the defining architecture of payments for the next thousand years. It is also the separation that every era after this one works to close.

The tally sticks matter for a second reason. When Parliament finally burned six centuries of them in 1834, the furnaces overheated and took the Palace of Westminster with them. Payment records, once made, are dangerous things to dispose of.

In 1609 the city of Amsterdam opened the Wisselbank, and in doing so proved the thesis of this essay. Merchants held accounts, where a payment was a book entry from one to another, and the bank’s ledger guilders, its “bank money”, traded at a persistent premium, the agio, over the actual coins in the vault. Economists at the BIS have described it, not entirely as a joke, as an early stablecoin ¹⁴. What matters more is that for the first time since Uruk the ledger entry was again openly the superior form of money, and everyone could see it priced daily.

England industrialised the other half of the machine. The Bank of England arrived in 1694, and its running-cash notes with it, and the earliest surviving English cheque is dated 1659, a £400 instruction to the scrivener-bankers Morris & Clayton. Instruments are the easy part, though, and clearing is the hard one. Around 1770, the walk clerks who trudged between London’s banks exchanging cheques worked out that it was easier to meet once a day at the Five Bells tavern on Lombard Street, swap everything and settle only the net.

The Bankers’ Clearing House invented the batch-and-net architecture over lunch, and it still sits inside every card scheme and every ACH today.

As before, pay attention to who invented it. Not the banks, whose partners would likely have vetoed the idea as collusion, but their most junior employees, co-creating an institution born out of sore feet, the desire for a good pint and common sense. The clearing house is perhaps the second great payments invention built by its users rather than its owners, and the banks only formalised what the clerks had already made work.

Returning to reality for a second, it is worth pausing here on what the notes in your pocket or wallet actually became, because this era is where it takes its modern form. A Bank of England note still reads “I promise to pay the bearer on demand”, wording that dates from when the note was a claim on gold in the vault ¹⁵. The gold is (sadly) long gone, and you can now redeem a note only for another note, yet the promise is not empty, because the note remains a liability of the central bank as an entry on its balance sheet.

A banknote is, in the plainest of terms, a non-interest-bearing IOU from the central bank to whoever holds it ¹⁶. That is the quiet radicalism of paper money once it stops being convertible. The value is not in the cotton and the ink, and it is not in any metal, but in the fact that everyone trusts the issuer and the issuer’s entry. The note is a ledger position you can fold into a wallet, which is worth remembering when we reach the day the ledger stops needing the paper at all.

The clearing model, meanwhile, had kept travelling whilst the note was becoming respectable. New York copied it in 1853 and cleared $23.9m on its first day. By 1871 every component of the modern system exists, with the account as the home of money, the note and cheque as its travelling claims, and the clearing cycle as its daily reconciliation. What nobody has yet is speed.

In 1871, ten years after its transcontinental line went up, Western Union began accepting money at one telegraph office for payout at another. Nothing physical travelled. An operator’s message moved the value, and the company’s internal accounts absorbed the difference. It was the bill of exchange at the speed of light, and the verb it produced, to wire money, has now outlived the telegram by decades.

Notice also what the wire was for. The telegraph was built to carry news and railway signals, not money, and payments simply moved in and occupied it, the way it would later occupy the phone line, the mobile network and the internet. This is the co-option habit at full strength, and it has a logic to it. The expensive part of any payment system is reach, and reach is cheapest when you borrow a network somebody else has already run to every town, every street and every house.

Something else was changing in this era, and it is people. Steamships and railways put ordinary citizens, not just merchants and their agents, across borders in numbers the world had never seen, and every traveller faced the old Lydian question in a new form, which is how to be trusted by parties who share no institution with you. The first great answer was the traveller’s cheque. Robert Herries had sketched the idea as “circular notes” in London around 1770, cashable at correspondent banks across Europe, and in 1891 American Express turned it into a mass product, as the company tells it after its president returned from a European trip fuming at how hard it was to get his letters of credit honoured ¹⁷.

Hold onto this thought, because it matters for what comes next. Travel and migration were becoming amplification vectors for payments innovation, demand pulling the system towards instruments that carry trust across jurisdictions, networks and continents. The traveller’s cheque answered that demand with paper. Within a lifetime, a plastic card would answer it at global scale. Whatever else the card networks went on to build, the void they grew up to fill was this same one, the demand for trust that travels.

Two upgrades transformed the wire into a systemic capability. For starters, in 1883 Vienna’s Postsparkasse introduced the postal giro, Georg Coch’s cashless transfer between accounts, and that design spread across the continent and made account-to-account credit transfer, rather than the cheque, the retail default in Germany, the Netherlands and the Nordics ¹⁸. That giro culture is worth holding onto, because it is precisely the fertile soil in which iDEAL, Swish and Wero later grow.

Second, the Federal Reserve began moving money over the wires in 1915, and in 1918 connected its twelve district banks with a dedicated leased telegraph network, settling transfers in central-bank money keyed in Morse code. That network has evolved, without ever being retired, into today’s Fedwire, which makes it the oldest living system in this essay, and arguably the world’s oldest continuously operating electronic interbank settlement system ¹⁹.

By mid-century, the wholesale problem was essentially solved, in that banks can move final money across a continent in an afternoon. The unsolved problem is the consumer standing at a shop counter, whose choices remain coin, note or cheque. *The next era belongs to whoever solves that, and the solution arrives from outside banking.*

In February 1950 Frank McNamara paid for dinner at Major’s Cabin Grill in Manhattan with a cardboard card and a signature, or so the company’s own telling goes, and it is worth saying that the famous forgotten-wallet story behind it was, his own publicist later admitted, invented afterwards, because payments has always understood what makes good marketing ²⁰. What is solidly documented is this. The first Diners Club cards really were cardboard, with plastic arriving only in 1961, and American Express, launching on paperboard in 1958, actually beat them to plastic in 1959 ²⁰²¹. Diners Club started with about 200 cardholders and a couple of dozen New York restaurants, grew to somewhere between ten and twenty thousand members within its first year, and passed forty thousand by the end of 1951. The multi-merchant charge card existed, and with it a new species of company that was neither buyer, seller nor bank, but a network selling trust between two unknown parties in a transaction.

1958 was the year it turned. American Express launched on the 1st October, and thirteen days earlier Bank of America had flooded Fresno, California with roughly 60,000 unsolicited, live BankAmericards in what was called the Fresno Drop. The fraud was spectacular, the losses were enormous, but the idea was unkillable, being a general-purpose card with a revolving line of credit, accepted everywhere.

Licensing turned it into a network, the network became Visa in 1976, and the rival bank association’s Master Charge became Mastercard. A supporting cast then arrived in a rush, with the PIN patented by Goodfellow in 1966, the ATM at Enfield in 1967, generally considered the world’s first, and the magnetic stripe developed at IBM, in the company’s own telling prototyped with a clothes iron ²². BACS and the ACH automated the clearing cycle, SWIFT standardised the cross-border message, and eventually the chip arrived, with France’s Carte Bancaire nationwide by 1992 and EMV written by 1996.

And here the standards story deserves a paragraph of its own, because it is how a cardboard novelty became a global infrastructure. As card production went international, the industry standardised everything, first through the American national standards of the early 1970s and then through the International Organization for Standardization, and by 1985 ISO 7810 had fixed the card’s exact dimensions at 85.60 by 53.98 millimetres and a nominal 0.76 millimetres thick, with companion standards governing the embossing, the magnetic stripe’s tracks, and from 1987 the chip ²³. Every card in your wallet today, and every terminal on Earth that accepts it, obeys those numbers. Standardisation is unglamorous, and it is also how you scale trust between machines that have never met, which by now you will recognise as this industry’s oldest problem in new clothes.

What the card did to the architecture of payments matters more than any one of those components. Authorisation became instant whilst settlement stayed slow, batched and netted at T+2. Underneath the plastic, in other words, the old clearing house never went anywhere, because card settlement still works exactly the way the walk clerks’ tavern arrangement did, everyone’s obligations pooled, netted and settled later in the day or the week. The Five Bells survives inside every card scheme as its settlement architecture. The gap between the promise to pay and the actual money is bridged by the scheme itself, through its rulebook, its interchange and its guarantee that the merchant gets paid even if the cardholder defaults. That bridge was the product, and it is expensive, as everyone on the paying side of it knows.

Before asking why the world paid that price, it is worth being plain about what the thing in the wallet actually is, because the plastic disguises its species. Hold the card up next to Era IV’s banknote. The note reads “I promise to pay the bearer”, and is an IOU from a central bank to whoever holds it.

The card belongs to the same family, a promissory instrument made portable, personal and reusable, except that what it represents is not money you already have but money an issuer is willing to advance you at the moment of purchase, which is to say credit.

This is not one feature of the card among many. It is the founding purpose. Diners Club was a charge card before it was anything else, the Fresno Drop was above all the mass mailing of a revolving credit line, and the network, the interchange and the guarantee were built to make that credit spendable between unknown counterparties.

Now contrast the account-to-account transfer, where there is no credit anywhere in the chain, because the value must already sit in the account before it can move. Strip the credit industry out of this history, and there is no card in it. Keep that distinction close for the rest of the essay and think of it as the moving of promises against the moving of money, because it is the deepest fault line buried underneath the taxonomy we began with.

Why did the world pay that price for half a century? Because the card genuinely earned it, and the economics of why are worth being precise about, since this essay will shortly call the era an anomaly and has no business doing so cheaply. Economists led by Rochet and Tirole later formalised the card network as a two-sided market, a platform that must recruit cardholders and merchants simultaneously, with the interchange fee acting as the balancing instrument that prices each side to keep both aboard ²⁴. Getting that machine to turn over, at 1950s technology, for millions of parties who share no institution, was a genuine coordination miracle, and the scholars who studied Visa’s build-out describe one of the most successful product innovations in commercial history ²⁵²⁶. The card solved distribution, credit at the point of sale, and global acceptance, all at once, decades before any account-to-account system could have. Whatever else this essay argues, it does not argue that the card was a mistake.

The card and the merchant then co-evolved into something neither could leave. Economists call it path dependence, and the textbook example is the QWERTY keyboard, which won early and held on through sheer installed base long after the argument for it had faded ²⁷. Brian Arthur gave the mechanism its name, being increasing returns to adoption ²⁸. Each new cardholder made the card worth more to every merchant, each new merchant made it worth more to every cardholder, and the whole edifice grew heavy with terminals, standards and habit that no rival could cheaply replicate. This is the real reason the card outlived its own logic by decades. It was locked in, and lock-in is a different thing from superiority, which is precisely why a cheaper rail could sit in plain sight for years before it started to take market share or simply win the markets outright.

For half a century the card was worth every basis point, because nothing else could say yes in two seconds at a counter in a city you had never visited other than cash.

The internet’s first payments decade was spent teaching the card to do a job it was never designed for. PayPal, born Confinity in 1998 and listed on Nasdaq by February 2002, succeeded precisely because it wrapped an account layer around cards and bank transfers, using an email address as an alias for a stored balance. Alipay, spun out of Taobao’s escrow in 2004, did the same for a country with almost no cards to wrap, and when smartphones arrived it made the QR code, a piece of printed paper, the acceptance device. A street vendor’s “terminal” now cost nothing. WeChat Pay followed in 2013 and made payments a feature of conversation, and together the pair came to carry roughly nine-tenths of Chinese mobile payments ²⁹.

The most radical proof came from Nairobi rather than Silicon Valley. M-Pesa, launched by Safaricom in March 2007 on feature phones and a network of airtime agents, made the SIM card the bank branch. Across the Vodafone group’s African markets it now serves over a hundred million financial-services customers moving roughly half a trillion dollars a year ³⁰.

The expensive part of payments was never the technology, but distribution and trust, which a telco happened to own already.

Without trying to sound like a broken record, notice what each of these was built from, because none of it was built for payments.

The QR code was invented in 1994 by Masahiro Hara at what was then a division of Denso, to track car parts on a Japanese production line ³¹. The SIM card was a way to bill phone calls, and the email address was a way to reach a person. Each was co-opted into a payment rail, the QR into an acceptance device, the SIM into a bank branch, the email into an account alias.

Biologists have a word for a feature that evolves for one job and is co-opted for another. That word is exaptation, and feathers are the classic case, having warmed a dinosaur long before they helped birds fly into the skies ³². Payments keeps doing this, for the same reason the wire did it a century earlier. Reach is the expensive part, and reach is cheapest when you borrow it.

Yet again, pay attention to who finished the building. M-Pesa’s designers intended a loan-repayment tool, and its users repurposed it into a money-transfer system within weeks, with the agent network and the “send money home” habit co-created between Safaricom and millions of Kenyans. WeChat Pay’s breakthrough was the 2014 digital red envelope, a ritual older than the technology by centuries, folded into software. The lessons repeat from Uruk to Lombard Street. Every payment system in this story that reached real scale was finished by its users, and the record is consistently unkind to the ones designed against that grain.

Meanwhile the banks built the era’s most underrated rail. The UK’s Faster Payments went live in May 2008, offering 24/7 instant account-to-account in a G7 economy years before anyone said “real-time payments” in a boardroom, and it carried 5.09 billion payments in 2024 ³³. And in 2014 Apple Pay completed the circle from the other direction, in that the card survived but as software, with the PAN retreating behind a device token (Apple Pay calls it the device account number) and the plastic behind glass. Every payment credential was now a claim on an account, rendered on whatever digital surface was nearest. *That leaves the question this whole history has been building towards, because if the account can speak for itself, instantly, what exactly is the card for?*

The reversion arrived, and it arrived from the south. India’s UPI launched in April 2016, and a decade on it clears over 240 billion transactions a year, roughly half the world’s real-time volume and more than four-fifths of India’s digital payments ³⁴. Brazil’s Pix launched in November 2020 and within four years outnumbered credit and debit cards combined, with over 90% of adults using it ³⁵. Neither is a wallet wrapped around cards. Both are what Uruk would recognise, being the institutional ledger speaking for itself, except that the ledger entry now clears in seconds, around the clock, with the central bank underneath. And both are co-creations in the fullest sense, designed by central banks and public bodies but built and distributed with hundreds of member banks and fintechs, which is a large part of why they scaled where single-owner schemes stall.

I owe the reader a caveat here, having spent a whole era separating authorisation from settlement, because that separation has not simply vanished on the new rails. Pix and SEPA Instant discharge the interbank obligation in central-bank money in something very close to real time, whilst UPI answers the payer in seconds and settles between banks on a deferred net cycle behind the scenes. The moment you are told the money has arrived and the moment the banks are square with each other are still two different moments. What the record shows is not that the gap disappeared, but that it collapsed from days to seconds or hours, and that on a growing number of rails the two have genuinely become one act.

Two textures matter enormously here, and the first is cost. The average merchant cost of accepting Pix runs at 0.22% of transaction value, against roughly 2.2% for credit cards in Brazil, a full order of magnitude, and the BIS economists who measured it describe the central bank as operating the platform deliberately as a public good ³⁶. India went further still, mandating by statute from 2020 that UPI transactions carry a zero merchant discount rate, a policy now sustained by government incentive payments, which is worth saying honestly, because zero is a political choice with a fiscal subsidy behind it rather than a discovered market price ³⁷.

And that political choice is under live pressure as I write. India’s Finance Ministry flatly denied any plan to reintroduce a merchant discount rate in June 2025, yet by March 2026 a parliamentary Standing Committee was recommending a tiered fee that would spare small merchants and charge large ones, with the Department of Financial Services itself telling the committee that a fee-free UPI is not financially sustainable ⁵¹. The incentive pot that keeps the rail free has meanwhile shrunk to ₹1,500 crore for the 2025 fiscal year, a fraction of what the ecosystem spends running it, and credit cards riding UPI already carry a discount rate of around 2% ⁵¹.

The second texture is intent. Neither system is merely a cheaper rail, because both were built consciously as digital public infrastructure, in India’s case as one layer of an identity-payments-data stack that the IMF credits with roughly doubling account ownership ³⁸. These are state actors building payment rails the way earlier governments built roads and grids, to create and enable commerce, inclusion and wealth in their own markets, and pricing them like infrastructure rather than like a franchise. Put those two textures together, and you can see why I describe what is happening to payment economics as a race towards zero, built on marginal-cost public rails, statutory zero pricing, and interchange caps in Europe. What India’s wobble adds is a caveat worth keeping, which is that zero itself may prove a waypoint rather than the settling price. A race whose finish line drifts between zero and a handful of basis points is still a different sport from one priced in whole percentages.

Honesty also requires the other side of this ledger, because the reversion has costs of its own, and its critics come with receipts. An instant, irrevocable, nearly free payment is instant, irrevocable and nearly free for a fraudster too. Authorised push payment fraud, the scam where a victim is talked into sending the money themselves, cost the UK £576 million in 2025, the bulk of it moving over the instant rails ⁴⁸, and the regulator’s answer, mandatory reimbursement from October 2024 capped at £85,000 per claim, amounts to bolting a dispute scheme onto a rail that was sold partly on not needing one ⁴⁹.

Brazil learned the lesson more brutally, with a wave of kidnappings built around forced Pix transfers pushing the central bank into nighttime transfer limits in 2021, and tighter caps on unregistered devices since ⁵⁰. The card’s toll, it turns out, was never pure rent. Part of it priced dispute rights, fraud liability and the guarantee, and when the toll is stripped away those costs do not vanish but resurface as fraud losses, reimbursement rules and state subsidy. I do not read any of this as the reversion failing. I read it as the unbundling the interlude wondered about, already underway, with the disputes, the guarantees and the credit being rebuilt as separate services on top of the new rails, priced on their own merits rather than as a percentage of every sale.

Without seeking to be outdone, the North is following at its own pace. Europe legislated that banks open up account data through PSD2 from January 2018, and the effect was measurable, with researchers documenting a surge of new payment firms across the continent after the directive ³⁹.

Worldpay, where I have spent part of my career, is itself a child of regulation, though of an older vintage than PSD2. It was the first Payment Services Directive, back in 2007, that invented the licence category a standalone payments company could inhabit, the payment institution, and it was the European Commission’s state-aid conditions on the RBS bailout that forced the bank to divest the business in 2010 and set it loose as an independent company ⁵².

The riverbank, in other words, does not only shape how the industry behaves. It decides which companies exist. UK open banking payments are meanwhile growing 57% a year, with regulation as the accelerant once again. Europe is now assembling Wero on top of SEPA Instant, fifty million users in, and the memorandum signed in February 2026 to link it with the EuroPA schemes, Bizum, Bancomat and Vipps MobilePay among them, spans roughly a hundred and thirty million users across thirteen countries in total ⁶⁰. The US launched FedNow in 2023, and three years later it carries over 1,800 institutions, with volumes still early and the direction unambiguous.

Tokenisation is the same story wearing newer, flashier and more expensive clothes, and it is worth me clarifying plainly what the word means, because the industry uses it for two different things. In one sense it is the security trick behind Apple Pay, where your card number is swapped for a surrogate so the real number never travels anywhere it shouldn’t or would be at risk. However, in the sense that matters here in this essay, it means something more radical. Tokenisation here describes money issued directly as an entry on a shared programmable ledger, where the token is not a pointer to the money but is the money itself. A stablecoin is the clearest example. Strip away all the noise, and it is a digital token, issued on a blockchain, designed to hold a steady value by being backed one-for-one by a reserve of safe liquid assets, usually dollars or other hard fiat currencies like the Euro.

Whoever holds the token holds the claim, and it moves from wallet to wallet with the transfer recorded on the chain with no bank in the middle updating an account.

It is a bearer instrument made entirely of ledger, which is a genuinely old idea rebuilt in code. The United States gave the reputable version a federal rulebook with the GENIUS Act in July 2025, requiring real reserves, real disclosure and a licence to issue ⁴⁰. Regulation again, and this time explicitly as legitimiser.

Here is where the circle actually closes. A blockchain is, in the flattest technical description, a ledger, being a shared, append-only record of who owns what, kept in step across many machines. The newest money technology on Earth is, structurally, the oldest, a clay tablet with a different keeper.

The central bankers have noticed the same shape, because the BIS’s own blueprint for the future of the system is something it calls a unified ledger, tokenised central-bank money and tokenised deposits on one programmable record ⁴¹. The difference is really who holds the pen. The temple ledger trusted one institution to write the next line, whereas a public chain is built to need no single trusted writer at all, and settles by agreement among many.

That difference is not trivial, and I would not wave it away, but the form is unmistakably the same, being value that lives as an entry rather than as a thing. Because the entry and its movement are now one act, clearing and settlement collapse into each other the way they did when a coin changed hands, except at any distance. The catch, and it is a real one, is the counterparty. A coin handed over settles instantly and finally, but you still hold whatever the issuer’s promise is worth, and with a private stablecoin that issuer is a company rather than a central bank, so the risk did not vanish but merely moved.

This is why the same regulators cheering instant settlement are so exercised about reserves, and the collapse of the algorithmic stablecoin TerraUSD in 2022, which fell to a fraction of its peg within days and to pennies within weeks, is the recent proof that a bearer instrument is only ever as good as whoever stands behind it ⁴².

Nor is Terra the only exhibit, because the reputable end of the market has a file of its own. USDC, arguably the best-behaved of the large stablecoins, slipped to 87 cents over a weekend in March 2023 when its issuer disclosed billions stranded at a failing bank ⁵⁵. Tether, the largest issuer of all, paid a $41 million penalty in 2021 after US regulators found its token had been fully backed on barely a quarter of the days they sampled ⁵⁴. And the BIS, whose unified-ledger blueprint I cited approvingly a moment ago, delivered its institutional verdict in 2025, which is that stablecoins fall short of the tests of sound money ⁵⁶.

Notice, finally, where all this places the stablecoin in this essay’s family tree. An IOU on a private issuer is kin to the banknote and the card, the promissory instruments, not to Pix or UPI, which move money that is already there. The form is the oldest in the book, and so is the failure mode.

Cash, meanwhile, has fallen from 44% of global in-store payment value to 15% in a decade, and more than a hundred jurisdictions now run live instant-payment systems ⁴³. And here I should be scrupulously fair to the incumbent, because the card is not fading, and anyone reading this essay as an obituary has misread it. Visa’s payments volume grew 8% in its 2025 fiscal year and Mastercard’s gross dollar volume grew 9%, card purchase transactions worldwide are still compounding at better than 9% a year, and two-thirds of American consumer spending still runs on cards ⁵⁷⁵⁸⁵⁹.

If my argument were that the card is dying, those numbers would bury it. The claim is narrower, and I think sharper. What the five-millennium record marks as anomalous is the card era’s structure, a percentage toll on every transaction in exchange for a guarantee, and the most telling evidence for the race I described at the outset is the incumbents’ own behaviour. The card networks are responding the way strong incumbents do, by buying A2A capability, tokenising themselves into wallets, settling in stablecoins, and competing hardest on the dispute and guarantee layer where their moat is real, which is, as I read it, the strategy you would choose if you thought the pure act of moving value was heading towards utility pricing. I am inferring that from public behaviour rather than reporting it, and they may simply be hedging. That contest is genuinely open, and I would not call it settled in anyone’s favour.

There is a live test of precisely that distinction between the level and the shape, and it concluded whilst I was writing this. In June 2026 a United States court approved a $38bn settlement between the card networks and American merchants, closing twenty-one years of litigation with roughly ten basis points off credit interchange for five years and a cap of 1.25% on standard cards for eight ⁶¹. Read as a price story it is the largest merchant win the industry has ever recorded. Read as a structure story it moves remarkably little, because the toll is still a percentage of every sale, the defaults at the checkout page are untouched, and the architecture that collects it sits exactly where it sat. The payments strategist Dwayne Gefferie put it more bluntly than I would have dared, arguing that even when you win on price the architecture still absorbs it ⁶². I think that is the anomaly in a single sentence, and it is why I keep insisting the shape rather than the level is the thing five millennia mark as strange.

What is not open, on a five-millennium reading, is the direction, because every era of this history has moved the ledger closer to the moment of payment, and the distance is now ten seconds.

The promise and the money now arrive together for the first time since the coin, and this time at any distance.