The Machine That Hired
The ATM was supposed to replace the bank teller. Instead, it made branches cheaper to run, so banks opened more of them, and teller employment doubled over four decades.
In 1865, William Stanley Jevons identified the same pattern in England's coal economy. More efficient steam engines didn't reduce coal consumption. They made coal useful in so many new applications that total consumption soared. The effect bears his name. And it is about to reshape financial services from both directions at once.
The Supply Side
Venmo needed five banking partners, state-by-state licenses across 49 US jurisdictions, and middleware connecting to over 12,000 financial institutions. It still only works in one country.
Every major market had to build its own version, some through government-run rails like PIX and UPI, others through private platforms like M-Pesa and Alipay. Roughly 80 countries now have real-time payment systems.
Almost none interoperate.
Fintechs have been structurally regional because each market requires its own payment rails, banking APIs, and licenses.
A public blockchain replaces that patchwork with a single open ledger, and self-custodial wallets remove the need for market-by-market banking partnerships. That is what allows a company like Sling Money to build a global payments product with 23 people and 3 licenses, even if limited today to the roughly 70 countries with fiat ramps. As Sling CEO Mike Hudack puts it, stablecoins transform payments "from a prefunding and reconciliation problem into an interoperability problem".
The companies building on that shift are not all startups. Stripe acquired stablecoin orchestration platform Bridge for $1.1 billion and wallet provider Privy, then launched stablecoin financial accounts in 101 countries, well beyond the 46 countries where it previously serviced merchants. The same Bridge infrastructure powering Sling's virtual accounts now sits inside a company handling $1.4 trillion in payments a year.
An exporter in Nairobi shows what that infrastructure makes possible. She receives payment from a US importer through a virtual dollar account. She spends at more than 150 million merchants via a stablecoin-linked card. Her idle balance earns 4 to 7 percent in an onchain lending vault. No bank account. No bank. Three years ago, that stack was a pitch deck. Today each layer exists, built by different companies, increasingly composable.
The 1.3 billion adults the World Bank classifies as unbanked are not unbanked because they don't want financial services. They are unbanked because the cost of serving them exceeded what anyone could charge. The average fee for a $200 remittance to Sub-Saharan Africa is 8.45%. That is nearly $17. For a family earning $150 a month, that is something akin to food for a week, or school supplies, or medicine someone will go without.
What happens when those costs drop is not theoretical. M-Pesa made mobile payments nearly free in Kenya, and financial inclusion surged from 27% to 85%; the IMF found the growth was generally additive, not substitution. India launched UPI with near-zero fees and digital payment volume grew from 18 million to over 228 billion transactions in under a decade.
More providers building more products in more markets, because the cost of doing so collapsed. Jevons' Paradox on the supply side.
The Cost Side
Now look inside the banks.
The industry spends $61 billion per year on financial crime compliance in North America alone. Forty-two percent of C-suite time at major banks goes to regulatory activity. Employee hours devoted to compliance grew 61% between 2016 and 2023. Banks are not financial services companies that happen to do compliance. They are compliance companies that happen to do financial services.
Most of that spending, compliance and IT alike, goes toward reconstructing information that the infrastructure was not designed to share. Walk into a bank exam and watch what examiners actually do. They reconcile ledger entries and verify that correspondent account balances match. They trace transactions across three or four intermediary banks through opaque bilateral relationships no single party can see end-to-end.
A shared ledger collapses that problem. When every counterparty writes to the same record, the reconciliation step disappears. Not because compliance gets weaker, but because the information is already there.
JPMorgan's Kinexys platform processes over $2 billion a day, with over $2 trillion settled since launch. The use case: a multinational using JPMorgan across a dozen markets needs to move money between its own accounts in real time. Core banking ledgers are siloed and batch-processed. Kinexys sits on top and makes the money programmable. Settlement drops from end-of-day to seconds, and capital that sat idle between batch cycles gets freed. JPMorgan has begun launching JPM Coin on the Canton Network, where Goldman Sachs, DTCC, and Broadridge are among the announced participants. Banks may prefer tokenized deposits over stablecoins. The underlying value is the same: shared infrastructure that collapses the reconciliation layer.
Make compliance cheaper per unit, and institutions can afford to serve more customers in more corridors.
The Convergence
More providers entering from outside banking because the barriers fell. Lower costs inside banking because the infrastructure improved. As regulatory frameworks like the GENIUS Act and MiCA provide clearer rules of the road, both forces push in the same direction: more financial services, for more people, at lower cost.
Cloud computing did not eliminate data centers. It made their capabilities accessible to anyone with an API key. Stablecoins are doing something similar to banking. The system does not disappear. It becomes infrastructure that others build on.
Jevons watched coal consumption climb as engines grew more efficient and called it a paradox. It was not a paradox. It was a pattern: when the unit cost of a fundamental service drops far enough, the market does not contract. It reaches everyone the old cost structure had excluded.
We are about to find out how many people that includes.
