The Abstraction of Value

Money is humanity's most successful abstraction. A coin, a banknote, a number in a database, a cryptographic entry—each stands for value only because enough other people accept that it does. Nothing else about it is real.

A century ago that abstraction still had a body. Money was coins and bills. A transaction meant a clerk making an entry in a paper ledger. Checks moved between cities by rail and truck. Sending money abroad meant a sequence of letters between correspondent banks, and interest was computed by hand.

Today roughly $7.5 trillion changes hands daily in foreign exchange alone, with hundreds of billions more in equities.¹ Trading algorithms execute in microseconds. A card transaction is authorized across three continents in about a second. Money has become information and moves at close to the speed of light.

Mostly.

The qualifier is the point of this chapter. The interfaces are modern; the plumbing is not. International wires still take days. US equities settled two days after the trade until 2024. Bank rails stop for weekends and holidays. A customer sees an app that updates instantly and assumes the money moved instantly, when what actually happened was a promise entered into a queue.

That gap between the visible interface and the underlying settlement is where most of finance's remaining friction lives, and closing it is what the next chapter is about. This one covers how the gap opened.

2026 Snapshot — Global Finance

Scale

Global financial assets exceed $500 trillion across equities, bonds, deposits, and derivatives. Banks hold something over $180 trillion. Total payment flows—wholesale and retail combined—run into the quadrillions annually, though the figure varies enormously depending on what gets counted.² SWIFT alone carries more than 45 million messages a day across some 11,000 institutions in over 200 countries.³

Infrastructure

The rails are a museum of the eras that built them. Visa and Mastercard descend from card systems designed in the 1950s and 1960s. SWIFT dates to 1973 and ACH to 1974. Fedwire and CHIPS handle US wholesale settlement, moving roughly $5 trillion and $1.8 trillion daily respectively. FedNow and RTP provide real-time retail settlement and are recent additions to a stack that mostly is not.

US equities moved to T+1 settlement in May 2024, an unglamorous change that meaningfully reduced counterparty risk; much of the world remains on T+2.⁶ ACH remains batch-processed, with same-day service available at a premium rather than as a default.

Digitization

Cards account for more than 80 percent of in-person retail payment in developed countries. Mobile payment is effectively universal in China through Alipay and WeChat Pay and growing everywhere else. Online banking is standard, and for younger demographics the mobile app is the bank.

Fintech has unbundled the traditional institution. Payments, lending, investing, and insurance have each been attacked separately by firms that took one profitable function and did it better, leaving incumbents with the expensive infrastructure and the regulatory burden.

What Is Still Broken

Four problems define the current system, and each is a target in the next chapter.

Settlement friction. Time between trade and settlement means counterparty risk and capital held against it. Every day of delay is capital that cannot be deployed.

Cross-border cost. Remittances average around 6.2 percent globally against a UN Sustainable Development Goal target of 3 percent—a gap representing tens of billions annually taken from people sending money home, typically the least able to absorb it.⁵

Exclusion. Roughly 1.4 billion adults have no bank account, concentrated in the developing world and disproportionately women.⁴ The binding constraint is usually identity documentation rather than poverty; without documents there is no account, without an account no credit history, and without credit history no access to capital.

Fraud. Card fraud losses exceeded $30 billion globally in 2023 and are growing with digital volume.⁸ Social engineering rather than technical compromise remains the dominant vector, which is why each technical improvement produces a smaller gain than expected.

Notable Players

The global banks—JPMorgan, Bank of America, Citigroup, HSBC, BNP Paribas, ICBC—remain the system's balance sheet. Their advantage is not technology but regulatory license and deposit access, which is precisely what fintech competitors cannot replicate and eventually have to rent.

The card networks, Visa and Mastercard, occupy the most durable position in finance: a two-sided network with sixty years of merchant acceptance, earning a fraction of nearly every retail transaction in the developed world. Multiple well-funded efforts to displace them have failed.

Real-time payment systems are increasingly public infrastructure rather than private product. India's UPI, Brazil's PIX, the UK's Faster Payments, SEPA Instant, and FedNow all reflect a view that instant payment is a utility. Notably, the countries that moved fastest are not the wealthiest—they are the ones without entrenched card economics to protect.

Fintech clusters by function: Stripe, Adyen, Block, and PayPal in payments; Nubank, Revolut, N26, and Chime in retail banking; Affirm and Klarna in credit; Robinhood and Wealthfront in investing; Plaid and Marqeta in the infrastructure layer that lets everyone else exist.

Digital assets have consolidated around exchanges (Coinbase, Binance, Kraken), stablecoin issuers (Tether, Circle), and institutional custody (Fireblocks, Anchorage, BitGo). Speculation still dominates the volume, but the payment use case has become real.

Open banking deserves mention as a regulatory rather than corporate actor. The UK's 2018 framework and the EU's PSD2 mandated that banks expose customer data through APIs to authorized third parties.⁹ This was the single most consequential structural change of the period, because it converted the incumbent's data monopoly into a shared resource by law rather than by competition.

The Century in Finance

The Banking Century, 1900–1970

Banking was physical. Branch networks, tellers, paper ledgers, and checks that cleared by being physically transported. A check written in California and deposited in New York took days to clear because a piece of paper had to make the trip.

The institutional architecture was built in response to failure. The Federal Reserve was established in 1913 after repeated banking panics. Glass-Steagall separated commercial from investment banking in 1933 after the Crash. Deposit insurance arrived the same year. The pattern—crisis, then structural reform—recurs throughout the century and is worth noting, because it is still how financial regulation gets made.

Consumer credit as a mass product began mid-century with Diners Club in 1950 and BankAmericard in 1958, the latter eventually becoming Visa.

The Electronic Revolution, 1970–2000

The ATM arrived in 1969 and did something more significant than dispense cash: it decoupled banking from banking hours, establishing the expectation of self-service that every subsequent development built on.

SWIFT was founded in 1973 to standardize international messaging between banks. Its design decision was consequential and is still constraining: SWIFT moves messages, not money. It tells correspondent banks to make entries. That architecture is why international transfers remain slow half a century later—the messaging was solved and the settlement never was.

NASDAQ launched as an electronic quotation system in 1971 and began the long displacement of floor trading. Glass-Steagall was repealed in 1999, permitting the consolidation that produced the universal banks whose failure would matter enormously nine years later. The first internet banks appeared in the mid-1990s.

The Fintech Era, 2000–2020

PayPal made internet commerce possible by solving payment for parties who had no reason to trust each other.

The 2008 financial crisis is the hinge of the modern period, and its most durable effect was not regulatory but reputational. Bailouts, foreclosures, and the absence of criminal accountability produced a lasting collapse in institutional trust, and that collapse is the necessary context for what followed. Bitcoin's genesis block, mined in January 2009, contains a reference to bank bailouts. Cryptocurrency is not comprehensible as a purely technical development; it was a political response to a specific event.

Smartphones then changed payment behavior faster than anyone anticipated, and they did so first in markets with weak card infrastructure. China leapfrogged directly from cash to QR-code mobile payment, achieving penetration the card-saturated West still has not matched. Absent legacy infrastructure turned out to be an advantage.

Neobanks unbundled retail banking. Open banking regulation forced incumbents to expose their data. Embedded finance made payment a feature of other products rather than a destination.

The Current Period, 2020–Present

FedNow launched in July 2023, bringing instant payments to the US years after the UK, EU, India, and Brazil.⁷ Adoption has been steady rather than rapid, because a payment network is worth using only when the other party is also on it.

Stablecoins scaled past $150 billion in circulation, with Tether dominant and Circle's USDC positioned as the regulated alternative.¹⁰ Most volume remains trading-related, but remittance and payroll use has become genuine—these are dollar-denominated instruments serving people who want dollars and cannot easily get a dollar bank account.

Buy-now-pay-later reintroduced installment credit to a generation that had rejected credit cards, largely by not calling it credit.

Fraud changed character. Voice cloning and synthetic media made social engineering dramatically more effective, and the fraud growth curve turned upward accordingly.

And in March 2023, Silicon Valley Bank failed in roughly thirty-six hours. This deserves emphasis as the most important financial event of the decade so far, because of the mechanism rather than the size. Depositors coordinated through group chats and withdrew through mobile apps—$42 billion attempted in a single day. Every prior model of bank runs assumed the physical friction of queuing at a branch. Instant payments and social coordination removed that friction, and no one had updated the assumptions that friction was quietly supporting.

How Money Actually Moves

The gap between apparent and actual settlement is worth making concrete.

A card payment authorizes in about a second and settles in days. The authorization is a promise; the money moves later through a chain running merchant to acquirer to network to issuer and back. Interchange fees are extracted along the way, which is why the networks are among the most profitable businesses in existence.

ACH collects transactions into batch files, processes them overnight, and settles the next day. It is cheap, high-volume, and structurally incapable of being instant.

A wire settles in real time, gross, and expensively—used for large amounts where speed justifies cost.

Peer-to-peer apps feel instant and mostly are not. Venmo and Cash App settle instantly within their own network by moving entries in their internal ledger, then reconcile with the banking system through ACH afterward. The instantaneity is an accounting convention layered over slow rails.

Cross-border is worst. A transfer routes from the sending bank to its correspondent in the destination country to the receiving bank, with each hop adding a day, a fee, and a foreign exchange spread. SWIFT coordinates but does not settle. The fees are frequently opaque, embedded in the exchange rate rather than disclosed as charges.

Modern Bottlenecks

Four constraints define what the next chapter has to overcome.

Settlement timing ties up capital and creates counterparty risk for no reason other than that the systems were built when reconciliation took time.

Cross-border complexity persists because compliance is national while money is not. Every jurisdiction requires its own know-your-customer and anti-money-laundering process, and each intermediary performs its own redundantly.

Identity is the binding constraint on inclusion. The 1.4 billion unbanked are largely unbanked because they cannot prove who they are to a standard the compliance regime accepts—which makes financial inclusion a documentation problem before it is a financial one.

Fraud improves faster than prevention, because the attack surface is human. Authentication technology has advanced considerably; the person who can be talked into authorizing the transfer themselves has not.

The AI Transition Begins

AI entered finance earlier than most sectors, because finance had labeled data, immediate feedback, and a direct monetary value on accuracy.

Fraud detection is the mature application. Nearly every significant institution now scores transactions in real time against behavioral baselines, and network analysis identifies coordinated fraud rings that per-transaction rules miss. It works, imperfectly, in a permanent arms race against attackers using the same tools.

Credit underwriting is where the stakes are highest. AI can assess creditworthiness from alternative data—transaction history, rent and utility payment, behavioral signals—which genuinely extends credit to people with no conventional file. It can also encode historical discrimination with a statistical justification attached, and the fair-lending problem here is unsolved rather than merely difficult.

Compliance absorbs enormous headcount, and AI monitoring for suspicious activity is displacing it. This is quietly one of the largest white-collar automation stories in progress, and it attracts little attention because nobody finds anti-money-laundering operations interesting.

Customer service and document processing follow the pattern described throughout this book: routine volume automated, exceptions escalated, headcount reduced at the entry level first.

Second-Order Impacts

Deposits became mobile, and banking stability assumptions did not update. The regulatory framework for liquidity was calibrated to an era when withdrawing money required effort. SVB demonstrated that this calibration is obsolete, and the framework has not been fully rewritten.

The unbundling shifted risk without shifting responsibility. Fintechs took the profitable customer-facing functions and left deposit-holding, regulatory compliance, and systemic obligation with banks—frequently by renting a small partner bank's charter. When these arrangements fail, the failure lands on customers who believed they were banking with the app.

Payment data became more valuable than payment fees. Knowing what people buy is worth more than the fraction charged for processing it, which is why technology firms entered payments without needing payments to be profitable.

Financial infrastructure became geopolitical. SWIFT's use to exclude states from the dollar system converted a technical utility into an instrument of statecraft, and accelerated work on alternative rails by every country that noticed.

Conclusion

Money is humanity's oldest information system, and for most of history the information had to travel on a physical token. The past century converted the token into a record, and the record into a database entry.

What did not get converted is the settlement layer underneath. The interfaces became instant while the plumbing stayed batch, and the resulting gap is where the industry's remaining rents are collected: in float, in interchange, in foreign exchange spreads, and in the days during which money exists in neither account.

The technology to close that gap exists now. Real-time rails are live in most major economies. Programmable settlement works. AI handles the risk and compliance functions that previously required the delay.

What has not been settled is who captures the value released when friction disappears. Every friction point in the current system is someone's revenue line, and the institutions collecting it are the institutions best positioned to slow the transition.

The next chapter takes up what happens when they can no longer slow it.

Endnotes — Chapter 53

1. BIS Triennial Survey (2022) reports $7.5 trillion in daily foreign exchange turnover; global equity markets add hundreds of billions more.
2. Global payment flows are estimated in the quadrillions annually including wholesale and retail; precise figures vary substantially by definition and are not directly comparable across sources.
3. SWIFT handles more than 45 million messages daily across over 11,000 financial institutions in more than 200 countries. It carries instructions, not funds.
4. World Bank Global Findex estimates 1.4 billion adults remain unbanked (2021), concentrated in developing economies, with women disproportionately affected. Lack of identity documentation is a leading stated reason.
5. World Bank Remittance Prices Worldwide: global average remittance cost was 6.2 percent (2023) against a Sustainable Development Goal target of 3 percent.
6. T+1 settlement was implemented for US equities in May 2024, halving the settlement window and the associated counterparty exposure; T+2 remains standard in many markets.
7. FedNow launched in July 2023, providing 24/7 instant payment settlement in the US—years after equivalent systems in the UK, EU, India, and Brazil. Bank adoption has grown steadily but is not universal.
8. Card fraud losses exceeded $30 billion globally in 2023 and continue to grow with digital transaction volume; social engineering rather than technical compromise remains the dominant vector.
9. Open banking frameworks—UK Open Banking (2018) and PSD2 in the EU (2018), with variants elsewhere—require banks to provide third-party access to customer data via APIs, converting an incumbent data advantage into shared infrastructure by regulation.
10. Stablecoin market capitalization exceeded $150 billion (2024), with Tether (USDT) largest and Circle (USDC) second. Most volume is trading-related, though payment and remittance use is growing.