← Embedded Finance Index 2026

Vault Core

Can a smart-contract-based, cloud-native core replace legacy mainframe banking infrastructure at scale?

Founded2014
HQLondon, United Kingdom
FoundersPaul Taylor
Latest roundSeries B, 2020 ($125M)
IndustryInfrastructure / Core Banking
The story

Thought Machine was founded in 2014 by former Google engineer Paul Taylor to build a cloud-native core banking platform — Vault Core — based on smart-contract-defined financial products. The strategic bet is that incumbent banks running mainframe cores need a fully programmable replacement, and that new banks (Zopa, SEB's UNQUO, Atom) want to launch on modern infrastructure. The company has steadily layered Vault Payments alongside Vault Core and built an ecosystem of pre-integrated partner connectors (Marqeta, Currencycloud, Wise, Salt Edge) so customer banks can assemble a full BaaS-grade stack on top of Vault.

Last 12 months
2025-03
2025-11
2026-05
Product timeline
2014
Thought Machine founded in London by Paul Taylor· founding
2018
Lloyds Banking Group partnership and Series A investment· partnership
2019
Opened Asia Pacific office in Singapore; SEB launches UNQUO on Vault Core· expansion
2020
Joined Banking Industry Architecture Network (BIAN); raised $125M Series B· funding
2025
General Bank of Canada adopts Vault Core· customer_win
2026
US Senate Federal Credit Union adopts Vault Core and Vault Payments· customer_win
Same company
Vault Core was analysed under more than one name on the same domain, so its stack is split across these dossiers. Each was analysed independently and may name providers the others do not.
Thought Machine3 layers →Vault2 layers →
The stack
8 layers disclosed in Vault Core’s embedded-finance stack
Payments / PSP
Vault Payments (self-operated)
Banking / BaaS
in the full report
Card issuing
in the full report
Ledger
in the full report
Lending
in the full report
Open banking
in the full report
Unified APIs
in the full report
FX & payouts
in the full report
Accounting gap: none
10 more providers across 7 layers, with confidence-scored evidence and source URLs, are in the Embedded Finance Index report.