What 3,902 fintechs actually run on.
Money moves on open rails. The data behind those movements doesn’t.
Layer disclosure · 3,902 fintechs
% naming ≥1 provider
$185BB2B SaaS embedded-payments TAM·<20% captured·BCG + Adyen, Oct 2024
Embedded finance =
Every app is becoming a bank.
Toast moves payments.
Shopify writes loans.
Lyft issues cards.
Brex runs treasury.
Airbnb sells flexibility. All of it powered by invisible banks behind a handful of APIs. Once a consumer-neobank story (
Chime,
Revolut,
N26), embedded finance is now the default playbook for vertical SaaS, marketplaces, B2B platforms, and even non-financial brands — anyone who can ship a payment, issue a card, write a loan, or hold a balance without sending users to a third party. The 18-month bank-charter project is now a two-week API integration.
814
Name their sponsor bank
Evidence from subprocessor lists, regulatory footers, trust pages, customer announcements, and ~478 press articles — produced by Apideck with content partners Open Banking Tracker and Embedded Finance Review.
Headline finding ·Stripe is named by 326 of 3,902 fintechs (8.4%) — the only vendor above 5% concentration in any infrastructure layer.
Section 02 · disclosure rates
Which layers are auditable. Which go dark.
For each layer of the embedded-finance stack: what share of 3,902 analyzed fintechs publicly named a vendor? The gap between the green and red rows is the dark half of the stack — the layers regulators, integrity researchers, and journalists can’t audit from outside.
Layer
% masked ?
% disclosed
PSP (payments)
31.2%masked
45.9%
Sponsor bank
2.9%masked
20.9%
Financial data connectivity
6.4%masked
16.2%
Card issuing
29.1%masked
16.0%
FX / payouts
22.4%masked
15.9%
KYC / identity
0.0%masked
3.7%
Regulatorily disclosed (named on trust/regulatory surface) Commercially disclosed (named via case studies, partnerships) Trade-secret layer (rarely disclosed publicly)
What this meansThe payments side of the stack is auditable from outside. The risk side isn’t. Public datasets of vendor relationships — including this one — will always over-represent payments and under-represent KYC, fraud, and sanctions screening. Treat the red rows as a floor, not a ceiling. The “% masked” column is the second-order finding: even when a layer IS disclosed, a meaningful fraction of companies talk about their own-brand product (“Acme Pay”, “Acme Checkout”) without identifying the underlying provider. ~34% of fintechs that mention a PSP, and ~37% of fintechs that mention BaaS, hide the actual infrastructure behind a brand wrapper — a structural masking effect that makes the auditable-looking layers thinner than they appear.
“Open banking opened the access door to financial data, but access and usability are not the same thing. A full internal transaction record carries around nine usable data points. What comes through an open banking API is typically two, sometimes four.”
— Ivan Dovica · CEO & Co-founder, Tapix by Dateio
Section 03 · dataset composition
What kind of fintech are these?
The 3,902 fintechs in this dataset split into 9 working categories. Infrastructure providers (BaaS, card issuing, PSP) sit alongside the platforms that consume them (vertical SaaS, marketplaces, B2B fintech) — the report is a snapshot of an ecosystem where buyer and seller often appear in the same chart. 790 entries were excluded as non-fintech (investor/advisory firms, non-financial orgs, press fragments) before this breakdown.
Fintech infrastructure
1,369·35.1%
Consumer neobank
400·10.3%
Consumer brand (embedded)
247·6.3%
Marketplace / platform
236·6.0%
Segments resolved via an enrichment cache + a local-profile fallback classifier sniffing each company’s public industry, tagline, and thesis. “Other” entries are kept in the dataset; we surface them as a visible gap rather than dropping them or guessing.