Instant Rails Elsewhere

43.0 What this chapter gives you#

1. You will be able to explain why Brazil, the United States and the euro area reached three different outcomes from the same forty-year-old technology.
2. You will be able to name the design choices that predict adoption, central operation or compulsion, mandated reach, a rule-based price cap and memorable addressing, and test any new system against them.
3. You will be able to separate Pix the scheme from SPI the settlement system and DICT the directory, which everyday usage runs together.
4. You will be able to explain why a Pix payment and a Faster Payment look identical to a customer and put risk and liquidity in completely different places for a treasurer.
5. You will be able to argue that the United States has more instant payment infrastructure than most countries and less instant payment, and say exactly what it is missing.
6. You will be able to quote the three Same Day ACH deadlines and say what happens to a file that misses all of them.
7. You will be able to state what the Instant Payments Regulation actually obliges, and the date each obligation bites for euro-area banks, non-euro-area banks and payment institutions.
8. You will be able to read the TIPS settlement dip of October 2025 as the expected cost of a well-executed mandate rather than as a failure.
9. You will be able to say who is paying for a “free” instant payment in each of the three jurisdictions, and whether they will still be paying in five years.
10. You will be able to diagnose what Britain has, what it lacks, and why the missing piece is alias addressing rather than speed.

On the same Tuesday morning, three people pay the same small amount of money to somebody they can see.

In São Paulo, a woman buys two kilos of tomatoes at a street market. The stallholder turns a laminated card round so she can photograph the square black-and-white pattern on it. She taps twice, the stallholder’s phone chimes, and the money is his, finally, irreversibly, at eleven o’clock in the morning. Neither of them paid anything to move it.

In Columbus, Ohio, a man pays his window cleaner. He has the window cleaner’s bank routing number and account number written on a slip of paper, and he types them into his bank’s website. It tells him the payment will arrive on Thursday. He also has the option to pay a fee to make it arrive today, and there is a third option, which is to write a cheque, and this is 2026.

In Rotterdam, a woman pays her plumber from her phone. Before she confirms, the screen shows her the name registered to the account she has typed, and it matches the name on the invoice. She presses send. The plumber’s bank credits him within ten seconds, and by law her bank may not charge her more for this than it would have charged for a payment that took a day.

Three countries, three answers, one question. The question is the same everywhere and it is not a technical question. Moving a number from one ledger to another has been technically trivial for forty years. The question is who decides, who must join, and who pays. Brazil answered it one way, the United States another, and the European Union a third, and the differences between the three explain almost everything about why instant payments are ubiquitous in one place, legislated into existence in another, and still an optional extra in the third.

All figures, dates, limits and cut-off times in this chapter are accurate at the time of writing, August 2026. Instant payment systems are among the fastest-moving infrastructures in finance, and several of the numbers below changed within the last eighteen months.

The plain version#

Imagine three towns, each with a river running through it and no bridge.

In the first town, the council decides that a bridge is not something you leave to chance. It builds the bridge itself, out of public money. Then it passes a rule: every haulage company in town above a certain size must use the bridge, and none of them may charge ordinary residents a penny to walk across. Businesses pay a small toll, but people do not. The bridge is open every hour of every day.

In the second town, the council does not build anything. It says that bridges are a matter for private enterprise. Two consortia eventually build two bridges, some distance apart. Neither bridge connects to every road. A haulier who wants to reach every address in town must have an arrangement with both, and many hauliers have an arrangement with neither, because the old ferry still runs and the ferry is cheap and everybody is used to the ferry. Some residents cross the river every day and have never used either bridge.

In the third town, the council also does not build the bridge. But it does something else. It passes a law: within eighteen months, every haulage company must be able to receive goods from every other haulage company across the river, within ten seconds, and may not charge more for the fast crossing than for the slow one. It then says: here is a piece of publicly owned road at the centre of the town that all the bridges must connect to. How you build your end is your business. That the connection exists is not negotiable.

The first town is Brazil. The second is the United States. The third is the European Union. And the striking thing, when you look at the traffic figures, is how well the outcome matches the design.

Brazil, and what happens when the state builds the road#

Brazil’s system is called Pix. It is run by the country’s central bank, the Banco Central do Brasil, and it opened in November 2020.

Two things about Pix matter more than everything else combined.

The first is that if you are a bank or a payment company in Brazil with more than five hundred thousand active customer accounts, you are not asked whether you would like to join. You are required to. There was no launch period during which some banks had it and others did not, no competitive advantage for early movers, no customer ringing up to ask why her bank did not support it. From very close to the beginning, if you had a bank account in Brazil, you could send and receive.

The second is that banks are legally forbidden from charging ordinary people to use it. Not discouraged. Forbidden, in a written rule, with exceptions so narrow that they prove the point. A shop pays to receive. A person does not pay to send.

Then there is a third thing, less dramatic but arguably the cleverest. To pay someone in Brazil you do not need their bank details. You need a key: their mobile number, their email address, their national tax number, or a long random string their bank generated for them. The central bank runs a directory that maps every key to an account. You type the key, the directory returns the account, and it also returns the name on that account so you can check you are paying who you think you are paying. Businesses print a square barcode instead, which contains the same thing.

The numbers that follow from those three decisions are difficult to argue with. In 2021, its first full year, Pix carried 9.4 billion transactions. In 2024 it carried 63 billion, moving BRL 26.4 trillion, which the central bank’s own deputy governor put at around two and a half times Brazil’s entire economic output for that year. At the five-year mark, in November 2025, close to 170 million people were using it, in a country of roughly 215 million, along with more than 20 million businesses. As at 15 August 2026, 889 institutions took part.

For a shop, the cost of taking a Pix payment has been measured at about 0.22 per cent of the sale, against roughly 1.1 per cent for a debit card and 2.2 per cent for a credit card. That is not a marginal saving. That is the difference between a business model working and not working.

The United States, and what happens when nobody is in charge#

The United States has the largest economy in the world and, for most of the last decade, some of the slowest ordinary payments in the developed world. The reason is not technical incompetence. It is that the United States has roughly nine thousand banks and credit unions, no legal power to make them all join anything, and a payment system that was already good enough that nobody had to.

The old system is called ACH, for Automated Clearing House. It is a batch system, in the exact sense of the Bacs cycle familiar to British readers: files of instructions are collected, sorted, exchanged and applied, with settlement in between. It carries wages, benefits, tax refunds, mortgage payments and most business-to-business invoicing. It is enormous. In 2025 it moved 35.2 billion payments worth USD 93 trillion.

It also got faster. Since 2016 there has been a same-day option, with three deadlines during the working day. Submit a file by 10:30 in the morning, New York time, and it settles at one o’clock the same afternoon. Submit by 2:45 and it settles at five. Submit by 4:45 and it settles at six. Miss all three and you are on tomorrow. That is not instant, but for a great many purposes it is close enough, and it costs a fraction of a wire transfer, and this is precisely why the United States was slow to build anything faster. The pressure was never quite unbearable.

When faster things were built, they were built twice. A group of the largest banks built one, called RTP, which went live in 2017. The Federal Reserve built another, called FedNow, which went live in July 2023. Both run around the clock. Both are genuinely instant. And there are two of them, which means that a bank must join both to be sure it can reach everyone, and many banks have joined neither.

The volumes tell the story plainly. In its first full year, 2024, FedNow settled 1.5 million payments. In 2025 it settled 8.4 million. That is fast growth from a small base, and it is a rounding error next to 35.2 billion ACH payments. Meanwhile, Americans do in fact send each other money instantly, in very large quantities, using a service called Zelle, which is not a payment rail at all but a directory and a set of rules sitting on top of the banks. In 2024 Zelle carried 3.6 billion payments worth more than a trillion dollars. It solved the addressing problem, which is the problem people actually experience, and left the plumbing alone.

Europe, and what happens when you legislate#

The euro area had the technology from 2017 and almost nobody used it. A pan-European instant payment scheme existed, banks could join it, and joining was voluntary, so a large proportion of banks did not. Money crossing Europe in ten seconds was possible in theory and unavailable in practice, because the bank at the other end had not signed up.

So the European Union passed a law. Under the Instant Payments Regulation, from 9 January 2025 every bank in the euro area had to be able to receive an instant euro payment. From 9 October 2025 every bank had to be able to send one, and had to offer a free check of the payee’s name before you press the button, and was forbidden to charge more for the instant version than for the slow version.

The effect was immediate and visible in the settlement figures. The Eurosystem’s instant settlement platform, TIPS, handled 1.35 billion transactions in 2024. In 2025 it handled 2.47 billion, a rise of 82.5 per cent, with the sharpest growth in the closing months of the year, exactly as the sending obligation bit. On 1 December 2025 it settled 5,759,570 payments in a single day.

Note what Europe did not do. It did not build a consumer product; there is no European Pix app. It legislated the two things Brazil got by other means, universal reach and price parity, and let the market do the rest.

The pattern#

Put the three next to each other and the pattern is not subtle.

Central operation, or at least central compulsion. Mandated reach. A cap on what the customer can be charged. Addressing a human being can remember. Where those four are present, instant payments take over. Where any of them is missing, they do not.

Where the plain version stops being true#

The story above is true in outline and misleading in four specific places, and each of them matters if you are building anything.

“Instant” describes the customer’s experience, not the banks’#

When a Pix payment lands, the two banks involved have genuinely settled with each other, individually, transaction by transaction, in central bank money, at that instant. When a FedNow payment lands, the same is true. That is unusual.

For a great many fast payment systems it is not true at all. In the United Kingdom, a Faster Payment reaches the recipient in seconds, but the obligation between the two banks is settled afterwards, on a net basis, across accounts at the Bank of England, against balances the banks have prefunded. India’s UPI, which moves more transactions than any system on earth, likewise gives the user an instant result over a deferred net settlement arrangement underneath. The customer cannot tell the difference and should not have to. The bank’s treasurer very much can, because the two models put risk and liquidity in completely different places. Real-time gross settlement means every payment consumes central bank money at the moment it is made, all night and all weekend, which means a bank must fund an account it cannot easily top up out of hours. Deferred net settlement means the banks owe each other for a while, which is cheaper and introduces exposure between them.

So “instant” is a statement about the payee’s balance, not about the plumbing. Two systems that look identical to a customer can be entirely different animals.

Public ownership correlates with adoption; it does not cause it#

It is tempting to read Brazil’s numbers and conclude that the state should build the rail. The evidence for that is real but weaker than the story suggests.

The Bank for International Settlements ran the comparison properly: thirteen jurisdictions, monthly data from April 2001 to December 2023, just over a thousand country-month observations, controlling for country and year. Public ownership of a fast payment system is associated with an increase of about 1.8 per cent in transactions per head. That is a genuine effect and it is the smallest of the four they found. Allowing non-bank payment firms to participate is associated with 3.5 per cent. Each additional use case is worth about 2 per cent, and so is each additional cross-border connection.

And there are counter-examples in plain sight. Sweden’s Swish, one of the most thoroughly adopted instant payment products anywhere, is owned and operated entirely by private banks. India’s UPI is governed jointly by private banks and the central bank through a purpose-built body. What Brazil got from public ownership was not superior engineering. It was the political authority to compel participation and to forbid consumer fees, and those are the levers that moved. A private consortium with the same two powers would have had much the same result. A central bank without them would not.

“Free” means somebody else is paying#

Pix is free to individuals because a rule says it must be. It is not free to operate. The central bank charges participating institutions for the settlement infrastructure and for the addressing directory. Merchants pay to receive. The 0.22 per cent that a Brazilian shop pays is small next to a card fee, but it is not zero, and the entire consumer-facing economics of the system rest on it.

The same is true everywhere. The Federal Reserve charges banks 4.5 cents to originate a FedNow customer credit transfer, a dollar for a liquidity transfer and a cent for a request for payment, and in 2026 waived the twenty-five dollar monthly participation fee and rebated the first 2,500 credit transfers a month, which is a subsidy with a purpose. The Eurosystem charges 0.2 of a euro cent per instant payment, split between the sending and receiving bank. In Europe, the Instant Payments Regulation does not say instant payments must be free. It says they may not cost the customer more than a non-instant one, which in most euro-area retail banking means free in practice and is a different legal statement.

If you are modelling a business on a rail because “it’s free”, find out who is paying and whether they are still going to be paying in five years.

The United States is not behind; it is differently arranged#

The comparison that makes the United States look primitive is unit volume on instant rails, and by that measure it is far behind Brazil and India. But the United States has more instant payment infrastructure than most countries, not less. It has two independent real-time networks, one public and one private, both settling in central bank money, both with transaction ceilings far above anything in Europe or the United Kingdom: RTP raised its per-transaction limit to USD 10 million on 9 February 2025, and FedNow raised its network limit from USD 1 million to USD 10 million on 12 November 2025. British readers should sit with that. The Faster Payments scheme maximum is GBP 1 million.

What the United States lacks is not rails. It lacks the three things Brazil legislated: compulsory reach, a price cap at the consumer end, and one directory. It also has something Brazil did not have, which is a card industry so profitable and so entrenched that account-to-account payment at the point of sale has no natural sponsor. Nobody in the American payments industry makes more money when a customer pays a shop directly from their bank account. In Brazil, the central bank did, in the sense that it was pursuing a public objective rather than a margin.

And the volume comparison flatters the wrong number in any case. Same Day ACH alone carried 1.4 billion payments worth USD 3.9 trillion in 2025, up 16.7 per cent on the year, and Zelle carried 2 billion payments worth nearly USD 600 billion in the first half of 2025 alone. Americans are not waiting three days to pay each other. They are doing it over three different systems, none of which reaches everybody.

The technical version#

Brazil: Pix, SPI and DICT#

Pix is a payment scheme owned by the Banco Central do Brasil, established by BCB Resolution 1 of 12 August 2020 and launched in November 2020. It is important to separate three things that colloquial usage runs together.

Pix is the scheme: the rulebook, the branding, the user experience standards, the participation criteria.

SPI, the Sistema de Pagamentos Instantâneos, is the settlement infrastructure. It is a real-time gross settlement system built and operated by the BCB specifically for instant payments, running 24 hours a day, every day of the year. Direct participants hold dedicated Instant Payment accounts, known as PI accounts, at the central bank. Every Pix transaction settles individually across those accounts in central bank money at the moment it is made. Brazil already had a conventional RTGS system, the STR, when Pix was designed; the decision to build a separate platform rather than extend the existing one was a deliberate one about service levels and technology. Legally, SPI was built for Pix, and other payment schemes cannot connect to it for instant settlement.

DICT, the Diretório de Identificadores de Contas Transacionais, is the addressing directory. It maps aliases, known as Pix keys, to transaction accounts. Four key types are permitted: the CPF or CNPJ tax identifier, a mobile telephone number, an email address, and a randomly generated key, the EVP. Participants that are direct participants in SPI must access DICT directly.

Participation criteria are set out publicly. An institution licensed by the BCB that holds more than 500,000 active customer accounts, counting demand, savings and prepaid payment accounts together, is obliged to join, and once it crosses that threshold it has 90 days to apply for membership in the transaction account provider category. Smaller institutions may join voluntarily; institutions outside the regulatory perimeter may participate indirectly through a direct participant, subject to a contract, demonstrated technical capacity and a minimum equity requirement.

Pricing to end users is governed by BCB Resolution 19 of 1 October 2020, which is worth quoting because its structure is the model other jurisdictions keep reaching for. Article 3 prohibits an account-holding institution from charging a natural person customer, including individual entrepreneurs, for the remittance of funds for transfer or purchase, or for the receipt of funds for the purpose of transfer. The single paragraph carves out transactions conducted through in-person channels or the call centre where an electronic means was available. Article 4 permits charging a natural person for the receipt of funds for the purpose of purchase, which is how a sole trader taking payment for goods ends up paying, and permits charging legal entities for both sending and receiving.

The functional set has widened steadily: billing (Pix Cobrança), cash withdrawal at merchants (Pix Saque), cash back (Pix Troco), scheduled payments (Pix Agendado), contactless payment over NFC (Pix por Aproximação, launched 28 February 2025) and direct debit (Pix Automático, launched June 2025). Pix Automático is the structurally interesting one: any participant may offer it to corporate customers regardless of where the payer’s account is held, which breaks the historic pattern in which recurring collection was the preserve of a handful of large banks that the biller had to bank with. Fraud handling runs through a Special Return Mechanism, the MED, which permits funds to be blocked and returned despite transactions being nominally irrevocable, with a further revision announced by the BCB.

Adoption, on the central bank’s figures: 9.4 billion transactions worth around BRL 5 trillion in 2021; 63 billion transactions worth BRL 26.4 trillion in 2024; close to 170 million users and more than 20 million businesses at the five-year mark in November 2025. As at 15 August 2026 the BCB listed 889 participants, of which 560 were credit cooperatives and 173 payment institutions, which is itself a comment on how wide the participation rules are drawn.

The United States: ACH, Same Day ACH, FedNow and RTP#

The ACH network is governed by the Nacha Operating Rules and cleared by two operators: FedACH, run by the Federal Reserve, and the Electronic Payments Network, run by The Clearing House. It carries credits and debits in batch, with deferred net settlement across Federal Reserve accounts.

In 2025 the network processed 35.2 billion payments valued at USD 93 trillion, the thirteenth consecutive year in which total value rose by at least a trillion dollars. December 2025 was the highest-volume month recorded, at 3.22 billion payments, and November 2025 set an average daily record of 151 million payments.

Same Day ACH is an expedited service within the same network, not a separate rail. The FedACH processing schedule, effective 12 September 2022, defines three same-day windows.

Items not eligible for same-day settlement, including anything submitted in the later overnight windows, settle at 08:30 ET on the next banking day. The per-transaction limit for same-day eligibility is USD 1,000,000, effective 18 March 2022. Items above that limit submitted in a same-day window are not rejected; the operators assign them to the next available window for next-day settlement. Nacha has approved an increase of that limit to USD 10,000,000 with an effective date of 17 September 2027. In 2025, Same Day ACH carried 1.4 billion payments worth USD 3.9 trillion, up 16.7 per cent by volume and 21.4 per cent by value, averaging 5.8 million payments per day across the year and 7.8 million per day in December.

The FedNow Service launched in July 2023. It settles customer credit transfers individually in real time across participants’ Federal Reserve accounts, using ISO 20022 messages: pacs.008 for the credit transfer, pacs.004 for the return, pacs.009 for a liquidity management transfer, pain.013 for a request for payment.

Operating hours are worth stating precisely, because “24/7” conceals a business-day convention. FedNow operates on a 24-hour funds transfer business day every day of the week including weekends and Federal Reserve holidays. The business day rolls over at approximately 19:01 ET each calendar day, with the new day beginning immediately as the previous one ends and no interruption in processing. Liquidity Management Transfers, which let a participant move funds between its Federal Reserve account and the balances supporting instant payments, run 19:00 to 07:00 ET on weekdays and continuously at weekends and on Federal Reserve holidays.

Value limits: the default per-transaction limit is USD 100,000, which a participant may adjust, and the network limit rose from USD 1 million to USD 10 million effective 12 November 2025.

The 2026 fee schedule, effective 1 January 2026:

Settled volumes, from the Federal Reserve’s own statistics, last updated 6 July 2026:

The 2026 second quarter is the interesting line: volume up 83.2 per cent on the quarter while value was almost flat, and average payment value more than halved. That is the signature of a wholesale-ish network beginning to carry retail traffic. Participation was reported by the Federal Reserve at more than 1,500 financial institutions across all fifty states in autumn 2025, with 97 more institutions and four certified service providers joining in the first quarter of 2026.

The RTP network, operated by The Clearing House, launched in 2017 and predates FedNow by six years. It settles from a prefunded joint balance held at the Federal Reserve. Its per-transaction limit rose to USD 10 million effective 9 February 2025, a tenfold increase on the previous USD 1 million cap. At the end of 2024 The Clearing House reported the network averaging about one million payments a day, with some 285,000 businesses a month originating payments through participating institutions; by mid-2025 it reported more than 1,000 banks and credit unions live, a 51 per cent increase year on year.

Zelle is neither of these. It is an overlay operated by Early Warning Services, owned by a group of large banks, which resolves an email address or mobile number to an account and moves the resulting instruction over whichever underlying rail the participating banks use. It is the closest thing the United States has to a DICT. It carried 3.6 billion payments and more than USD 1 trillion in 2024, and 2 billion payments worth nearly USD 600 billion in the first half of 2025.

Why the United States is unusual, stated plainly. There is no legal power to compel roughly nine thousand depository institutions to join a payment system. There are two competing instant networks, so reach is a commercial question rather than a given. Addressing is by routing transit number and account number, credentials which are both hard to remember and, being reusable, unsafe to publish. Consumer pricing is unregulated. Cheques remain in use. And the existing batch rail is cheap, extremely reliable and, since 2016, same-day capable, which removed most of the urgency. The Bank for International Settlements notes a further constraint: FedNow, like Mexico’s CoDi and Sweden’s Swish, is open only to banks, in some cases for legislative reasons, and non-bank participation is one of the strongest correlates of adoption they measured.

Europe: T2, TIPS and the Instant Payments Regulation#

Terminology first, because the old name persists and is now wrong. TARGET2 was replaced on 20 March 2023 by T2, following consolidation with the securities settlement platform T2S. TARGET Services today comprise T2 for high-value and wholesale euro payments, T2S for securities, TIPS for instant payments and, since 16 June 2025, the Eurosystem Collateral Management System.

T2 is the euro area’s real-time gross settlement system: the direct analogue of CHAPS within the Bank of England’s RTGS, at continental scale. In 2025 it settled an average of 431,067 payments per day in euro, worth EUR 1,932.8 billion, across 255 operating days, for a yearly total of 111,910,103 transactions worth EUR 492,859.0 billion. That total exceeded the previous peak of 107,999,982 transactions set in 2024, and the value grew 6.3 per cent, ahead of euro-area nominal GDP growth of 3.8 per cent.

The business day matters for anyone building payment operations against it. The real-time settlement window concludes at 18:00 CET and may be interrupted by an optional maintenance window from 03:00 to 05:00 CET. The cut-off for customer payments is 17:00 CET; all other categories may settle until 18:00 CET. In 2025, 99.7 per cent of volume and 96.3 per cent of value had settled by the 17:00 customer cut-off, with the remaining payments in the final hour typically few in number and very large in value, as banks square positions and refinance on the money market.

TIPS, TARGET Instant Payment Settlement, launched in November 2018. It settles instant credit transfers in central bank money, 24 hours a day, 365 days a year. Participants hold a TIPS Dedicated Cash Account, a DCA, funded from their RTGS position; the alternative structure is an Ancillary System Technical Account, an ASTA, used where a clearing mechanism sits between the participant and the platform. Pricing is EUR 0.002 per instant payment, split equally between originator and beneficiary at EUR 0.001 each.

TIPS is a settlement platform, not a scheme. The scheme is SCT Inst, the SEPA Instant Credit Transfer scheme of the European Payments Council, launched in November 2017. Participation in TIPS, directly or indirectly, is mandatory for payment service providers that have signed up to SCT Inst and are reachable via T2. Until 9 October 2025 the scheme-level default maximum amount was EUR 100,000; that ceiling has been removed.

The Instant Payments Regulation, Regulation (EU) 2024/886, was adopted on 13 March 2024. It amends the SEPA Regulation, the cross-border payments regulation, the Settlement Finality Directive and PSD2. Its four substantive obligations are that a payment service provider offering ordinary credit transfers must also offer instant ones (Article 5a); that charges for instant credit transfers may not exceed charges for other credit transfers of corresponding type (Article 5b); that providers must offer a free verification of payee service before the payer authorises, returning a match, close match, no match or other result (Article 5c); and that providers must screen their own customer base against targeted financial sanctions at least daily rather than screening each transaction in flight (Article 5d).

The implementation timetable is staggered and precise.

The Eurosystem offers a verification of payee service built on solutions developed by Banco de Portugal and Latvijas Banka, conforming to the European Payments Council’s VoP scheme rulebook, effective 5 October 2025. Separately, from October 2025, non-bank payment service providers meeting the conditions in the TARGET Guideline may access TARGET Services directly, including T2 and TIPS, a change with the same intent as Brazil’s wide participation categories.

The effect on volumes was sharp and unambiguous. TIPS transactions rose from 1.35 billion in 2024 to 2.47 billion in 2025, growth of 82.5 per cent. Euro-denominated TIPS activity averaged 2,735,053 payments per day in 2025, worth EUR 2.2 billion, up 110.8 per cent from a daily average of 1,297,241 in 2024. December 2025 averaged 3,845,376 payments a day and the maximum single day was 5,759,570 on 1 December 2025. Germany overtook Italy as the largest contributor by volume, rising from 15.2 to 19.8 per cent of the total, and contributed 27.4 per cent of value.

The stress shows in the settlement ratio. Across 2025, 98.0 per cent of euro traffic settled by volume and 95.8 per cent by value. The value ratio fell to 93.1 per cent in October, coinciding with the sending obligation, as participants adapted to the shorter processing timeline, newly connected participants struggled with genuine 24/7 operation, and much larger payments arrived following the removal of the EUR 100,000 ceiling. By December it had recovered to 96.5 per cent. That is what a well-executed mandate looks like from the inside: a measurable two-month degradation, then recovery.

TARGET Services are also becoming multi-currency. The Danish krone joined T2 and TIPS in April 2025, with T2 settling an average of 11,425 krone payments per day worth DKK 1,641.9 billion in 2025, and the migration of Danish instant payments contributed materially to TIPS growth. Cross-currency capability was implemented in TIPS in June 2025, and the ECB is exploring interlinking TIPS with fast payment systems elsewhere.

What the successful systems have in common#

Set Brazil, the euro area, India, Thailand and the United Kingdom side by side and five design choices recur wherever adoption is high. They are not all necessary, but the more of them you have, the steeper the curve.

Central operation or central compulsion. Someone with authority has to be able to say that this exists and everyone connects to it. In Brazil that authority is ownership of the scheme and the infrastructure. In the euro area it is a regulation with dates in it. What matters is not the flag on the building; it is that reachability stops being a commercial decision.

Mandated participation, and therefore guaranteed reach. This is the single largest determinant, and it is upstream of everything else. A payment system a customer cannot rely on reaching the other party is not a payment system; it is a feature. Brazil compelled institutions above 500,000 active accounts. Europe compelled everyone, in two stages. The United States compelled nobody, and the outcome is exactly what you would predict.

Low or zero cost at the consumer end, established by rule. Pix forbids charging individuals. The Instant Payments Regulation forbids charging more than for a slow transfer. Neither leaves it to competition, because competition in retail banking does not reliably drive a per-transaction fee to zero, particularly where the incumbent alternative pays the bank a share of interchange.

Simple addressing, backed by a directory, with a name check. A payment instrument that requires the payer to obtain and correctly transcribe an account credential has a ceiling on its use. Aliases remove that ceiling. Brazil built DICT. India built the virtual payment address. Europe legislated verification of payee, which is a name check rather than an alias directory, and is a weaker version of the same idea. The United States has aliases, but they live in a private overlay rather than in the rail.

A settlement account that is open when the rail is open. An instant system running at three in the morning on a Sunday needs participants to hold funded positions at the central bank outside banking hours, and needs a way to top them up. The Eurosystem gives participants TIPS DCAs. The Federal Reserve gives them Liquidity Management Transfers with defined overnight and weekend windows. The BCB gives direct participants remunerated PI account balances, liquidity lines, and a window after the close of the RTGS system to move funds across. Skip this and the system works beautifully until a large payment arrives on a bank holiday.

There is a sixth, less often stated: one scheme, not several. Two competing instant networks in one currency area do not double the reach; they halve the certainty. The United States is the demonstration.

What that predicts for other markets#

For the United Kingdom, the diagnosis is uncomfortable, because Britain was early. Faster Payments arrived in 2008, before almost anyone, and being early froze in the design assumptions of 2008. The country has central operation, effective reach, and no consumer charge for retail Faster Payments in practice. What it does not have is alias addressing. A British payment is still addressed by six-digit sort code and eight-digit account number, exactly as it was in 1968, and Confirmation of Payee is a check on that credential rather than a replacement for it. Britain has the Europe answer to fraud without the Brazil answer to addressing. Open banking has partially filled the gap, but as an API layer that initiates payments over Faster Payments it inherits the addressing model rather than replacing it. Any future British improvement will have to solve identity resolution, and the scheme maximum of GBP 1 million per Faster Payments transaction will look increasingly conservative next to the USD 10 million now permitted on both American instant networks.

For the United States, expect convergence on reach without convergence on scheme. Both networks are growing, both have raised limits far beyond international norms, and the FedNow figures for the second quarter of 2026 show average payment value falling as volume climbs, which is retail adoption beginning. What will not happen without legislation is a single directory or a consumer price cap, so Zelle will continue to be the alias layer and card interchange will continue to fund the resistance at the point of sale.

For the euro area, the reach problem is solved and the product problem is not. TIPS settles the payments; it does not give a consumer a reason to use one. That gap is where private schemes layered on SCT Inst are competing, and their success or failure over the next few years will show whether legislated infrastructure is sufficient without a consumer-facing brand. Brazil’s answer suggests it is not: Pix succeeded partly because the central bank specified minimum user experience requirements and a single recognisable name rather than leaving presentation to 889 institutions.

For markets that have not yet built, the BIS evidence gives a clear order of operations. Admit non-bank payment service providers, because that was the strongest single correlate they measured. Launch with several use cases rather than one, because each additional use case is worth about as much as public ownership. Plan cross-border connections early. And, though the data were not available to test it directly, mandate participation, because that is the mechanism by which Brazil obtained in one year what voluntary schemes in Europe failed to obtain in seven.

The counter-pressure is fraud, and it is the one thing every fast system has had to retrofit. Irrevocable, instant, push payments to an account you cannot inspect are an excellent instrument for a confidence trick, which is why Brazil built the MED into an otherwise irrevocable scheme, why Europe legislated verification of payee alongside the sending obligation rather than afterwards, and why the United Kingdom has operated a mandatory reimbursement regime for authorised push payment fraud since October 2024. Speed and reversibility trade against each other, and the trade cannot be designed away, only priced and allocated. Every jurisdiction that has built a fast rail has ended up building a slow, human, contested process alongside it for the payments that should not have happened. That process is not a failure of the design. It is part of the design, and the ones that planned for it from the start have had a considerably easier time than the ones that did not.

43.98 Common wrong ideas#

Wrong: if a payment lands instantly, the two banks have settled. Right: that is true of Pix over SPI and of FedNow, and false of Faster Payments and UPI, which give an instant customer result over deferred net settlement.

Wrong: public ownership is what makes a fast payment system succeed. Right: the BIS comparison found public ownership the smallest of four effects at about 1.8 per cent, behind non-bank participation at 3.5 per cent; what Brazil actually used was the authority to compel participation and forbid consumer fees.

Wrong: Pix is free. Right: it is free to individuals because BCB Resolution 19 says it must be, while merchants pay about 0.22 per cent of the sale and participating institutions pay the central bank for settlement and for the directory.

Wrong: the Instant Payments Regulation made instant euro payments free. Right: it forbids charging more for an instant credit transfer than for a corresponding non-instant one, which usually means free in practice and is a different legal statement.

Wrong: the United States is technologically behind on instant payments. Right: it runs two independent real-time networks settling in central bank money, both now with USD 10 million ceilings against the Faster Payments maximum of GBP 1 million; what it lacks is compulsory reach, a consumer price cap and one directory.

Wrong: Americans wait days to pay each other. Right: Same Day ACH carried 1.4 billion payments worth USD 3.9 trillion in 2025 and Zelle carried 2 billion in the first half of that year; the traffic is spread across systems, none of which reaches everybody.

Wrong: Zelle is an instant payment rail. Right: it is an overlay operated by Early Warning Services that resolves an email address or mobile number to an account and rides whichever underlying rail the banks use; it is the closest thing the United States has to DICT.

Wrong: two competing instant networks give a country better coverage than one. Right: they do not double the reach, they halve the certainty, because a bank must join both to be sure of reaching everyone and many have joined neither.

Wrong: TARGET2 is the euro area’s real-time gross settlement system. Right: TARGET2 was replaced by T2 on 20 March 2023; TIPS is the instant settlement platform, and SCT Inst is the scheme that runs over it.

Wrong: an instant, irrevocable rail removes the need for a dispute process. Right: every jurisdiction that has built one has retrofitted a slow, human, contested process alongside it, from Brazil’s MED to Europe’s verification of payee to Britain’s mandatory reimbursement regime.

43.99 Chapter summary in 20 lines#

1. Moving a number from one ledger to another has been technically trivial for forty years, so the interesting questions are who decides, who must join and who pays.
2. Brazil built the rail publicly, the United States left it to the market, and the European Union legislated, and the traffic figures match the designs.
3. Pix is a scheme owned by the Banco Central do Brasil, launched in November 2020, settling individually and in real time across dedicated PI accounts on SPI, every hour of every day.
4. Institutions licensed by the BCB with more than 500,000 active accounts are obliged to join, so reachability was never a commercial decision.
5. BCB Resolution 19 forbids charging a natural person to send, while a shop pays about 0.22 per cent against roughly 1.1 per cent for debit and 2.2 per cent for credit cards.
6. DICT maps four kinds of key, tax identifier, mobile number, email address and a random string, to accounts, and returns the account name so the payer can check.
7. Those three decisions produced 63 billion transactions worth BRL 26.4 trillion in 2024 and close to 170 million users by the five-year mark.
8. The United States has roughly nine thousand depository institutions, no legal power to compel any of them to join anything, and a batch rail that was already good enough.
9. ACH carried 35.2 billion payments worth USD 93 trillion in 2025, and has had three same-day windows since 2016, which removed most of the urgency to build anything faster.
10. When faster rails were built they were built twice, RTP in 2017 and FedNow in 2023, so reach became a commercial question rather than a given.
11. FedNow settled 8.4 million payments in 2025, a rounding error next to ACH, while Zelle carried 3.6 billion in 2024 by solving the addressing problem and leaving the plumbing alone.
12. The euro area had SCT Inst from 2017 and almost nobody used it, because joining was voluntary and the bank at the other end had not joined.
13. Regulation (EU) 2024/886 required euro-area providers to receive instant payments from 9 January 2025 and to send them, check payee names and match prices from 9 October 2025.
14. TIPS volumes rose 82.5 per cent to 2.47 billion in 2025, with a measurable settlement-ratio dip in October and recovery by December, which is what a well-executed mandate looks like from the inside.
15. “Instant” is a statement about the payee’s balance, not about the plumbing, and real-time gross and deferred net settlement put risk and liquidity in entirely different places.
16. Nothing is free: Brazilian merchants, FedNow’s 4.5 cents per credit transfer and the Eurosystem’s 0.2 of a euro cent are all somebody paying for the rail.
17. The BIS evidence ranks non-bank participation, additional use cases and cross-border links above public ownership as correlates of adoption.
18. Five things recur wherever adoption is high: central operation or compulsion, mandated reach, a rule-based price cap at the consumer end, alias addressing with a name check, and a settlement account open whenever the rail is open, with a sixth being one scheme rather than several.
19. Britain was early, and being early froze in the assumptions of 2008: it has reach and no consumer charge, but still addresses payments by sort code and account number, with Confirmation of Payee as a check on that credential rather than a replacement for it.
20. Speed and reversibility trade against each other and the trade cannot be designed away, only priced and allocated, which is why the slow human fraud process beside a fast rail is part of the design rather than a failure of it.

Sources: Banco Central do Brasil, on Pix participation criteria, SPI and DICT, BCB Resolution 1 of 12 August 2020, BCB Resolution 19 of 1 October 2020, the Pix Statistics participant tables as at 15 August 2026, and the “Pix at 5” press article of November 2025; Federal Reserve Financial Services, on the FedNow Service volume and value statistics updated 6 July 2026, operating hours, the 2026 fee schedule, the transaction limit increase effective 12 November 2025, and the FedACH processing schedule effective 12 September 2022; Nacha, on 2025 ACH Network volume and value, Same Day ACH limits and the increase to USD 10 million effective 17 September 2027; The Clearing House, on the RTP network transaction limit effective 9 February 2025 and participation figures; Early Warning Services, on Zelle volumes; European Central Bank, TARGET Services Annual Report 2025, and the ECB pages on the Instant Payments Regulation, verification of payee and TIPS; European Payments Council, on the SCT Inst and VoP scheme rulebooks; Regulation (EU) 2024/886; Bank for International Settlements, “Fast payments: design and adoption”, BIS Quarterly Review, March 2024, and BIS Bulletin no 52 on Brazil’s Pix; International Monetary Fund, “Pix: Brazil’s Successful Instant Payment System”, 2023.