Core banking modernization: Mambu vs Thought Machine vs Temenos

Core banking modernization is the process of replacing or transforming the central system that manages a bank's financial ledger, accounts, transactions, and product calculations — typically from a legacy mainframe or on-premises system to a cloud-native platform. The goal is to reduce maintenance costs, enable faster product innovation, support real-time digital banking features, and comply with modern regulatory requirements (open banking APIs, DORA). It is one of the most complex and high-risk technology programmes a bank undertakes, typically taking 3–7 years for large institutions.

Mambu is a SaaS cloud-native core banking platform designed for digital banks, neobanks, and bank-as-a-service providers. Its composable architecture allows banks to combine Mambu's product engine with third-party fintech services for KYC, payments, and lending. Mambu is best suited for organisations launching new digital banking propositions that need to go live quickly (weeks to months), don't want to manage infrastructure, and need a highly configurable product engine for retail banking, lending, and SME products. It is used by N26, OakNorth, and many neobanks globally.

Thought Machine Vault's key differentiator is its Smart Contract product engine — all financial products are defined as Python-based Smart Contracts that describe how the product behaves over its lifetime. This programmability allows banks to define any financial product without depending on the vendor's product development roadmap. Vault runs on a single event-driven Universal Product Ledger that handles all product types, reducing integration complexity. Its clients include JPMorgan Chase, Lloyds Banking Group, and Standard Chartered — banks that chose Vault specifically for the control its programmable architecture provides.

Yes, Temenos Transact (with 3,000+ bank clients globally) remains highly relevant, particularly for banks requiring broad product coverage (retail, corporate, private banking, treasury, capital markets in one platform), deep multi-jurisdiction regulatory support, or global scale. Temenos has modernised its architecture for cloud deployment on AWS, Azure, and GCP. However, for digital-first or neobank programmes where speed to market and API-first architecture are the priority, Mambu or Thought Machine are typically faster and more suitable. Temenos is most competitive for complex global banks where Mambu's product range or Thought Machine's client base are insufficient.

The strangler fig pattern (named after the vine that gradually replaces its host tree) migrates products and customers to the new core incrementally rather than in a single big-bang cutover. The new core is deployed alongside the legacy system; new products are launched on the new core; then existing customers and products are migrated segment by segment over 3–7 years. The legacy system handles decreasing volumes until it is decommissioned. This approach is lower risk than big-bang migration — a problem with one product's migration doesn't affect the rest of the bank — but requires running two cores simultaneously with reconciliation between them.

Core banking replacement costs vary enormously by bank size, complexity, and approach. For a neobank or digital bank using Mambu on a greenfield basis, total implementation costs (software, implementation, integration, testing) typically run €2–10M. For a mid-sized retail bank doing a full replacement, €50–200M over 3–5 years is typical. For a large global bank doing a multi-entity replacement, costs can exceed €500M–1B. These figures include software licensing, implementation services, data migration, integration development, testing, parallel running costs, and change management — not just the software licence.

Composable banking (associated with Mambu's "composable banking" approach and Gartner's "composable architecture" concept) is the practice of assembling a banking platform from modular, best-of-breed components — a core banking engine (Mambu), a KYC/onboarding service (Onfido), a payments orchestrator (Modulr), a lending engine, a card processing service — rather than relying on a single monolithic all-in-one system. Each component connects via APIs, allowing the bank to swap individual components as technology evolves without replacing the entire stack. This approach reduces vendor lock-in and allows faster adoption of fintech innovation compared to traditional all-in-one core banking systems.

TSB's 2018 core banking migration is the most cited failure in financial services technology history. TSB attempted a big-bang cutover from a Lloyds Banking Group legacy system to a new Proteo4UK platform (built on Sabadell's Proteo) over a single weekend. The migration catastrophically failed, locking 1.9 million customers out of their accounts for days to weeks, causing 80,000 customers to be switched to fraudulent accounts, and ultimately costing TSB over £330M. The key lessons: never attempt a big-bang core migration without extensive parallel running validation; test at production scale with production data volumes; have an executable rollback plan; and ensure executive accountability extends to post-migration support capacity, not just cutover execution.