Thought Machine and AWS target core banking's COBOL problem with AI

Thought Machine, the London-based cloud-native core banking vendor, has partnered with Amazon Web Services to release a joint AI-powered migration pipeline that the pair say can convert legacy mainframe code into tested, cloud-native financial products in a fraction of the time previously required. The announcement targets one of the most stubborn bottlenecks in global financial services: the tens of thousands of institutions still running core operations on ageing COBOL mainframe infrastructure estimated, industry-wide, to underpin trillions of dollars in daily transactions.

The product, Vault Forge, powered by Amazon Bedrock's frontier model infrastructure, works in four automated stages: discovery and reverse-engineering of legacy application logic via AWS Transform; consolidation of redundant product variants; synthesis of Python-native financial products by specialised AI agents; and human-in-the-loop validation gates before sandbox deployment. The critical design choice, which separates this approach from straightforward code-transpilation tools, is that it targets business intent rather than raw code. Rather than rewriting COBOL line-by-line, the pipeline extracts interest calculations, fee schedules, repayment workflows, and account lifecycle rules as clean, high-level specifications before re-expressing them in Thought Machine's Vault platform architecture.

Why the economics of migration have stalled modernisation

Core banking modernisation has long been described as the financial sector's equivalent of performing open-heart surgery while the patient runs a marathon. Traditional approaches require banks to untangle product logic that has been hardcoded directly into database infrastructure over decades, making each migration simultaneously slow, expensive, and operationally hazardous. Thought Machine's CEO Paul Taylor framed the ambition bluntly: "We are fundamentally rewriting the economics of core modernisation. For the first time, banks can deploy AI to rapidly convert complex legacy mainframe specifications into simulation-tested, cloud-native financial products."

The company says the joint solution compresses what are typically multi-year programmes into days for the initial proof-of-concept phases, an efficiency claim that, if validated at scale, would materially alter the cost-benefit calculus for institutions that have deferred modernisation precisely because the upfront disruption outweighed the projected gains. Thought Machine is supporting adoption through a Core Modernisation Accelerator programme that embeds its engineers directly within bank teams throughout delivery.

Convergence implications: cloud infrastructure meets financial regulation

The strategic read-across for cross-sector investors extends well beyond the immediate fintech modernisation narrative. AWS's involvement signals that hyperscaler cloud platforms are moving aggressively up the value chain inside regulated financial services, shifting from infrastructure provider to active transformation partner. That repositioning has material implications for the managed-services and systems-integration incumbents (think Accenture, IBM, Infosys) that have historically owned the professional-services revenue attached to core banking programmes.

Regulatory tailwinds are also shaping the timing. Supervisory bodies across the UK, EU, and US have intensified scrutiny of operational resilience and third-party technology dependencies, creating pressure on boards to retire legacy systems that concentrate risk. An AI-guided pipeline that operates entirely within the bank's own secure cloud environment, and provides auditable validation gates at each stage, is explicitly designed to answer those governance objections rather than sidestep them. Charith Mendis, AWS's Head of Worldwide Banking, noted that "customer expectations and regulatory demands are evolving at pace," positioning the joint solution as a compliance-readiness tool as much as a cost-reduction play.

For capital allocators tracking the intersection of AI infrastructure spend and financial services transformation, this deal illustrates a broadening pattern: hyperscalers are no longer competing purely on compute price, but on domain-specific AI tooling that embeds them as strategic partners inside heavily regulated verticals. Thought Machine, which has raised more than £500m and counts over 65 banks among its clients, provides AWS with a credentialled distribution channel into Tier 1 institutions that would otherwise require years of direct enterprise sales cycles to penetrate. The partnership is therefore as much a go-to-market convergence play as it is a product announcement.