Daniel sent us this one — he's been thinking about that weird gap between hitting send on a bank transfer and the money actually arriving. You know, the one where your account says "pending" and the money has clearly left, but nobody on the other end has it yet. The question is: if money is just database entries, why can't moving it between accounts be instant the way it is with cryptocurrency? The answer apparently involves something called settlement, which works completely differently depending on whether you're sending money across town or across the world. So where does your money actually go during those three days?
Into a limbo that costs the global poor about six to eight percent per transfer, but we'll get to that. To understand this, you have to start with what your bank balance actually is. It's a row in a database. Specifically, it's a liability on your bank's books. When you deposit a hundred dollars, the bank isn't holding your hundred dollars in a box with your name on it. They're recording that they owe you a hundred dollars. That's it. That's the whole thing.
My life savings is an IOU from people who charge me overdraft fees.
An IOU from an institution that is, to be fair, heavily regulated and insured. Now, when you send money to someone at a different bank, you're asking Bank A to reduce its liability to you and somehow get Bank B to increase its liability to the recipient. Two separate databases, two separate companies, no shared ledger between them. That's the fundamental asymmetry with cryptocurrency. Bitcoin runs on a single shared ledger where every participant agrees on the state of the world. The banking system runs on thousands of private ledgers that don't trust each other and need a reconciliation layer.
That reconciliation layer is settlement.
That's the whole game. Settlement is the process by which two banks actually make good on the IOUs they've been trading all day. And here's the thing that surprises most people: at the interbank level, settlement can happen instantly. The Federal Reserve runs something called Fedwire, which is an RTGS system — Real-Time Gross Settlement. It settles central bank money between member banks in real time, transaction by transaction, irrevocably. The UK has CHAPS, the eurozone has TARGET2. These systems move trillions daily and they settle immediately.
The technology for instant settlement exists and has existed for decades. Which makes the three-day ACH transfer feel less like a technical limitation and more like a choice.
It is absolutely a choice, but it's a choice driven by economics and risk, not laziness. Fedwire settles instantly because both banks hold reserve accounts at the Federal Reserve. The Fed debits one reserve account and credits the other. But holding those reserves is expensive — it's capital that banks can't lend out or invest. So banks use Fedwire for their own large, time-sensitive transfers. For your hundred-dollar venmo to your friend at a different bank, they batch those up.
Batch them up how?
Think of it like this. Over the course of a day, Bank A's customers send maybe ten million dollars total to Bank B's customers, and Bank B's customers send about nine-point-eight million to Bank A's. Instead of settling every coffee purchase and rent payment individually, the banks net those positions. At the end of the day, Bank A owes Bank B two hundred thousand dollars. They settle that single net amount through Fedwire. That's why your transfer sits in "pending." Your bank has debited your account — removed its liability to you — but it hasn't actually sent anything to the other bank yet. It's just tallying up what it owes.
"pending" means "we've deducted it from you but we haven't paid them yet, and we're hoping enough money flows the other direction today that we barely have to move anything.
That is a shockingly accurate summary. The delay isn't the database update — that's instant. The delay is the netting cycle. The Automated Clearing House, ACH, processes transfers in batches. In the US, same-day ACH exists now, but standard ACH still takes one to two business days because it's running through those batch settlement windows. Your bank is optimizing its reserve requirements.
Alright, so domestic transfers are slow because banks are cheap about holding reserves. What happens when I send money internationally?
This is where it gets genuinely baroque. Let's trace a hundred dollars from a Chase account in New York to a Barclays account in London. You initiate the transfer. Chase immediately debits your account — your hundred dollars is gone. Chase then sends a SWIFT message. Now, SWIFT does not move money. I need to say that twice. SWIFT is a messaging network. It's a Belgian cooperative that processes about forty-two million messages a day, and every single one is just a formatted payment instruction. It's the financial equivalent of sending a very formal email that says "please pay this person.
The global payment backbone is basically an email system with nicer formatting.
With very specific field codes and security protocols, but yes. The SWIFT MT103 message is the standard customer transfer instruction. It says: Chase New York instructs its correspondent bank — let's say JPMorgan London — to pay Barclays London one hundred dollars on behalf of this customer. But the message itself moves no funds. Settlement happens through something called correspondent banking, and this is where nostro and vostro accounts come in.
I've heard those terms. They sound like a magic spell from a children's book about accounting.
They're wonderfully straightforward once you translate them. Nostro is Latin for "ours." Vostro is "yours." A nostro account is an account that Bank A holds at Bank B, in Bank B's country and in Bank B's currency. From Chase's perspective, the account they hold at Barclays in London is a nostro account — "our account with you." From Barclays' perspective, that exact same account is a vostro account — "your account with us." Same account, two names, depending on whose books you're looking at.
When Chase sends my hundred dollars to Barclays, what actually moves?
Nothing moves across any border. Chase debits your account. Then, through its correspondent bank in London, it instructs that its nostro account at Barclays be debited by a hundred dollars, and Barclays credits the recipient's account by the same amount. The money never left London. The dollars never crossed the Atlantic. What happened was a series of ledger entries: Chase USA reduced a liability to you, and Chase's London operation — or its correspondent — reduced an asset it held at Barclays UK, which then increased a liability to the recipient. No armored truck drove through the Channel Tunnel.
It's just different people editing different spreadsheets and trusting that everyone will honor the edits.
That trust is the entire reason this takes days. Each bank in the chain needs to verify the instruction, check sanctions lists, screen for anti-money laundering flags, confirm that the nostro account actually has sufficient funds, handle currency conversion through the FX markets if needed, and then reconcile its own books. If the transfer passes through three correspondent banks — which is common for corridors that don't have direct banking relationships — each one adds its own verification window. Plus time zones. If you initiate a transfer at four PM New York time, London is closed. Nothing happens until the next business day.
The actual database updates are instant, but the verification-and-trust choreography across multiple institutions in multiple jurisdictions is what creates the three-day gap.
That's just for major currency corridors. If you're sending money from, say, a small bank in Pakistan to a credit union in the Philippines, you might be looking at five to seven hops through correspondent banks, each taking their fee. The World Bank tracks this — the average cost of sending remittances to developing countries is still above six percent globally. In some corridors, particularly in sub-Saharan Africa, it's over eight percent. That's not a technical friction cost. That's a tax on people who can least afford it, extracted by a correspondent banking network that has no incentive to get faster or cheaper because the people paying the fees aren't the ones choosing the system.
The correspondent banking network is essentially a trust graph, and if you're a small bank in a developing country, you're at the edge of the graph with the longest path to everyone else.
That's exactly the right way to think about it. There are maybe ten to fifteen global correspondent banks — JPMorgan, Citibank, HSBC, Deutsche Bank, a handful of others — that handle the vast majority of cross-border volume. Smaller banks don't have direct relationships with banks in every country, so they maintain an account at one of these global correspondents and route everything through them. It's a hub-and-spoke model. Efficient for the hubs, expensive for the spokes, and it creates enormous systemic risk. If one of those major correspondent banks fails, the entire network seizes up.
Which brings us to the obvious question. Cryptocurrency settles in minutes, sometimes seconds, across borders, without correspondent banks. Why can't fiat currency do that?
Because cryptocurrency collapses trust and settlement into a single layer, and the fiat system has deliberately kept them separate for reasons that aren't stupid. When you send Bitcoin, the transaction is broadcast to the entire network, miners or validators include it in a block, and once you have enough confirmations, the entire network agrees that those coins now belong to the recipient. There's no separate settlement step because the transaction and the settlement are the same event. The ledger update is the transfer.
That also means there's no undo button.
Finality in crypto is probabilistic — after six confirmations on Bitcoin, the transaction is considered settled, but there's no central authority that can reverse it if you were scammed or if you sent to the wrong address. The fiat system deliberately introduces a gap between the payment instruction and the settlement so that there's a window for checks, reversals, and interventions. That's not a bug. It's a feature that protects consumers and enables regulatory compliance. The speed of crypto is a design trade-off, not a pure technological advantage. You're trading finality risk and consumer protection for settlement speed.
When someone says "crypto is faster because the technology is better," the more precise version is "crypto is faster because it doesn't have to do any of the things the banking system was designed to do.
And some of those things — like the ability to reverse a fraudulent transaction — are valuable. Others — like the correspondent banking fees that extract six percent from remittance payments — are rent-seeking on an outdated architecture.
What's actually changing? I know there are real-time payment systems popping up.
Several, and they're impressive. The UK launched Faster Payments in 2008 — it was the first major real-time retail payment system. India's UPI launched in 2016 and it's now processing about ten billion transactions a month with instant settlement. It's the largest real-time payment system on the planet. The US finally launched FedNow in July 2023, and as of mid-2026 it has about nine hundred participating financial institutions. These systems settle retail payments in seconds by running on top of the central bank's real-time gross settlement infrastructure.
They're domestic only.
That's the catch. UPI is extraordinary within India, but the moment you try to send money from India to Dubai, you're back in the correspondent banking swamp. FedNow works between US banks. Faster Payments works between UK banks. None of these systems talk to each other across borders. The real-time revolution has been walled off by national boundaries.
We've solved the domestic problem with technology that's been around for years, but the cross-border problem is still running on a system designed in the 1970s.
SWIFT knows this. They launched something called GPI — Global Payments Innovation — which is an overlay on top of the existing SWIFT network that adds tracking, transparency, and faster processing. In a 2021 pilot, they got fifty percent of cross-border payments down from three days to under thirty minutes. But here's the thing: that's an optimization on top of the correspondent banking architecture. It's not a redesign. The money is still moving through nostro and vostro accounts at correspondent banks. GPI just makes the messages faster and the fees more transparent.
It's like putting a GPS tracker on a stagecoach. You can see exactly where your money is stuck, but it's still stuck.
The stagecoach drivers are still taking their cut. Which brings us to the interesting question: what about CBDCs?
Central bank digital currencies. The thing China's been pushing and everyone else has been nervously watching.
China's digital yuan, the e-CNY, had about a hundred and twenty million wallets as of early 2026. The European Central Bank is targeting 2027 for its digital euro pilot. The theoretical promise of a CBDC is that it enables direct settlement between any two parties using central bank money, with no commercial bank intermediary and no correspondent banking chain. It would be like giving every citizen and business a direct account at the central bank, with transactions settling instantly on the central bank's ledger.
Which sounds great from a speed perspective and terrifying from a privacy perspective.
That's the trade-off. If every transaction settles on the central bank's ledger, the central bank sees every transaction. Every coffee, every rent payment, every political donation. The Chinese model doesn't even pretend to care about privacy — the e-CNY is designed for surveillance. The European model is trying to build in privacy protections, but the fundamental architecture makes anonymity difficult. A commercial bank doesn't care what you spend your money on, as long as you're not laundering it. A central bank with a full transaction ledger has very different incentives.
The privacy we currently have in banking is partly an accident of the correspondent banking architecture. The central bank doesn't see your individual transactions because they're buried inside net settlements between commercial banks.
The settlement delay and the multi-layered banking system create a de facto privacy shield. It's not a privacy system by design — it's a privacy system by accident of architecture. And that's one of the reasons the transition to real-time settlement and CBDCs is going to be so politically fraught. You're not just changing the speed of payments. You're changing who can see what you do with your money.
Alright, so we've got this landscape: domestic real-time rails are here, cross-border is still stuck in the 1970s, CBDCs are coming but they bring their own problems. What should someone actually do if they need to move money internationally today?
Use services that have pre-solved the trust problem. Companies like Wise, formerly TransferWise, and Revolut maintain pre-funded local accounts in every country they operate in. When you send money through Wise from the US to the UK, your dollars go into Wise's US bank account, and Wise pays out pounds from its UK bank account. The money never actually crosses a border. It's two domestic transfers that happen to be coordinated by a company that holds balances in both jurisdictions.
They're doing what the correspondent banks do, but they've pre-funded the accounts and they're optimizing for speed instead of extracting fees at every hop.
Because they're moving large volumes in both directions, they can match flows internally. If lots of people are sending dollars to pounds and roughly the same amount is going pounds to dollars, Wise barely needs to move money between its own accounts. It's netting, but at the retail level and across borders. The technology isn't revolutionary — it's the business model that's different. They've internalized the correspondent banking function and made it consumer-friendly.
The innovation isn't faster settlement. It's a company saying "we'll hold money everywhere and settle with ourselves.
Which is, when you think about it, what the correspondent banks have been doing for decades. They just had no incentive to pass the speed or cost savings on to consumers. The correspondent banking network is a trust graph that charges rent at every node. Wise and Revolut are trust graphs that charge a flat fee and settle in minutes. Same underlying mechanism, completely different incentive structure.
This whole system is a fifty-year-old solution to a problem that predates the internet, and we've just been layering band-aids on it ever since.
The band-aids work, to be fair. FedNow is real. UPI is real. SWIFT GPI has meaningfully improved cross-border speeds for many corridors. But the underlying architecture — the idea that banks don't trust each other and need a separate reconciliation layer — hasn't changed. The question is whether CBDCs will finally collapse the trust and settlement layers the way cryptocurrency did, but with central bank backing and regulatory compliance.
Or whether the settlement delay is actually load-bearing. If settlement becomes instant everywhere, do we still need commercial banks as intermediaries? What's the bank's role if the central bank is running the ledger directly?
That's the existential question for the banking industry over the next decade. Banks currently earn float income on the gap between when they debit your account and when they settle with the other bank. If settlement is instant, that float disappears. Banks earn fees on correspondent banking. If CBDCs enable direct central bank settlement, those fees disappear. Banks provide credit — mortgages, business loans, credit cards — and that function doesn't go away, but it becomes decoupled from the payment system in a way it hasn't been historically.
The three-day pending period isn't just an annoyance. It's a business model.
It's a business model built on a trust architecture that made sense when international communication meant telex machines and paper ledgers. The fact that it's survived into an era of real-time global communication is a testament to how deeply embedded it is in the legal and regulatory framework of global finance. You can't just replace the technology. You have to replace the laws, the treaties, the compliance regimes, and the risk models that were built on top of it.
When I see "pending" on a transfer, what I'm actually seeing is the ghost of a world where banks sent each other paper checks by mail and needed three days to make sure nobody was cheating.
The ghost is still collecting rent. That's the part that should bother people. The technology for instant settlement has existed for decades. The delay persists because the institutions that profit from the delay are the same institutions that would need to implement the change. It's not a conspiracy — it's just misaligned incentives. No individual bank can unilaterally speed up cross-border settlement because it requires coordination across the entire correspondent banking network. And the network, as a whole, benefits from the float and the fees.
Alright, let me try to synthesize this into something I can actually use. When I send a domestic transfer and it says pending, my bank has debited my account but hasn't settled with the receiving bank yet. The money is in a weird limbo where it's no longer my bank's liability to me but not yet the other bank's liability to the recipient. For international transfers, it's worse — my money is hopping through nostro accounts at correspondent banks, each one verifying and taking a cut, and the SWIFT message is just the instruction telling each bank what to do. None of this is slow because the computers are slow. It's slow because the trust architecture requires multiple institutions to verify, reconcile, and settle before anyone considers the transfer final.
That's it. And if you need to move money internationally, use a service that maintains pre-funded local accounts — they've already done the settlement step before you even initiated the transfer. You're essentially buying money that's already in the destination country.
The next time someone tells me crypto is faster because the technology is better, I'll point out that it's faster because it doesn't have a separate settlement layer, and that's both its strength and its weakness. No reversals, no chargebacks, no regulatory compliance checks. Speed is a trade-off, not a pure win.
The fiat system could be just as fast if we were willing to live with the same trade-offs. But we've collectively decided, through decades of regulation and institution-building, that we want the safeguards. The question now is whether we can build a system that has both — the speed of a shared ledger with the consumer protections and privacy of the banking system. That's the CBDC experiment, and we're living through it.
Which means the next five to ten years are going to be a hybrid mess. Legacy correspondent banking for high-value cross-border transfers, real-time rails for domestic retail, CBDCs for government-issued digital cash, and crypto continuing to exist as the pressure that forces the legacy system to improve. The interesting question is which one wins for cross-border payments.
My bet is on the non-bank providers — the Wises and Revoluts of the world — absorbing more and more of the consumer cross-border volume, while the correspondent banking network consolidates around high-value corporate and institutional transfers. CBDCs will probably handle government disbursements and maybe some wholesale settlement between banks. But the idea of a single global payment rail is still a long way off.
Next time I send a wire transfer, I should check the fine print for the settlement date.
It's a tiny window into an enormous invisible machine. Most people never see it. And now you know what's happening behind the curtain — a bunch of banks editing their own spreadsheets and sending each other very formal emails about it.
The global financial system, ladies and gentlemen. Spreadsheets and emails.
Spreadsheets and emails, secured by a trust graph that costs the global poor six to eight percent per transfer. It's remarkable that it works as well as it does.
Now: Hilbert's daily fun fact.
Hilbert: In the 1950s, the Namib Desert's Welwitschia mirabilis plant, which can live for over a thousand years, was discovered to produce a sugary nectar specifically to attract ants that would then defend it from herbivores — essentially hiring a private security force in exchange for snacks.
A plant with a defense budget.
Nature's been doing correspondent banking this whole time.
This has been My Weird Prompts. Thanks to our producer Hilbert Flumingtop. If you enjoyed this episode, tell someone about the show — word of mouth is how we grow. You can find every episode at my weird prompts dot com. I'm Corn.
I'm Herman Poppleberry. Check your settlement dates.
