The State of Neobanks 2026: what 357 verified-active digital banks reveal
Most industry reports count press releases. This one counts survivors. The neobankbeat dataset tracks every neobank we can verify as currently operating — 357 of them as of July 2026 — with custody model, licence, card rails, stablecoin support and geography checked entity by entity. Defunct banks are removed, not averaged in. That gives the numbers below an unusual property: they describe the industry that actually exists, not the one that was announced.
Ten findings. Every figure is reproducible from the open dataset.
1. The founding boom is over — and has been for years
Among today's survivors, foundings peaked at 45 in 2019 (40 in 2018, 35 in 2021 as the fintech bull market echoed). Then the cliff: 22 survivors founded in 2023, 16 in 2024, and just 4 verified so far from 2025. Some of that is pipeline lag — young companies take time to surface — but the direction is unambiguous. The era of "another challenger bank" as a fundable pitch ended around 2022. What gets funded now is either a niche with real underwriting edge or a stablecoin-native architecture.
2. A third of the new generation is self-custodial
Of neobanks founded in the 2020s that survive today, 38 of 125 — 30% — are web3-native. In the 2010s cohort that figure was 4% (9 of 215). The industry's marginal energy has visibly moved to the self-custodial model: wallets with cards, smart accounts with IBANs, apps where no company holds the balance. The incumbent-challenger fight was the last decade's story; custody itself is this decade's.
3. Stablecoin support is now a category marker — except where it matters most
103 of 357 neobanks (29%) support stablecoins in some form. But the distribution is the story:
| category | stablecoin support | share |
|---|---|---|
| web3-native | 47 of 47 | 100% |
| hybrid | 49 of 54 | 91% |
| traditional | 7 of 256 | 2.7% |
The traditional wave — the one with hundreds of millions of users — has barely started. That 2.7% is the most interesting number in this report: it's either the ceiling of a niche or the floor of the next migration. Post-GENIUS Act and MiCA, our money is on the latter, and every traditional neobank that flips the switch moves ~millions of users onto stablecoin rails overnight. (Flagged as our reading, not a fact.)
4. Deposit insurance is rarer than the marketing suggests
Only 92 of 357 (26%) are licensed banks holding deposits directly. The rest stack up as: 39 self-custodial software (no deposits to insure), 20 partner-bank models, 8 e-money institutions, 5 MiCA CASPs, 5 payment institutions, 5 VASP/MSB arrangements — and a long tail where the regulatory posture isn't cleanly classifiable from public sources (we mark those unclassified rather than guess). If you take one action from this report: find your app's row in the directory and read what actually protects your balance. The safety deep dive explains each model.
5. Cards are near-universal; the economics behind them aren't
299 of 357 issue a card (167 Visa, 146 Mastercard programmes, some issue both networks). 131 advertise cashback; 190 offer some form of yield. But "up to" does heavy lifting across the industry — headline rates typically gate behind subscriptions, staking or balance tiers. The interchange-only business model that funded the first wave's free accounts is visibly straining; yield and subscription tiers are where monetisation moved.
6. Europe hosts the most neobanks; Latin America grows the biggest ones
Active-presence counts: Europe 131, Asia 120, North America 111, Latin America 102, Africa 79, MENA 72, Oceania 61 (multi-region players counted everywhere they operate). But scale inverts the ranking: the largest customer bases are emerging-market — Nubank's 131M leads the entire industry, with WeBank, bKash, OPay and GCash in the same tier. The West has the density; the Global South has the giants. Why that is is the least-told story in fintech.
7. One in three neobanks now picks an audience first
117 of 357 (33%) serve a named niche: SMB & startups (29), the underbanked (21), freelancers & creators (14), gen z & students (9), immigrants (9), kids & family (8), faith-based (7), plus travel, women-first, climate, seniors and more. The "better bank for everyone" pitch is dying; the "right bank for someone" pitch is compounding — see the full niche map.
8. No-KYC remains a rounding error — by design
Exactly 12 of 357 work without identity checks, all of them self-custodial wallets, none of them able to issue a card without KYC. The line is structural: software that never holds funds escapes intermediary regulation; everything touching fiat doesn't. We mapped the whole territory here.
9. Transparency is still the industry's weak spot
We could verify official terms-of-service documents for only 126 of 357 — and this is after an audit in which we repaired 60 dead legal links and removed 53 that no longer resolve anywhere. For an industry holding people's money, the state of basic legal-document hygiene is genuinely poor. (Our rule: unverifiable links get removed, never guessed — a dead terms link is worse than none.)
10. The categories are dissolving at both ends
Traditional players are adding stablecoin rails; web3-native apps are acquiring e-money licences and IBANs; hybrids already straddle the line. We expect our own three-way classification to get harder to maintain every quarter — which is the finding, really: the interesting boundary in banking is no longer bank vs fintech. It's custodial vs self-custodial, and it cuts through every category we track.
Method, in one paragraph
neobankbeat tracks consumer- and SMB-facing digital-first banking apps: traditional challengers, hybrid fiat+crypto apps, and self-custodial money apps with banking features. Verified-active only — defunct entities and pure infrastructure/BaaS are excluded. Fields are compiled from public sources: filings, licence registers (ESMA, EBA, FCA, NMLS), official terms documents. Unverifiable fields are null, never fabricated. The dataset is MIT-licensed and machine-readable at /data.json; the full methodology is here. Cite freely with a link and the as-of date (July 2026).