Money Never Dies. It Molts.
Banking has been declared finished before — by the goldsmith, by the gold standard, by the branch, by the cheque book. Every time, the institution that “died” turned out to be a costume the underlying function was wearing. The data on what’s actually happening in 2026 points to something narrower, and more interesting, than extinction.
Every generation of bankers has watched something they were certain was permanent turn out to be provisional. The goldsmiths of seventeenth-century London thought their vaults were the point; it turned out the receipts they issued against those vaults were the point, and the receipts became banknotes. The Bretton Woods architects thought gold convertibility was the anchor of the system; in 1971 the anchor was cut loose and the system did not sink — it re-anchored itself in trust, law, and central bank credibility instead. Each time, a specific technology of money died. The function it performed did not.
That distinction — between a technology and a function — is the place to start when asking whether banking is heading toward the fate of the systems it once replaced. It is a fair question to ask today, when mobile financial services have onboarded more first-time account holders in a decade than branch banking managed in half a century, when central banks representing nearly the entire global economy are actively piloting digital currencies, and when letters of credit — instruments older than the printing press — still move the fuel, machinery, and grain the world runs on.
01 The pattern obsolescence actually follows
Look closely at every transition on the timeline above and a pattern repeats: the old form does not vanish, it gets demoted to a special case, while a new form absorbs the everyday load. Coins did not eliminate barter — barter still happens between states that distrust each other’s currency. Banknotes did not eliminate coins — they eliminated the need to carry coins for large transactions. Electronic transfers did not eliminate cash — they eliminated cash’s monopoly on speed. What obsoletes is convenience, not concept. The function — storing value, moving it, pricing risk on it, extending credit against a promise — has had an unbroken run stretching back to the first grain ledgers of Mesopotamia. Only its packaging has ever been retired.
The institutions that actually disappeared were the ones that confused themselves with the function they performed — not the ones whose packaging changed.
02 The forces actually moving the needle
Three forces are converging on the banking function simultaneously in a way that has not happened before, and each threatens a different layer of the stack.
-
Sovereign digital money. Central bank digital currencies have gone from a curiosity to a near-universal research agenda in six years. This threatens the payments layer of banking — the part where a bank is simply a trusted ledger-keeper between a payer and a payee — because a central bank can, in principle, keep that ledger itself.
-
Private dollar-tokens. Stablecoins have grown roughly twelvefold since 2020 and now settle trillions of dollars in transfers a year. The United States chose in 2025 to regulate this private-money layer rather than compete with it through a retail digital dollar of its own — a significant bet on private rails over state ones.
-
Algorithmic underwriting. The slower but deeper disruption is to credit judgment itself. Machine-driven underwriting and embedded finance are unbundling that judgment from institutions that once held a monopoly on it, letting platforms price risk directly off transaction data.
Only three of those 146-plus countries — the Bahamas, Jamaica, and Nigeria — have actually launched a fully live retail CBDC so far; the rest sit somewhere between research and pilot. That gap between exploration and deployment is itself the story: the direction of travel is nearly unanimous, but the pace is deliberately slow, because getting the payments layer wrong at national scale is not a mistake any central bank wants to make twice.
The stablecoin line is the sharper of the two curves, and that matters. Sovereign digital currencies are advancing through careful pilots; private, dollar-pegged tokens are already moving markets. With almost the entire stablecoin market denominated in US dollars, the fastest-growing new form of digital money in the world is, for now, a private extension of the dollar system rather than a neutral global rail — a detail with real consequences for any economy that depends on dollar liquidity for trade.
03 So will banking become obsolete?
Not in the sense the question usually intends. There will not be a future without institutions that gather deposits, price credit risk, and stand behind the promises that make trade and investment possible — that function is closer to a fixed feature of any economy complex enough to have strangers transact with each other on credit than a Victorian relic awaiting its replacement. But yes, in a narrower and more uncomfortable sense: the specific bundle a mid-sized commercial bank performs today — deposits, payments, cross-border settlement, retail lending, advice, all under one roof — is already being taken apart piece by piece, and that bundle is what is genuinely at risk of going the way of the goldsmith’s vault.
The next decade will not obsolete banking. It will obsolete the full-service branch bank as the default shape banking takes — the way the horse-drawn carriage was not obsoleted by the disappearance of the “getting somewhere” function, but by a car that did that one thing better and left the stable to become a garage.
What survives the unbundling follows a reasonably reliable filter: functions requiring trust backed by a balance sheet and a regulator survive inside institutions; functions requiring only accurate ledger-keeping get taken by whoever does it cheapest — increasingly a central bank, a stablecoin issuer, or a platform. Deposit insurance, capital adequacy, and the lender-of-last-resort relationship with a central bank are not features a fintech can casually replicate; they are the actual moat. A trade finance desk that structures a documentary credit against a shipment of machinery six months before it exists, prices the country risk correctly, and stands behind that promise when the counterparty bank on the other side has never heard of the buyer — that is underwriting judgment plus a balance sheet plus a legal system that recognises the guarantee. It is closer to un-automatable than most of a bank’s income statement.
04 A market that has already run this experiment
-
Some economies offer an early look at how this settles, because they have already lived through one full cycle of “will the old system survive the new technology” — and the answer was neither extinction nor survival unchanged, but exactly the molting described above. Mobile financial services did not replace conventional banking; they extended banking’s core function — moving and storing value safely — to tens of millions of people the branch network was never going to reach economically, using rails the branch network does not own.
Central bank research into a domestic digital currency in such markets tends to reach a similarly unglamorous conclusion: worth studying carefully, requiring infrastructure that does not yet fully exist, and best pursued by learning from pilots elsewhere rather than rushing ahead of them.
05 What to watch next
- Whether CBDCs stay wholesale or go retail. A wholesale CBDC, used only between banks and the central bank, reinforces existing banks. A widely adopted retail CBDC lets ordinary depositors hold money at the central bank directly — the scenario that genuinely threatens deposit funding for commercial banks.
- Whether stablecoin regulation converges or fragments. A world where major economies each ring-fence their own regulated stablecoin regime looks very different from one where a handful of dollar-pegged tokens become the default rail for cross-border trade settlement, bypassing correspondent banking almost entirely.
- Who ends up owning the underwriting data. The institution — bank, platform, or telecom — holding the richest, most current picture of a borrower’s cash flows will end up pricing that borrower’s risk, regardless of what its business card says.
None of this resolves into a headline as clean as “banking is dying” or “banking is safe.” It resolves into what every previous transition actually produced: a smaller number of institutions doing the parts of banking that require balance sheets, legal standing, and regulatory trust, connected to a much larger and more crowded layer of technology providers doing the parts that only require moving information accurately. The goldsmith’s vault is gone. The banknote it produced is still, in spirit, in every wallet on earth.