What's better, an instant payment or one that takes two days?

You'd think the obvious answer is instant, but it depends who you ask.

For a merchant, instant is everything. Today, they often make a sale, and the inventory is gone, but wait days (or in some cases weeks and months in B2B) to get paid. This means they have to borrow to cover the gap, or just accept this risk in their business. Consumers too, are much more reactive and likely to use a product if payments are instant.

But that's just one perspective.

For a corporate treasurer or a bank, the slower payment could actually be far more cost-effective with almost zero negative consequence. This is thanks to the magic of netting and liquidity savings mechanisms in most large payments clearing networks today. TCH, Faster Payments, CLS, and insert your favorite acronym here all make payments much more efficient for the balance sheet.

This contrast in perspectives is really important now, because stablecoins and tokenized deposits are re-wiring how money works. Stripe, Visa and Mastercard are all in on stablecoins for faster settlement, while the banks are backing tokenized deposits for many of the same instant payment benefits, but with some important liquidity advantages.

Then last month, seventeen of the biggest banks in America announced they'd clear and settle tokenized deposits onchain through The Clearing House, and half my feed asked the same question. Why didn't they just use a stablecoin?

By the end of this rant you'll know why they never would. You'll also know why the banks' favorite trick is about to get stolen right back.

Here's the plan:

In the past, every payment was physical. I would hand you $10 and right there, it's settled, instantly. Most payments worked this way. To settle a debt, you had to move cash or paper records from one place to another.

This was slow and clumsy, so banks created "netting." Instead of moving two separate payments, they added them up and moved only the difference. So if a bank has to pay all of the other banks $10m, and all of the other banks owe the bank $9m, it's far easier for them to just send the $1m out to some central payment system operator who can "net" off all those positions.

This simple trick saved banks from keeping huge piles of cash sitting idle while waiting for the next truck or mail delivery. It turned a mountain of physical work into a tidy accounting shortcut.

Over time, the trick became software.

Banks created "Liquidity Savings Mechanisms" (LSMs). The kind of jargon term only a banker could love. My general principle for this type of jargon is that it's always easier if you break it down word by word:

The best example is The Clearing House (TCH), which runs the CHIPS network. The biggest banks use CHIPS to settle high-value dollar payments. Every payment goes into a central queue, where a patented algorithm spends all day hunting for offsets. Bank A's outgoing $50m meets Bank B's incoming $48m, the algorithm cancels them against each other, and only the difference touches anyone's funding. Payments still settle in seconds. The liquidity barely moves.

The numbers are outrageous:

Before LSMs, banks kept massive idle piles of cash in reserve to cover the gaps while payments settled. Because the queue now clears debts continuously through the day, those reserves shrank, and the capital went back to work in lending and investment. TCH puts the savings for CHIPS participants at $5.5bn a year.

CLS does the same trick for FX. On an average day it settles over $8 trillion across 18 currencies, and multilateral netting shrinks what members have to fund by over 96%. On its record day it settled $19.1 trillion of payment instructions with $72 billion of actual funding.

That's 0.38%.

For every $100m settled, a member funded $380k.

Proof, if you ever needed it, that bankers can get blood from a stone.

And one more thing. TCH also runs RTP, America's fastest payment rail. RTP settles every payment one by one, instantly, from a prefunded account at the New York Fed. To go instant, RTP gave up netting and prefunded instead.

Keep that trade in your head. Stablecoins made the same one.

Netting has one requirement. Flows have to offset. And a huge amount of the world's money can't, because the plumbing to net it was never built.

If you look at the major use cases of stablecoins that have emerged, they plug holes in the financial system where it was either not global, instant, or 24/7.

This can be someone in the global south finally able to access Netflix or Claude, an importer in Bolivia finally able to access trade finance at an affordable cost, or it could be a giant institution choosing to trade 24/7.

On Tokenized, El Dorado CEO Guillermo Goncalvez described his customers in Bolivia, Paraguay and Ecuador: import economies where China is the main trading partner, where multinationals route payments through their US subsidiaries and small importers of electric vehicles, health equipment and basic materials have nothing like that.

Those smaller firms now pay Chinese suppliers in stablecoins.

He says Toyota Bolivia accepts USDT directly, five or six of Bolivia's biggest banks offer USDT wallets (buy and sell within office hours, Monday to Friday), and El Dorado has just opened a brick-and-mortar office in Bolivia because small importers wanted somewhere they could come in, ask questions and fill in their customs forms. Then he described his diary:

These are all huge problems that need solving, and the more efficient, legacy systems simply didn't. So whenever we discuss tokenized money I come back to my first principle:

Stablecoins are money that moves. They can go anywhere, anytime. All you need is a compatible wallet

Deposits are money at rest. They live inside a bank, and for another bank to accept one, you need a clearing intermediary both sides trust. Put another way, you need an RTP-like network for them to have that same instant movement, and those clubs tend not to welcome new members easily.

Stablecoins made real-time payments a new global default. They came with one catch, a hidden tax. Prefunding.

Stablecoins settle in seconds yet still require 100% prefunding. To send a stablecoin you need to have that stablecoin. In fact, that pre-funding requirement is true for nearly all instant payment methods.

Stablecoins under the GENIUS Act (section 4(a)(11), for the lawyers) also bans issuers from paying yield on the float. So in a scenario where you're moving stablecoin dollars very quickly, the underlying fiat is often locked, and doesn't work weekends. It's sitting there, pre-funding everything you do in stablecoin land. It doesn't have the efficiency that comes from something like CLS.

Netting is where the capital comes back in TradFi. Most corridor flows offset, because money going out meets money coming in. An LSM queues obligations, cancels them against each other, and settles only the net.

Imagine a bank pushing $1 billion a day through a corridor, funded ahead of a two-day settlement cycle. Settled gross, that's roughly $2 billion parked ($1bn per day), and at a 4% opportunity cost, an $80 million a year friction tax. Compress the gross flow by even 90% with netting and the bank funds $200 million across that cycle instead. Same payments, a tenth of the capital.

Speed then works on the remaining $200m. The net number sits in transit for minutes instead of days, and you can run net cycles hourly instead of daily, which caps the credit exposure that builds between them.

A stablecoin settles the gross number instantly, and makes you prepay it.

Now put those two ideas together for a second.

Industry estimates put $15 trillion or more in nostro and vostro balances worldwide, parked to cover settlement across time zones and closed windows.** Capital whose full-time job is waiting for a rail to open. Yes, it's netted, liquidity saved and compressed, but it's still a LOT.

So imagine a best-of-both-worlds scenario.

Bank people will tell you stablecoins can't net, so they lose. Crypto people will tell you netting is what slow rails do to cope, so who cares. I think both camps are about to be surprised.

Nobody prefunds payments because it's some ancient law of the universe. They prefund because there isn't a central intermediary (like a clearing house). And yet nothing about a token forbids all of the things a payments engine does, like a queue, a matching engine, and a net settlement cycle. A blockchain is a shared ledger with programmable rules, which makes it arguably the best netting machine ever built.

Someone just has to write the smart contract for netting.

However, to do that, a dollar has to be a dollar. And today's stablecoins don't work that way: USDG, USDC, and the litany of new stablecoins like KlarnaUSD or MoneyGram's MGUSD don't all easily exchange between each other for $1. USDC ranges between 0.9995 and 1.00013 USD. They also don't convert back to $1 with zero friction, and as yet nobody is building netting.

Think about it this way: a clearing house does three jobs.

Now the start-up universe is reverse-engineering clearing for the tokenized money age.

The Better Money Company is the clearinghouse for exactly this. Founded by ex-a16z investor Sam Broner and backed by a $10m seed led by a16z crypto, it swaps any compliant stablecoin for any other at par, with Paxos, Bridge, MoonPay, and many other partners live, with settlement on 15 different assets. Singleness of money, as a service.

Ubyx, founded by ex-Citi payments veteran Tony McLaughlin, is a clearing system where banks and fintechs redeem any supported stablecoin at par into ordinary accounts, under a common rulebook. Ripple, Paxos and a dozen other issuers have signed on, Barclays invested, BitGo acts as a settlement agent. One connection, every coin, out at par. The same move Visa pulled so a small bank could accept any card without building a network.

Glacis Labs built ZeroDelta, a multichain clearing layer that matches and nets digital asset transfers across 40+ chains, then settles only the residual onchain. Over $1 billion cleared to date, and a $6.8m seed led by Lightspeed Faction (with Franklin Templeton and Coinbase Ventures on the ticket) announced six days before I wrote this. It's the CHIPS trick without the banking license.

And the wildest idea. Cycles maps who owes what to whom across a whole network and cancels the loops. If A owes B, B owes C, and C owes A, that's a triangle nobody needs liquidity to clear. It's netting one layer up the stack, at the obligation level, and it's either brilliant or a decade early. Founded by Cosmos co-founder Ethan Buchman, who calls clearing "a financial superpower," it just raised $6.4m to clear the most debt, for the most people, with the least money moved.

Even Circle is circling this. Its payments network (CPN) is a closed, permissioned network where member banks and fintechs settle directly with each other in USDC, 24/7, under Circle's rules. That's a clearing club in everything but name. The netting layer is the brick still missing from the wall, and everyone above is fighting to lay it.

The harder question is always how, and with whom.

So coming back to the question. Why didn't TCH use a stablecoin?

On June 5th, seventeen of the largest US banks (JPMorgan, Citi, Bank of America, Wells Fargo and friends) said The Clearing House will build on-chain clearing and settlement of tokenized deposits, plus a connectivity layer into RTP and CHIPS, running 24/7.

So these banks now get 24/7 money movement capability and 26:1 netting, and it works with all the systems they have today. Logical. They didn't need to do this with a stablecoin, because TCH has been the intermediary between banks that helps clear and accept different bank deposits since 1853.

The banks read the stablecoin pitch but looked at gross, prefunded settlement and said no thanks; we have something much more efficient, and we solved interoperability in 1970. For the flow between banks today, especially large, wholesale payments, this is a major upgrade.

Directionally though, you can read the bet. Keep the deposit, keep the netting, and add the part stablecoins proved everyone wants. Programmable, 24/7 settlement, applied to the net number instead of the gross one.

The downside is, as big as TCH is, it's not one ledger to rule them all. Stablecoins can go between any clearing network, any payment rail, and to any wallet with compatible software. You don't need TCH, Faster Payments in the UK, UPI in India, and CLS, and countless others to integrate bilaterally with a rulebook for stablecoins.

They just work. They just go. You probably do need to figure out how to make all of that compliant (and, we're still waiting on the final regulators' rules for stablecoins under GENIUS). So while it won't happen fast, my take is that stablecoins and tokenized deposits will become part of a 24/7 money loop.

Clearing was a great solution for deposits in 1853, 1970, and today.

But we also have new opportunities, for new kinds of money, with new tradeoffs.

That space for innovation is exciting and genuinely staggering.

If you look at our choices for how money could settle it looks like this:

Everyone is now racing to box four from a different starting square.

The big banks are taking what they do and extending it to be 24/7, the stablecoin start-ups are making that work for the existing stablecoin industry, and in the middle, in my day job at Tempo*, I'm spending most of my time thinking about how a settlement chain and smart contracts can be built with liquidity savings mechanisms baked in. And that works for deposits, central bank money or stablecoins.

Because surely, that is the future of money.

So what's better, an instant payment or one that takes two days?

It was a false choice all along. The treasurer was never anti-instant. The merchant was never anti-efficiency.

The winning rail settles a small net number, fast, whenever you want.

ST.

PS. Come talk to us at Tempo if you're looking at how to build much more efficient and instant payments rails. (Seriously, you should).

* Ok, netting and speed. But one of them is fifty years old and the other one got all the venture funding.

** Nostro/vostro estimates range from $15 trillion to $27 trillion depending who's counting. Directional, not gospel.

If you enjoy this kind of content, I can guarantee you’ll love being in a room of 1,500 other folks who love to go deeper into where finance meets AI. That’s a huge theme for us at this year’s Nerdcon in San Diego on the 19th and 20th November. I’m bringing my audience, the operators, the people who read this newsletter. And it’s the perfect place to find your next hire, client, or just get inspired. Let’s make events awesome again.

Remember, if you're enjoying this content, please do tell all your fintech friends to check it out and hit the subscribe button :)

Want more? I also run the Tokenized podcast and newsletter.

(1) All content and views expressed here are the authors' personal opinions and do not reflect the views of any of their employers or employees.

(2) All companies or assets mentioned by the author in which the author has a personal and/or financial interest are denoted with a *. None of the above constitutes investment advice, and you should seek independent advice before making any investment decisions.

(3) Any companies mentioned are top of mind and used for illustrative purposes only.

(4) A team of researchers has not rigorously fact-checked this. Please don't take it as gospel—strong opinions weakly held

(5) Citations may be missing, and I've done my best to cite, but I will always aim to update and correct the live version where possible. If I cited you and got the referencing wrong, please reach out