Finance, fintech & payments

Finance, fintech & payments

Industry ID: 07 | Slug: finance-fintech | Researched: 2026-09-15 | Analyst: agent

1. Definition and boundaries

In scope: banking (retail, commercial, universal, community); capital markets and investment banking; payments and card networks; consumer and commercial lending, including BNPL; insurance and insurtech, including reinsurance; wealth and asset management, including private credit and alternative asset management; RegTech and compliance technology; embedded finance and banking-as-a-service.

Explicitly out of scope: crypto and digital assets, which belong to sector 21. The boundary is regulatory perimeter, not technology. A stablecoin issued by a permitted payment stablecoin issuer under the GENIUS Act, held as an insured deposit, or settled between banks is in scope here. A token traded on an offshore exchange is sector 21's. Kinexys and Citi Token Services sit here because they are deposit liabilities of insured banks recorded on a ledger; USDT's secondary-market trading does not.

Named boundary disputes:

1. Stablecoins. Circle is simultaneously a crypto company and a conditionally chartered national trust bank. This dossier covers it as a regulated issuer under T-07-04 and T-07-20; sector 21 should cover its token's market behaviour. Double-counting here is likely and should be reconciled at aggregation.
2. Private credit. Sits between finance and private equity. Because the 2026 story is its interconnection with banks (T-07-03), it is treated as core here rather than as an asset-management adjunct.
3. Agentic commerce. The payment rail is ours; the AI agent and the merchant's catalogue are not. T-07-09 deliberately covers only the credential, settlement and liability layer.
4. Insurance. Included in full, including reinsurance and catastrophe risk. Climate modelling itself belongs to a climate sector; the pricing consequence is ours.
5. Fintech infrastructure that is really AI infrastructure. Stripe's $7bn+ acquisition of OpenRouter is counted here as a fintech M&A event, but the asset is AI infrastructure. This is precisely the kind of classification choice that makes fintech funding totals non-comparable between trackers.

2. Subcategories

3. Market structure

Concentration: the sector is not one structure but four.

• Banking: regulated oligopoly with a long tail. 4,278 FDIC-insured institutions as of Q1 2026, but a handful of GSIBs dominate the aggregates. Barriers to entry are regulatory (a charter) rather than technological, which is exactly why charter access — OCC trust charters for stablecoin issuers, sponsor-bank arrangements for fintechs — is the contested asset.
• Card networks: duopoly. Visa and Mastercard. Visa's fiscal Q3 2026 (quarter ended 2026-06-30) produced $11.6bn net revenue on 71.7 billion processed transactions, with revenue growing 14% against payments volume at 10% — pricing power in plain sight.
• Private credit: fragmented at the manager level, concentrating fast at the top. Five firms (Apollo, Ares, Blackstone, Carlyle, KKR) hold $1.5tn of perpetual capital, roughly 40% of their combined AUM.
• Financial market infrastructure: regulated utility. The Clearing House clears roughly $2tn daily through CHIPS.

Where margin actually sits. Not in the balance sheet. Aggregate US bank ROA was 1.37% in Q2 2026 — a good quarter by post-crisis standards and still a thin return on assets. The high-margin positions are (a) the network toll, (b) the asset-management fee on perpetual capital, and (c) the data-and-decisioning layer between them. Banks increasingly monetise by financing the balance-sheet risk-taker rather than being it: loans to non-depository financial institutions grew 22.4% year-on-year in Q2 2026 against 6.8% total loan growth.

Barriers to entry. A charter; capital; a distribution relationship; and, increasingly, the compliance capacity to operate across fifty state regimes rather than one federal one. That last barrier is new in 2026 and favours scale.

Who has pricing power. Card networks, systemically important infrastructure operators, the largest alternative managers, and — cyclically, right now — reinsurers' customers rather than reinsurers, since five consecutive quarters without a $10bn loss have handed pricing to cedants.

Market-size figures. Stated only with modeller and date:

4. Who matters

Leading companies. JPMorgan Chase (jpmorganchase.com), Bank of America (bankofamerica.com), Citigroup (citigroup.com), Goldman Sachs (goldmansachs.com), Capital One (capitalone.com), Visa (usa.visa.com), Mastercard (mastercard.com), BlackRock (blackrock.com), Block (block.xyz), Affirm (affirm.com), Klarna (klarna.com), Circle (circle.com), Chime (chime.com), Swiss Re (swissre.com).

Notable startups. Stripe (stripe.com, $159bn Feb 2026 tender), Ramp (ramp.com, $44bn Jun 2026), Brex (brex.com, being acquired for $5.15bn), Plaid (plaid.com), Revolut (revolut.com, $75bn Nov 2025 secondary), Taktile (taktile.com, $110m Series C Jun 2026), Flutterwave (flutterwave.com, $3.2bn Jun 2026), BioCatch (biocatch.com, $2.4bn Visa deal), Securitize (securitize.io).

Active investors. Goldman Sachs Alternatives (led Taktile's Series C); Blackstone, Blue Owl Capital, Ares, KKR and Carlyle on the private credit side. Note: this research did not verify individual venture firms' 2026 fintech deal activity and deliberately does not list unverified names.

Platforms and standards bodies. The Clearing House (theclearinghouse.org) — CHIPS, RTP, and operator of the planned 17-bank tokenised deposit network; Agentic Commerce Protocol (Stripe/OpenAI, open standard); NPCI (npci.org.in) for UPI; Banco Central do Brasil for Pix; ECB digital euro scheme rulebook (v0.91, July 2026).

Regulators. Federal Reserve Board (federalreserve.gov); FDIC (fdic.gov); OCC (occ.gov); CFPB (consumerfinance.gov); SEC (sec.gov); FinCEN and OFAC (Treasury); state regulators, notably NYDFS (dfs.ny.gov) and California DFPI; Financial Stability Board (fsb.org); Bank of England FPC (bankofengland.co.uk); ECB (ecb.europa.eu); FCA (fca.org.uk).

Research institutions. Federal Reserve Bank of New York Center for Microeconomic Data (newyorkfed.org/microeconomics); Federal Reserve Bank of Richmond; Brookings Institution (brookings.edu); Swiss Re Institute; US Government Accountability Office (gao.gov); BIS (bis.org).

Trade organisations. Bank Policy Institute (bpi.com); The Clearing House Association; Consumer Bankers Association (consumerbankers.com); American Bankers Association (aba.com); Independent Community Bankers of America (icba.org).

Consumer and civil-society groups. State attorneys general acting in a de facto consumer-protection role (see §8); the National Association of Attorneys General (naag.org). This research did not verify the 2026 activity of specific consumer advocacy organisations — recorded as a gap in §13.

5. Products, business models, technologies, customers

Major products. Deposit accounts and payment services; commercial, consumer and mortgage credit; card issuing and acquiring; underwriting and market-making; property, casualty and life insurance and reinsurance; asset management mandates and funds; direct lending facilities; compliance and fraud technology.

How money is actually made today, and how that is changing.

The three classic engines are net interest margin, fee and interchange, and fee on assets. All three are shifting:

• Net interest margin has stopped compressing rather than recovered — 3.31% in Q1 2026, 3.32% in Q2. Because there is no rate move to pass through, deposit repricing has stalled; Bank of America's rate paid fell 16bp to 1.47% with management stating that with rates static there is "no great deal of impetus" to change it. The growth now comes from volume, and specifically from lending to financial intermediaries rather than to end borrowers.
• Fee and interchange is under two simultaneous pressures: account-to-account rails at the low-value end, and the prospect of agents intermediating the checkout. Visa's response is instructive — revenue growing 14% against 10% payments volume, "other revenue" up 45%, and a $2.4bn acquisition of a fraud company. The network is monetising services and trust, not just the toll.
• Fee on assets is migrating from public to private markets and from institutional to retail wrappers. $607bn now sits in 567 evergreen vehicles.

Technologies that matter. Permissioned distributed ledgers for deposit tokenisation (Kinexys at $7bn daily); tokenised money-market fund share classes as collateral; agentic payment credentials (Shared Payment Tokens); behavioural biometrics and fraud ML; decision engines for credit and fraud (Taktile); large language models deployed internally at scale but without disclosed financial results (see T-07-19).

Major customer segments and what they buy on.

• Retail consumers buy on convenience, price and trust, in that order, and switch rarely. BNPL's growth is a convenience-and-access story; 45% of originations went to deep-subprime borrowers.
• Small and mid-sized businesses buy on speed of decision and integration with their existing software — which is why spend-management platforms (Brex, Ramp) command premium valuations.
• Corporates and treasurers buy on settlement certainty, working-capital efficiency and 24/7 availability — the actual demand driving tokenised deposits.
• Institutional allocators buy on access, yield and, in private markets, on manager brand.
• Merchants buy on conversion and cost of acceptance, the two axes on which A2A and agentic rails compete.

6. Geography

Capital concentrates in the US. The US took $15bn of the $28.6bn global H1 2026 fintech funding, 52% of the total; the UK $2.7bn and India $1.9bn. This mirrors the macro brief's Americas-at-66% figure for all-sector VC. European fintech remains the structurally weak leg despite hosting the highest-valued private fintech (Revolut, $75bn, 65 million customers, $1bn annualised revenue).

Regulation concentrates in three blocs with divergent direction. This is the single most important geographic fact of 2026: the US and the UK are moving in opposite directions on the same question. The CFPB's Section 1033 open banking rule is enjoined and being rewritten to permit data-access fees; the FCA published an open finance roadmap in April 2026 with Treasury legislation expected to expand its rule-setting powers. A firm building a data-access-dependent business now faces a liberalising UK and a retreating US.

Payments innovation concentrates outside the US. The two most consequential retail payment systems of the last decade are Brazilian and Indian, both state-built.

Non-US market covered in depth — Brazil and India (regional source: Atlantic Council GeoEconomics Center, 2025-04-21). Pix has 140 million users, roughly 65% of Brazil's population, and processed 15.4 billion transactions in Q2 2024. UPI has roughly 350 million users and carries over 75% of India's retail digital payments. The finding that matters is counter-intuitive: cards grew faster after the instant rails arrived, not slower. Brazilian credit card growth accelerated from 12.7% annually (2018–20) to 31.7% (2020–22); Indian credit cards moved from −7.3% annually to a 24.2% CAGR. The instant rail formalised cash and pulled unbanked users into the system; cards then moved up-market into credit. A second regional finding cuts the other way: Google Pay and PhonePe together take over 80% of UPI transactions, so a public rail did not prevent front-end concentration — it relocated the rent.

Europe (regional source: ECB, digital euro project, rulebook v0.91, July 2026). The euro area is attempting the same thing by central-bank fiat: pilot in 2027, potential first issuance 2029, contingent on EU legislation passing during 2026, with bank-side implementation cost estimated at €4–5.8bn.

UK (regional source: Bank of England Financial Stability Report, 2026-07-07). The FPC has been more specific than US authorities on non-bank leverage — hedge fund equity leverage balances up roughly 40% year-on-year — and on AI-related valuation risk, naming circular financing among technology companies and maturity mismatches between short-lived AI chips and longer-duration debt.

Insurance demand concentrates where risk is repricing. GAO found US premiums rose 3% nationally over 2019–2024 but 25%+ in southern coastal regions, with Florida, Louisiana and Oklahoma carrying the greatest burden relative to household income.

7. Historical trend patterns

Ten to twenty-five years of this sector's cycles, with specific past false positives:

The recurring cycle. Credit expands into a new channel that sits outside the prevailing regulatory perimeter → the channel is celebrated as innovation → a credit event reveals that the risk never left the system, only moved → the perimeter is redrawn. Subprime mortgage securitisation (2003–08), non-bank mortgage servicing (2010s), marketplace lending (2014–16), crypto lending (2020–22), and now private credit.

Prior hype waves and how they resolved:

• Marketplace/P2P lending (2013–16). Framed as disintermediating banks. Resolved by the discovery that origination without a balance sheet is a fee business dependent on someone else's funding: LendingClub's 2016 crisis, Prosper's retrenchment, and eventual conversion to bank ownership. False positive: "banks will be disintermediated." What actually happened is what is happening again in 2026 — banks financed the disintermediators.
• Blockchain for interbank settlement (2015–18). Consortium after consortium (R3, Utility Settlement Coin, various trade-finance ventures) produced pilots and few production systems. False positive: "consortium DLT will replace correspondent banking by 2020." Relevant now: the 17-bank Clearing House tokenised deposit network targeting H1 2027 is the same structure that failed before. What is different this time is that a single-bank version already works at scale ($7bn daily on Kinexys) and that there is a competitive threat with a real float. Those are genuine differences. They are not proof.
• Insurtech (2018–21). Lemonade, Root, Hippo listed at high multiples on the thesis that technology would beat underwriting discipline. It did not. False positive: "loss ratios are a technology problem." The 2025 listing of Slide Insurance at $17, trading at $16.29 by January 2026, is a much more sober valuation of the category.
• Neobanks as bank replacements (2018–22). Resolved into: real customer acquisition, real primary-account share in some cases, and public-market valuations far below private peaks. Chime listed at $27 in June 2025, spiked 59%, and sat at $26.07 seven months later.
• BaaS as infrastructure play (2020–23). Ended with Synapse's failure and a supervisory response that now accounts for over 25% of FDIC and over 20% of OCC formal enforcement actions. False positive: "charter access is a commodity."
• "Embedded finance will be 10% of transactions." Still a forecast, still unattributed to a published methodology, still repeated.

What has actually cycled, reliably: the reinsurance underwriting cycle (currently softening into a rising loss trend — see T-07-16), the bank capital regulation pendulum (currently loosening for the first time since 2008 — T-07-06), and the credit cycle itself.

The lesson for 2026. Two of this dossier's twenty trends are classified as overhyped and three as cooling or reversing, and in every case the disconfirming evidence was available in primary sources. The pattern to watch for is a claim whose only support is a participant with a commercial interest in it being true.

8. What is changing now (as of 2026-09-15)

Grounded in the macro brief, five things have changed:

1. The rate assumption inverted. The macro brief records the FOMC holding at 3.50–3.75% on 2026-07-29 with three dissents in favour of a hike. Bank plans built in 2024–25 on an easing path are now mispositioned. Concretely: Bank of America raised its 2026 NII guidance while noting the forward curve had gone from two cuts to zero. The sector-wide result is a margin plateau (3.31% → 3.32%) with growth coming from volume. Every trend thesis in this sector that assumes cheap capital has been stress-tested against this and the ones that failed are marked as such.

2. Capital requirements are falling. The 2026-03-19 joint proposals imply a roughly 2.4% CET1 reduction for the largest banks and up to 7.8% for smaller non-CBLR banks, on top of an eSLR change effective 2026-04-01. Q2 2026 already shows Tier 1 down 17bp and an 87% dividend payout ratio. Capital is being released into a credit cycle whose direction is uncertain.

3. Risk is visibly circulating between banks and non-banks. NDFI lending up 22.4% and securities financing up 29.7% year-on-year, against 6.8% total loan growth. Simultaneously private credit is producing its first coordinated stress indicators: redemptions at 4.8% of NAV, bad PIK at 6.4% of loans, roughly 40% of borrowers with negative free cash flow. The FSB can identify roughly $220bn of the bank-to-fund exposure; commercial estimates exceed $500bn.

4. The federal consumer-finance perimeter has been withdrawn and states have moved in. CFPB headcount from ~1,700 to ~1,300, budget request below half of prior years, rescission agenda across six major rules, litigation docket moving to DoJ. New York's FAIR Act, seven states coordinating on AI enforcement, seven states writing to BNPL lenders.

5. The AI story in this sector is an input story, not yet an output story. Every large bank reports use-case counts; none has disclosed an attributable financial result. Meanwhile the macro brief's AI-driven financial-stability concerns show up here in a specific, under-noticed form: roughly a fifth of direct lending is secured against software revenue, at the same moment the FPC is warning about AI-related valuation and financing risk.

One thing that did not change on schedule, consistent with the macro brief's warning about deadline slippage: the EU AI Act's high-risk employment obligations were deferred from 2026-08-02 to 2027-12-02. Vendor compliance marketing built on the August 2026 date is stale.

9. The five lists

Five most important current trends

1. T-07-01 — Net interest margin plateau replaces the expected easing-cycle squeeze
2. T-07-02 — Private credit's first genuine stress test: gates, PIK and shadow defaults
3. T-07-06 — US prudential deregulation: capital requirements fall for the first time since 2008
4. T-07-04 — GENIUS Act rulemaking redraws the bank-stablecoin boundary
5. T-07-07 — Consumer-finance enforcement migrates from the CFPB to the states

Five fastest-growing signals (by rate of change, not importance)

1. T-07-03 — NDFI lending, +22.4% YoY against 6.8% total loan growth
2. T-07-05 — Tokenised deposits: Kinexys from $5bn to $7bn daily in two months
3. T-07-04 — GENIUS rulemaking: six-plus federal proposals across five agencies in nine months
4. T-07-11 — Private markets retailization: evergreen private credit AUM +45% YoY
5. T-07-09 — Agentic commerce rails — fastest-growing in capability, not in usage; adoption remains unmeasured

Five trends most likely to affect businesses

1. T-07-06 — Capital rules determine credit availability and pricing for every corporate borrower
2. T-07-02 — Private credit stress determines refinancing availability for mid-market companies
3. T-07-07 — Fifty-state compliance replaces single-federal compliance for consumer-facing firms
4. T-07-05 — Tokenised deposits change corporate treasury settlement
5. T-07-17 — Metered data access changes the unit economics of every data-dependent fintech

Five trends most likely to affect consumers

1. T-07-08 — BNPL as interest-bearing credit, concentrated in deep-subprime borrowers
2. T-07-16 — Property insurance affordability, with premiums up 25%+ in coastal regions
3. T-07-07 — Uneven consumer protection depending on which state you live in
4. T-07-13 — Account-to-account rails and, in the EU, a digital euro
5. T-07-11 — Private markets arriving in retirement accounts

Overlaps, stated explicitly: T-07-07 appears on both the business and consumer lists because regulatory fragmentation raises cost for firms and lowers protection consistency for individuals — the same mechanism, two incidences. T-07-04 appears on both the "most important" and "fastest-growing" lists because rulemaking velocity and strategic consequence coincide here, which is unusual. T-07-02 and T-07-03 are two faces of one phenomenon and should be read together; they are separated because the observable data differs (fund disclosure vs bank call reports).

10. Overhyped / overlooked / cooling / reversing

Most overhyped

1. Bank generative-AI productivity (T-07-19). The specific evidence that hype outruns substance: JPMorgan reports ~1,000 AI use cases under development, Bank of America ~300 approved and 100+ deployed with all 200,000 employees given access and 90 live installations — and not one large US bank has published an attributable cost saving, efficiency-ratio delta or revenue figure, through two full earnings cycles (Q4 2025 reported January 2026, Q2 2026 reported July 2026). The most concrete disclosure is a 1,070-person headcount reduction through attrition on a base of 200,000, i.e. 0.5%. The most-quoted number in the whole discourse — "up to 20% industry cost reduction" — is a McKinsey projection, not a result. A deliberate search for independent measurement returned only vendor ROI marketing and content-farm guides.

2. Stablecoins as imminent deposit disintermediation (T-07-20). The evidence: the headline $1.3tn community-bank deposit-loss figure is an ICBA advocacy estimate conditional on a rule that has not been made (whether affiliates may pay yield), and Brookings authors Liang and Dudley — a former Treasury Under Secretary and a former New York Fed President, with no bank or issuer interest — call the concern "overstated." Meanwhile observed US domestic deposits grew 0.8% in Q2 2026, the eighth consecutive quarterly increase, and Bank of America reported a fourth consecutive quarter of consumer deposit growth with rate paid falling. The entire global USD stablecoin float (~$280–300bn) is about 2% of US bank loans alone.

Most overlooked

1. Software concentration in direct lending (T-07-14). Roughly 25–26% of direct lending portfolios sit in software, and outstanding SaaS loans went from ~$8bn in 2015 to over $500bn by end-2025 — about 19% of total direct loans, underwritten against ARR rather than EBITDA. Attention has missed it because private credit coverage focuses on aggregate default rates and AI coverage focuses on equity valuations; nobody is joining the two. This is the specific, nameable mechanism by which the macro brief's AI-valuation concern transmits into credit losses, and roughly 70% covenant-lite issuance means the signal arrives only at payment default.

2. Bank lending to non-depository financial institutions (T-07-03). The fastest-growing large asset class in US banking, at more than three times total loan growth, with the FSB able to identify only about $220bn against commercial estimates above $500bn. Overlooked because it is a boring line item in a quarterly regulatory release rather than a narrative.

3. That instant payment rails have historically grown card volumes. Brazilian and Indian card growth accelerated sharply after Pix and UPI. Missed because the disruption framing is more interesting than the data.

Trends that appear to be cooling

1. Property-catastrophe reinsurance pricing (T-07-16). Indicator that turned: five consecutive quarters with no single insured loss above $10bn, and only 11 billion-dollar insured events in H1 2026 against a ten-year average of 16. Rates softened through the April 2026 renewals despite the Middle East conflict. Note carefully: the price is cooling, the risk is not. Swiss Re maintains 5–7% annual structural growth in insured losses.

2. The sponsor-bank BaaS model (T-07-18). Indicator that turned: over 25% of FDIC and over 20% of OCC formal enforcement actions now target sponsor banks, and 80% of sponsor banks report struggling with multi-partner compliance. Incumbents are buying capability inside the charter instead (Capital One/Brex, Citi/Kard). Embedded finance demand is not cooling; cheap charter access is.

3. Fintech deal formation. Indicator that turned: 1,605 deals in H1 2026, down 25.7% year-on-year and 40% from H1 2024, even as dollars rose 22.7%. Dollars up, companies down.

Trends that may reverse, and the mechanism

1. US open banking (T-07-17) — already reversing. Mechanism: a Kentucky federal court enjoined the Section 1033 rule as likely exceeding statutory authority; the CFPB is rewriting it to permit data-provider fees after a set number of free requests, and has it on the rescission list. Free mandated data access — the foundational assumption of US aggregation, cash-flow underwriting and pay-by-bank — becomes metered.

2. Bank capital deregulation (T-07-06). Mechanism: a credit event during implementation. Capital rules are among the most politically reversible instruments in the sector; a 2.4% CET1 release granted in 2026 can be withdrawn by a different Board. Any strategy built on it has low persistence by construction.

3. The private credit stress narrative could reverse benignly. Mechanism: if the Fed cuts and refinancing reopens, PIK conversions unwind and the shadow default rate re-converges on the headline. The falsifiable indicator is the gap between the ~2% headline default rate and the ~6% PIK-implied distress rate — watch whether it narrows or widens over the next three quarters of BDC filings.

11. Risks and major uncertainties

Sector-specific risks

• Pro-cyclical capital release. Requirements are falling while NDFI exposure grows 22.4% and private credit shows early stress. If those coincide adversely, the buffer is thinner than it was.
• Valuation opacity in private credit. Quarterly, manager-determined marks with ~70% covenant-lite issuance means the information arrives late by design.
• Liquidity mismatch in retail wrappers. $607bn in evergreen vehicles promising periodic liquidity against illiquid assets, with gates already triggered.
• Concentration in tokenisation infrastructure. One asset manager and one transfer agent underpin the institutional tokenised-fund market.
• Consumer credit blind spot. ~$157bn of annual BNPL origination largely invisible to the credit bureau data that underpins US consumer risk models, concentrated in deep-subprime borrowers.
• Regulatory whipsaw. Nearly every US regulatory position in this dossier is reversible by a change of administration.
• Insurance affordability as a political risk. 25%+ premium increases in coastal regions combined with 305–331 day rate-approval delays in Colorado and California is a configuration that produces political intervention, not market clearing.

Genuine unknowns — distinguishing "we don't know" from "nobody can know"

We don't know (knowable, just not published or not retrieved):

• Bank AI financial returns. Banks have the numbers internally; they have not disclosed them.
• The true size of bank exposure to private credit funds. Supervisors could measure it with a reporting requirement.
• What share of stablecoin float would otherwise be US insured deposits. Measurable with the right survey.
• BNPL loss rates by product type. The Fed measured originations but not losses.
• Mastercard's 2026 quarterly figures and KPMG's fintech-specific H1 2026 total — both exist and were simply outside this research's search budget.

Nobody can know:

• The 2026-09-16 FOMC decision, one day after this research date, and the rate path beyond it.
• Whether agentic commerce produces material volume, because the demand has never existed in any form.
• Whether AI compresses SaaS revenue enough to impair ARR-based loans, and on what timeline.
• Whether a 17-bank consortium ships shared infrastructure in H1 2027. Consortium governance failure is not forecastable.
• Which catastrophe occurs next and where.

12. Scenarios to 2030

Base — "the plateau holds." Rates stay in a 3–4% band, NIM sits near 3.3%, credit normalises without an event. Capital deregulation completes; banks deploy released capital into NDFI lending and buybacks. Private credit's shadow default rate converges down toward the headline as refinancing reopens. Tokenised deposits ship late but ship; stablecoins settle into a large but non-systemic payments niche. Agentic commerce becomes a modest checkout channel. Fintech consolidates into incumbents. Falsifiable early indicator: aggregate FDIC NIM stays within 3.20–3.45% for four consecutive quarters and BDC non-accruals stay below 3% of fair value through 2027.

Upside — "productivity arrives and credit holds." A major bank discloses an attributable AI efficiency figure in 2027 and peers follow; efficiency ratios improve measurably. The exit window stays open and fintech IPOs reprice the private book upward. Tokenised deposit and stablecoin rails interoperate, cutting cross-border settlement cost materially. Falsifiable early indicator: any GSIB publishing a quantified, AI-attributed efficiency-ratio improvement in a 10-Q or earnings deck — as of 2026-09-15 none has.

Downside — "the shadow default rate was right." The 6% PIK-implied distress figure proves the better estimate. Evergreen vehicles gate broadly; retail holders are locked in; NDFI exposure crystallises through drawn facilities exactly when fund NAVs fall. Bank capital, just reduced 2.4%, proves thinner than needed. Falsifiable early indicator: non-traded BDC tender fill rates below 100% at three or more of the five largest managers in the same quarter, combined with the NDFI lending growth rate turning negative — the second being the sign banks are pulling lines.

Disruption — "the checkout moves." Agentic commerce reaches material volume faster than expected; the payment credential and the fraud liability move to the agent platform; card networks are relegated to settlement utilities and the consumer relationship migrates to whoever runs the agent. Falsifiable early indicator: Visa or Mastercard beginning to disclose agentic transaction volume as a separate line — a company only breaks out a number when it has become material.

Regulatory — "the pendulum swings back." A credit event or an election reverses the deregulatory direction: capital requirements are re-raised, the CFPB is refunded and rebuilt, Section 1033 is reinstated in its original form, and stablecoin yield is definitively prohibited. Firms that built strategies on the 2026 posture find them obsolete. Falsifiable early indicator: the March 2026 capital proposals failing to be finalised by mid-2027, or a second Federal Reserve Governor dissenting publicly.

Failure — "a 2008-shaped surprise from an unfamiliar direction." The failure mode is not a bank run; it is a correlated repricing of AI-exposed software credit that transmits through direct lending into the bank facilities financing those lenders, while capital buffers are at their lowest point since the reforms began, with quarterly marks hiding the deterioration until it is simultaneous. Falsifiable early indicator: software-sector non-accruals rising in BDC 10-Qs at the same time as the H.8 NDFI line decelerates — deterioration in the asset and withdrawal by the financier at once.

13. Data gaps and limitations

1. No disclosed bank AI financial results anywhere. The central gap of this dossier. Searched deliberately; found only vendor ROI marketing and Tier C content farms. Cannot be closed from public sources until a bank chooses to disclose.
2. FDIC Quarterly Banking Profile web pages returned HTTP 403 to automated fetching on 2026-09-15. Q2 2026 headline figures (net income $90.1bn, ROA 1.37%, NIM 3.32%) were verified through two independent secondary sources, and Q1 2026 figures directly from the FDIC press release — but noncurrent-loan rates, charge-off rates by category, and the problem-bank list count for Q2 2026 could not be verified at all. Phase 2 should use FFIEC CDR bulk download.
3. Two Federal Reserve System publications differ by more than 2x on US BNPL market size ($70bn real transaction value vs $156.7bn nominal originations), a definitional divergence recorded in full in T-07-08's contradictions.
4. Swiss Re and Gallagher Re differ by ~10% on H1 2026 insured cat losses ($42bn vs $46bn) and by 42% on economic losses ($100bn vs $142bn). Recorded in T-07-16.
5. FSB's own bank-to-private-credit exposure figure ranges from $220bn to over $500bn depending on whether supervisory or commercial data is used. Supervisors cannot see this market precisely.
6. Morningstar PitchBook's $607bn all-asset-class evergreen figure is smaller than With Intelligence's $644bn private-credit-only figure, so the two universes are not nested. Neither is wrong; both are widely quoted as if comparable.
7. KPMG's fintech-specific H1 2026 figure was not retrieved within the 25-search budget, so the Crunchbase $28.6bn figure stands single-sourced at the sector level. Given the macro brief documents a $510bn/$560.4bn divergence on all-sector VC, the fintech subset certainly diverges too.
8. Mastercard 2026 quarterly figures not independently verified. Visa's were (direct from the IR PDF). The Mastercard entity record is marked low confidence accordingly.
9. Non-English primary sources were partly inaccessible. NPCI's UPI statistics page is disallowed by robots.txt for automated agents; Banco Central do Brasil's Pix statistics page is JavaScript-rendered and returns no data to a plain fetch. Regional evidence therefore rests on the Atlantic Council (2025-04-21) and the ECB, and the Brazil/India transaction data is a year or more stale. This is a real limitation for a sector whose most interesting payments developments are in Portuguese and Hindi-speaking markets.
10. The embedded finance market-size series ($115.8bn → $251.5bn, 16.8% CAGR) has no published methodology and no named modeller in the source consulted. Recorded as an estimate and explicitly flagged as weak; it should not be quoted.
11. Paywalled or licensed: Trepp loan-level CMBS data, PitchBook's database, With Intelligence's underlying data, most broker reinsurance reports. Phase 2 licensing budget should prioritise Trepp (for CRE) and one private-markets database.
12. No consumer or civil-society organisation activity verified for 2026. Given the CFPB's retrenchment, advocacy groups are likely more consequential than usual and this is a genuine hole.
13. OCC leadership could not be verified. One source referenced an "Acting Comptroller Hood" in mid-2026; occ.gov was inaccessible to automated fetching. The entity record flags this as unverified rather than guessing.
14. The 2026-09-16 FOMC decision falls one day after this research date and is unresolved. Every rate-dependent claim in this dossier is conditional on it.

14. Ranking scorecard

15. Sources

1. "FDIC-Insured Institutions Reported Return on Assets of 1.26 Percent and Net Income of $80.5 Billion in First Quarter 2026" — Federal Deposit Insurance Corporation — https://www.fdic.gov/news/press-releases/2026/fdic-insured-institutions-reported-return-assets-126-percent-and-net — 2026-05-27 — Tier A
2. "FDIC Approves Proposal to Implement GENIUS Act Requirements and Standards" — Federal Deposit Insurance Corporation — https://www.fdic.gov/news/press-releases/2026/fdic-approves-proposal-implement-genius-act-requirements-and-standards — 2026-04-07 — Tier A
3. "Report on Vulnerabilities in Private Credit" — Financial Stability Board — https://www.fsb.org/uploads/P060526.pdf — 2026-05-06 — Tier A
4. "Financial Stability Report July 2026" — Bank of England — https://www.bankofengland.co.uk/financial-stability-report/2026/july-2026 — 2026-07-07 — Tier A
5. "Agencies issue final rule to modify certain regulatory capital standards" — Board of Governors of the Federal Reserve System — https://www.federalreserve.gov/newsevents/pressreleases/bcreg20251125b.htm — 2025-11-25 — Tier A
6. "Speech by Vice Chair for Supervision Bowman on Basel III and bank capital rules" — Board of Governors of the Federal Reserve System — https://www.federalreserve.gov/newsevents/speech/bowman20260312a.htm — 2026-03-12 — Tier A
7. "FOMC Minutes, July 28–29 2026" — Board of Governors of the Federal Reserve System — https://www.federalreserve.gov/monetarypolicy/fomcminutes20260729.htm — 2026-07-29 — Tier A
8. "Buy Now, Pay Later Beyond Pay in 4: A Comprehensive Product Overview" (FEDS Notes) — Board of Governors of the Federal Reserve System — https://www.federalreserve.gov/econres/notes/feds-notes/buy-now-pay-later-beyond-pay-in-4-a-comprehensive-product-overview-20260605.html — 2026-06-05 — Tier A
9. "Buy Now, Pay Later: Recent Developments and Implications" (Economic Brief 26-05) — Federal Reserve Bank of Richmond — https://www.richmondfed.org/publications/research/economic_brief/2026/eb_26-05 — 2026-02-11 — Tier A
10. "Quarterly Report on Household Debt and Credit, 2026 Q2" — Federal Reserve Bank of New York — https://www.newyorkfed.org/medialibrary/interactives/householdcredit/data/pdf/HHDC_2026Q2.pdf — 2026-08-01 — Tier A
11. "Homeowners Insurance: Premiums Generally Tracked Inflation but Rose More in Disaster-Prone Areas" (GAO-26-107867) — US Government Accountability Office — https://www.gao.gov/products/gao-26-107867 — 2026-02-27 — Tier A
12. "First-half 2026 insured catastrophe losses: below trend, rising risks" — Swiss Re Institute — https://www.swissre.com/institute/research/topics-and-risk-dialogues/climate-and-natural-catastrophe-risk/first-half-2026-insured-catastrophe-losses.html — 2026-08-11 — Tier A
13. "Visa Reports Fiscal Third Quarter 2026 Results" — Visa Inc. — https://s1.q4cdn.com/050606653/files/doc_financials/2026/q3/Q3-2026-Earnings-Release_vF.pdf — 2026-07-28 — Tier A
14. "Visa and Partners Complete Secure AI Transactions, Setting the Stage for Mainstream Adoption in 2026" — Visa Inc. — https://usa.visa.com/about-visa/newsroom/press-releases.releaseId.21961.html — 2025-12-18 — Tier A
15. "Bank of America First Quarter 2026 Earnings Call transcript" — Bank of America Corporation — https://d1io3yog0oux5.cloudfront.net/_6239653d48bde8c1848e60d3fc022eb9/bankofamerica/db/775/10537/webcast_transcript/2026-04-15+BAC+1Q+Earnings+Call_ADA.pdf — 2026-04-15 — Tier A
16. "Stripe powers Instant Checkout in ChatGPT and releases Agentic Commerce Protocol codeveloped with OpenAI" — Stripe — https://stripe.com/newsroom/news/stripe-openai-instant-checkout — 2025-09-29 — Tier A
17. "Buy it in ChatGPT: Instant Checkout and the Agentic Commerce Protocol" — OpenAI — https://openai.com/index/buy-it-in-chatgpt/ — 2025-09-29 — Tier A
18. "BPI, TCH, CBA Comment on FDIC GENIUS Act Stablecoin and Tokenized Deposit Rule" — Bank Policy Institute — https://bpi.com/bpi-tch-cba-comment-on-fdic-genius-act-stablecoin-and-tokenized-deposit-rule/ — 2026-06-09 — Tier A
19. "Digital euro project status and scheme rulebook (draft v0.91)" — European Central Bank — https://www.ecb.europa.eu/euro/digital_euro/html/index.en.html — 2026-07-01 — Tier A
20. "FCA news 2026" (incl. Nikhil Rathi, "Rethinking regulation for the age of AI") — Financial Conduct Authority — https://www.fca.org.uk/news — 2026-09-01 — Tier A
21. "Private Credit Redemptions, Defaults, and Wrappers, Oh My!" — Chartered Alternative Investment Analyst Association — https://caia.org/blog/2026/04/20/private-credit-redemptions-defaults-and-wrappers-oh-my/ — 2026-04-20 (updated 2026-07-13) — Tier B
22. "Private Credit Outlook 2026: Market Faces First Big Test" — With Intelligence — https://www.withintelligence.com/insights/private-credit-outlook-2026/ — 2026-01-07 — Tier B
23. "Do the Recent Bankruptcies of First Brands and Tricolor Suggest Trouble Ahead in Private Credit?" — Cambridge Associates — https://www.cambridgeassociates.com/insight/do-the-recent-bankruptcies-of-first-brands-and-tricolor-suggest-trouble-ahead-in-private-credit/ — 2025-11-11 — Tier B
24. "Next steps for GENIUS payment stablecoins" — Brookings Institution (Nellie Liang and William C. Dudley) — https://www.brookings.edu/articles/next-steps-for-genius-payment-stablecoins/ — 2026-03-03 — Tier B
25. "The GENIUS Act in 2026" — ABA Banking Journal (American Bankers Association) — https://bankingjournal.aba.com/2026/07/the-genius-act-in-2026/ — 2026-07-20 — Tier B
26. "US Banking Regulators Propose Reforms to Capital Requirements" — Mayer Brown — https://www.mayerbrown.com/en/insights/publications/2026/03/us-banking-regulators-propose-reforms-to-capital-requirements — 2026-03-19 — Tier B
27. "Funding Haze And Deregulatory Pursuits: The CFPB In 2026" — Covington & Burling LLP — https://www.cov.com/-/media/files/corporate/publications/2026/01/funding-haze-and-deregulatory-pursuits-the-cfpb-in-2026.pdf — 2026-01-02 — Tier B
28. "As CFPB retreats, state AGs and bank regulators step up" — American Banker — https://www.americanbanker.com/news/as-cfpb-retreats-state-ags-and-bank-regulators-step-up — 2026-01-21 — Tier B
29. "Open banking regulation in 2026: federal regulation resurfaces as states bring data sharing into focus" — Consumer Finance Monitor (Ballard Spahr) — https://www.consumerfinancemonitor.com/2026/06/26/open-banking-regulation-in-2026-federal-regulation-resurfaces-as-states-bring-data-sharing-into-focus/ — 2026-06-26 — Tier B
30. "Fintech Funding Surges 23% In H1 2026 As Investors Concentrate Their Bets On AI And Financial Infrastructure" — Crunchbase News — https://news.crunchbase.com/fintech/funding-rises-deals-slump-h1-2026/ — 2026-07-15 — Tier B
31. "Capital One to acquire fintech startup Brex for $5.15B" — Crunchbase News — https://news.crunchbase.com/ma/capital-one-acquisition-fintech-startup-brex/ — 2026-01-23 — Tier B
32. "August 2026: Top five fintech M&A stories of the month" — FinTech Futures — https://www.fintechfutures.com/m-a/august-2026-top-five-fintech-m-a-stories-of-the-month — 2026-08-28 — Tier B
33. "A look back, and a look ahead, at fintech IPOs" — This Week in Fintech — https://www.thisweekinfintech.com/a-look-back-and-a-look-ahead-at-fintech-ipos/ — 2026-01-16 — Tier B
34. "Top 5 fintech IPOs of 2025" — American Banker / PaymentsSource — https://www.americanbanker.com/payments/news/top-5-fintech-ipos-of-2025 — 2025-12-30 — Tier B
35. "Big banks continue the hunt for AI-driven efficiencies" — CIO Dive — https://www.ciodive.com/news/banks-pursue-ai-efficiency-gains/809898/ — 2026-01-16 — Tier B
36. "The Real Bank Earnings Story Was the AI Spending Boom" — PYMNTS — https://www.pymnts.com/news/banking/2026/real-bank-earnings-story-was-ai-spending-boom/ — 2026-07-15 — Tier B
37. "Gallagher Re estimates global insured cat losses at $46bn for H1'26, 28% below 10-year average" — Artemis.bm (reporting Gallagher Re) — https://www.artemis.bm/news/gallagher-re-estimates-global-insured-cat-losses-at-46bn-for-h126-28-below-10-year-average/ — 2026-07-15 — Tier B
38. "Fast payments in action: Emerging lessons from Brazil and India" — Atlantic Council GeoEconomics Center — https://www.atlanticcouncil.org/blogs/econographics/fast-payments-in-action-emerging-lessons-from-brazil-and-india/ — 2025-04-21 — Tier B (regional source; data to Q2 2024 and therefore stale)
39. "JPMorgan's Kinexys Blockchain Hits $4 Trillion, Adds Five APAC Currencies" — The Defiant — https://thedefiant.io/converge/tradfi-and-fintech/jpmorgan-kinexys-4-trillion-apac-currencies — 2026-06-29 — Tier B
40. "BlackRock expands tokenized cash with new blockchain-based money market offerings" — CoinDesk — https://www.coindesk.com/business/2026/08/03/blackrock-expands-tokenized-cash-with-new-blockchain-based-money-market-offerings — 2026-08-03 — Tier B
41. "Evergreen Funds Reach $607 Billion Despite Redemptions" — WealthManagement.com (citing Morningstar PitchBook) — https://www.wealthmanagement.com/alternative-investments/evergreen-funds-grow-to-607b-despite-redemptions — 2026-07-09 — Tier B
42. "Embedded finance grows up and regulators take notice" — FinTech Global — https://fintech.global/2026/06/23/embedded-finance-grows-up-and-regulators-take-notice/ — 2026-06-23 — Tier B (market-size figures carry no published methodology — see §13.10)
43. "Commercial Mortgage-Backed Securities Distress Is Peaking — Again" — Commercial Observer — https://commercialobserver.com/2026/08/cmbs-distress-2026/ — 2026-08-31 — Tier B
44. "FDIC: Banks see strong earnings growth in Q2" — Financial Regulation News — https://financialregnews.com/fdic-banks-see-strong-earnings-growth-in-q2/ — 2026-08-27 — Tier B
45. "Looking Under the Hood: A Closer Look at the FDIC's Q2 2026 Banking Report" — Premier Insights — https://www.premierinsights.com/blog/looking-under-the-hood-a-closer-look-at-the-fdics-q2-2026-banking-report — 2026-08-27 — Tier C (used only for figures cross-checked against source 44; never as sole support)
46. "Banks Rush to Tokenize Deposits as Stablecoin Networks Beat Them to Shared Payment Rails" — TechTimes — https://www.techtimes.com/articles/324351/20260813/banks-rush-tokenize-deposits-stablecoin-networks-beat-them-shared-payment-rails.htm — 2026-08-13 — Tier C (consortium membership cross-checked against source 39)
47. "Insurify Projects Home Insurance Rates Will Rise for the 5th Consecutive Year, After a 12% Increase in 2025" — Insurify — https://insurify.com/homeowners-insurance/report/home-insurance-price-projections/ — 2026-01-01 — Tier C (vendor with named methodology: Quadrant Information Services rate filings; paired with GAO source 11, never sole support)
48. "Ant International, Visa, and Mastercard Agree on Agent Identity Standard" — Forkast — https://forkast.news/ant-international-visa-and-mastercard-agree-on-agent-identity-standard-now-comes-the-hard-part/ — 2026 (undated at page level, undated: true) — Tier C (sole basis for the Mastercard agent-identity claim; that entity record is marked low confidence accordingly)

More research in this sector · Explore trend records