379 verified-active digital banks, measured — custody, licenses, cards, stablecoins, geography and the fine print. The opening third is free below; the full 60-page designed PDF is free for subscribers. data as of 13 August 2026.
The State of Neobanks is a monthly measurement of the digital banking industry, produced from the neobankbeat open dataset — a hand-verified directory of every neobank we can confirm as currently operating. Defunct entities are removed, not averaged in; unverifiable fields are left empty, never estimated. What follows describes the industry that exists, not the one that was announced.
A consumer- or SMB-facing, digital-first money app offering everyday banking — an account or balance, usually a card, payments — without a branch network. Three structural waves qualify: traditional fiat challengers, hybrid apps holding fiat and custodial crypto together, and web3-native self-custodial apps where the user holds the keys. Pure infrastructure (BaaS, issuer-processors), superapp payment features without banking ambition, and anything we cannot verify as live are excluded.
"neobankbeat, The State of Neobanks — August 2026 (№ 02), data as of 13 August 2026." Link: neobankbeat.com/report
Figures are compiled from public sources for comparison and research, and are not financial advice. Cashback and yield figures are "up to" headline rates that change constantly — always confirm with the issuer. © 2026 neobankbeat · Francesco Andreoli · MIT.
The founding boom that created this industry is over: among today's survivors, new-neobank formation peaked at 43 in 2019 and has collapsed to 11 in 2025 — a quarter of the peak. What replaced volume is structural change — the marginal new neobank is dramatically more likely to be self-custodial (30% of the 2020s cohort vs 5% of the 2010s), more likely to be niche-first, and near-certain to touch stablecoins.
Meanwhile the industry's centre of gravity sits where the marketing isn't: Latin America, Africa and Asia grow the giants (WeBank's 400M and Nubank's 131M dwarf every Western player), while Europe hosts the greatest density of players (144 active). And beneath everything runs the report's core tension: only 35% of neobanks are licensed banks — the remaining 247 rest on partner banks, e-money safeguarding, crypto licenses, or no custodian at all. The gap between what apps imply and what their legal structure delivers remains the industry's biggest consumer risk, and its least covered story.
Banking's interesting boundary is no longer bank vs fintech — it is custodial vs self-custodial, and every quarter moves more of the industry across it.
The series now has a baseline, so this edition reports the headline metrics against July. Net, the dataset moved from 357 to 379 verified-active neobanks — 28 added and 6 removed as apps launched, relicensed and quietly died. The mix kept tilting the same way: hybrid and web3-native both grew faster than traditional, and stablecoin support widened again.
| metric | august | july | Δ | definition |
|---|---|---|---|---|
| Verified-active neobanks | 379 | 357 | +22 | live and onboarding |
| — traditional / hybrid / web3-native | 260 / 61 / 58 | 256 / 54 / 47 | +4 / +7 / +11 | wave split |
| Stablecoin support | 121 (32%) | 103 | +18 | any verified support |
| Licensed banks¹ | 132 (35%) | — | n/a | charter holders |
| Partner-bank (BaaS) model¹ | 81 | — | n/a | rent a charter |
| Self-custodial (incl. MPC) | 54 | 42 | +12 | no custodian exists |
| No-KYC apps | 14 | 12 | +2 | zero identity checks |
| Niche-first neobanks | 126 (33%) | — | — | named audience |
| Largest reported user base | WeBank · 400M | — | — | self-reported |
¹ Regulation-type coverage was expanded and backfilled since № 01, so these two rows are not month-over-month comparable — the shift reflects more complete classification, not net industry movement. Category, stablecoin, self-custody and KYC deltas track real additions and removals.
28 added · 6 removed since № 01. The full births-and-deaths list — with the cause behind every removal — is on the next page.
A directory is only as honest as its removals. Neobanks rarely fail with a bang — usually an app just stops updating and support goes quiet. The 6 that left the active list since № 01 died three ways: a partner or rail collapsing beneath them, an acquirer switching them off, or a regulator pulling the licence. Who, and why:
the defining risk of the rent-a-charter model: you die when your sponsor does
Fi Money IN · savings app · 3.5M users · ~$169M raised
Partner Federal Bank ended the relationship (11 Mar 2026) amid the RBI's tightening of bank–fintech tie-ups and thin unit economics; 3.5M customers were redirected to the bank's own app as Fi pivoted to AI.
Juno US · crypto-friendly checking + savings
Collateral damage in the 2024 Synapse/Evolve collapse: when Evolve Bank lost access to Synapse's ledger, customer funds froze across 50+ fintechs (~$95M went missing industry-wide). Juno wound down its Treasury account and pivoted to on-chain.
Kard FR · family & teen banking
Its e-money provider terminated the contract, leaving no rail to operate on — the company went into liquidation.
not a failure so much as consolidation — the users move, the brand does not
Pomelo US · remittance + credit card, Philippines corridor
Acquired by Zepz (WorldRemit / Sendwave) in Jan 2026; the product was paused during integration and the team folded in — an exit, not a failure.
Z1 BR · teen neobank
Absorbed by crypto neobank NG.CASH; the Z1 brand was retired.
the rarest and most abrupt: the licence is pulled and the doors shut
Will Bank BR · digital bank
Liquidated by Brazil's Central Bank (Jan 2026) — nominally for breaching Mastercard obligations, but tied to the collapse of the Banco Master conglomerate (a severe liquidity crisis and an ~R$11.5B fraud probe) that felled several linked institutions.
Sources are on each entity's profile; the running death log is the public changelog, and the deeper pattern is dissected in "why neobanks die" and the deposit-risk essay "who actually holds your money?"
Verified additions since № 01 — community-submitted or found in our discovery sweep, then checked before listing. The pattern mirrors the whole dataset's drift: heavier on hybrid and web3-native, stablecoin-first, and increasingly niche.
AMP Bank GO · Always.bank · BFinance · Bitget Wallet · Blink · Brighty · Brookwell · COCA · Dolafy · Esh Bank · Flex · Flouci · Haventree Bank · Hyperbeat · Karta · Lava · Mine · Moto · PaySika · Plata · Slush · Solid · Sony Bank · Startale · SurfCash · Takenos · Tangem · Veera
Every figure here is live and updated continuously. Explore all 379:
▸ the directory · world map by country · sortable database · feature matrix
▸ the changelog — every add & death as it happens · the blog — a deep dive per chapter
▸ machine-readable: data.json · MCP server — point your AI assistant at it
Foundings among survivors: 43 (2019) → 27 (2023) → 18 (2024) → 11 (2025). "Another challenger bank" stopped being fundable around 2022; what raises now is niche underwriting edges and stablecoin-native architecture.
47 of 146 survivors founded in the 2020s (32%) are web3-native, vs 5% of the 2010s cohort. The industry's marginal energy has moved to the model where no company holds the balance.
All 58 web3-native and 56 of 61 hybrids support stablecoins — but only 7 of 260 traditional neobanks do. That 2.7% is either a ceiling or the floor of the next migration. We think floor.
132 of 379 hold a charter. The rest: 81 partner-bank models, 48 e-money institutions, 53 self-custodial software, and a long unclassifiable tail. Deposit insurance is rarer than landing pages suggest.
313 of 379 issue a card (182 Visa, 149 Mastercard programmes). 141 advertise cashback, 205 offer yield — nearly all behind "up to" tiers. Interchange-only economics are visibly straining.
Active presence: Europe 144, Asia 132, North America 126, LatAm 113, Africa 91. But the largest customer bases are all emerging-market.
126 of 379 serve a named niche — SMB (33), underbanked (23), freelancers (15), gen z (8), immigrants (8), kids (7), faith-based (7) and a dozen more. "Right bank for someone" beats "better bank for everyone".
14 of 379 are usable with no identity check at all; all are self-custodial wallets; none issues a card without KYC. The line is structural, not cultural.
Official terms documents verifiable for only 145 of 379 — after an audit that repaired 60 dead legal links and removed 53 that resolve nowhere. For an industry holding money, basic document hygiene remains weak.
Traditional players add stablecoin rails; web3-native apps acquire e-money licenses and IBANs. Our own three-way classification gets harder to maintain each quarter — which is itself the finding.
Every number above is a few lines of code against neobankbeat.com/data.json — schema in Appendix B.
"Neobank" is one word covering three different machines. The differences — who holds the money, what can freeze it, what insures it — matter more than any feature comparison, so this taxonomy underpins every page that follows.
Fiat money, custodial accounts, a card, and a regulatory wrapper that permits holding customer funds. From Simple and Nubank through the London cluster to today's licensed digital banks across the Gulf and Asia. Contains all the giants.
Fiat plus custodial crypto in one app, arriving ~2017 from two directions: banking apps adding trading (Revolut, Cash App) and exchanges adding cards (Crypto.com, Coinbase, Binance). One counterparty holds everything.
Self-custodial apps where the user holds the keys and the company holds nothing: wallets that grew cards (MetaMask, Phantom), and card programmes built on smart accounts (Gnosis Pay, EtherFi Cash, Payy). No deposit, no deposit insurance, no custodian to fail.
| traditional | hybrid | web3-native | |
|---|---|---|---|
| Who holds funds | Bank or partner bank | The company (fiat + crypto) | The user (keys) |
| Balance is | Fiat deposit | Fiat + custodial crypto | Stablecoins / crypto |
| Deposit insurance | Usually (direct or pass-through) | Fiat sometimes; crypto never | None — no deposit exists |
| Main failure mode | Bank failure / ledger gaps | Custodian failure | Key loss, issuer or contract risk |
| KYC | Always | Always | Card-only or none |
| Stablecoin support | 7 of 260 (2.7%) | 56 of 61 (92%) | 58 of 58 (100%) |
| Archetypes | Nubank, Chime, Monzo | Revolut, Cash App, Crypto.com | MetaMask, Gnosis Pay, Payy |
The 2010s produced a traditional-wave industry with a hybrid fringe. The 2020s cohort is a different animal: web3-native is its second-largest wave, larger than hybrid, and growing while overall formation shrinks.
| term | as used in this report |
|---|---|
| Neobank | Digital-first consumer/SMB money app: account or balance + payments, usually a card, no branch network |
| Verified-active | We can confirm the product is live and onboarding today; defunct or paused entities are removed |
| Custodial | The company (or its partner bank) legally holds customer funds |
| Self-custodial | The user holds the keys; the company cannot move or freeze the balance |
| MPC self-custody | Key split via multi-party computation — self-custody without a single seed phrase |
| Licensed bank | Holds a banking charter; deposits sit on its own balance sheet with direct deposit insurance |
| Partner-bank model | Unlicensed app fronting a chartered bank; insurance applies pass-through, contingent on accurate ledgers |
| E-money institution | EU/UK license to issue e-money; funds safeguarded in segregated accounts, not deposit-insured |
| CASP | Crypto-Asset Service Provider authorised under MiCA; passports across EU/EEA |
| Stablecoin support | Any verified support: balances, transfer rails, funding a card, or on/off-ramps |
| KYC: card only | The wallet is permissionless; identity is required only to obtain the card |
| "Up to" rate | Headline cashback/yield gated by tiers, staking, subscriptions or balances — not the typical rate |
Wave colours used throughout: ■ traditional · ■ hybrid · ■ web3-native. Orange marks findings and editorial emphasis.
Read this chart carefully: it under-counts early years (more time to die) and the last two (not yet surfaced). Even so, the shape is unmistakable — a build through the mid-2010s, the 2018–19 peak (85 survivors in two years), a COVID dip, the 2021 bull-market echo (36), then contraction to 11 verified from 2025.
Three eras explain it. 2013–2016: smartphone banking becomes viable; the archetypes launch. 2017–2021: venture abundance funds a challenger for every country and demographic; BaaS makes launching one a procurement decision. 2022→: rates rise, fintech funding halves and halves again, and the marginal pitch shifts from "bank, but nicer" to products with a structural edge — underwriting a niche, or removing the custodian entirely.
The 2025–26 cohort surfacing over coming months skews heavily web3-native and stablecoin-first in our intake pipeline — next editions will quantify it.
| year | what happened | why it mattered |
|---|---|---|
| 2009–13 | Simple, Moven, GoBank; Fidor in Germany | Proved a bank could live in an app — mostly on partner charters |
| 2013–15 | Nubank founded; UK mints Atom, Tandem, Monzo, Starling licenses | The two enduring models appear: LatAm scale and UK charter-first |
| 2016–17 | N26 gets a full license; Revolut adds crypto; Crypto.com & exchange cards | Wave two begins — fiat and custodial crypto in one app |
| 2018–19 | Peak formation: 85 of today's survivors founded in 24 months | BaaS turns launching a neobank into a procurement decision |
| 2020 | COVID; Kakaobank/jiban giants scale; Australia's Xinja collapses | First proof that licenses without economics don't survive |
| 2021 | Bull-market echo cohort (36 survivors); Nubank IPO at ~$41B | The category gets a public-market benchmark |
| 2022 | Rates rise; funding halves; FTX fails | Custody stops being a philosophical question |
| 2023 | Gnosis Pay ships the first self-custodial Visa card at scale | Wave three gets its archetype: smart account + card rails |
| 2024 | Synapse collapse strands partner-bank customers; MiCA phases in | The pass-through model's weakest joint fails in public |
| 2025 | GENIUS Act; FDIC recordkeeping rule; wallet-card wave (MetaMask, Phantom) | Stablecoins become supervised instruments; wallets become neobanks |
| 2026 | MiCA grandfathering closes; traditional-wave stablecoin pilots begin | The migration this report exists to measure |
Deeper cuts on each era: the blog series at neobankbeat.com/blog, starting with "the three waves of neobanks".
This dataset's defining choice is survivorship: we track who is alive, and remove who is not. But the removals teach as much as the roster. Four failure archetypes recur:
Xinja (Australia) returned its license and customer deposits in 2020; Volt followed in 2022. Both were fully licensed and adequately capitalised — the model failed on cost of funds vs. cost of growth, not on regulation. Contrast Up, which thrived inside Bendigo's charter, and Judo, which found profit in SMB lending: same market, different unit economics.
Synapse (2024) was infrastructure, not a neobank — which is exactly why its bankruptcy stranded end users of apps that were "FDIC-insured" in marketing copy. Insurance protects against bank failure, not against the ledger between you and the bank going dark. The FDIC's recordkeeping rule is the regulatory scar tissue.
Dozens of 2018–21 vintage niche apps quietly sunset after Series A: the audience was real, the CAC advantage was real, but the deposit base needed to fund the roadmap never arrived. This is the tail our monthly removals mostly consist of.
Custodial crypto neobanks whose balance sheets or user trust didn't survive 2022 (and whose users learned the difference between an account and a claim). The lasting effect shows up in this report as wave three's growth: the market's answer to custodial failure was less custody.
No failure archetype involves the product being bad. Neobanks die of balance-sheet physics, middleware, and funding weather — which is why this report spends its pages on custody, licenses and economics rather than app-store screenshots.
Subscribe to the (free) newsletter and the full 60-page designed PDF downloads instantly, right here — plus every monthly edition lands in your inbox. No spam, no paid tier.
free · via Substack · hit subscribe — your download starts as you finish
already subscribed, or the box didn't load? just download →
If it didn't, download the PDF directly. Not subscribed yet? Use the box above — the September edition ships there.