Banks Train Single AI Model on Transaction Data to Replace Fragmented Systems
Nvidia has released a blueprint for banks to consolidate their scattered AI systems into one model trained on transaction data they already own. The single model handles fraud detection, credit scoring and risk assessment together instead of separately.
The architecture works like large language models, except the training material is financial behavior rather than text. A payment at midnight looks different when it's the fourth transaction in ten minutes, from an unfamiliar device, in a city where the customer has never purchased before.
Nvidia's 2026 State of AI in Financial Services report found 65% of financial institutions already use AI, with nearly 90% deploying or assessing it. The real bottleneck isn't adoption-it's the sprawl. Most banks have built too many disconnected AI systems, and fragmentation is now slowing them down.
What Consolidation Looks Like in Practice
Revolut published results from PRAGMA in April, a model trained on 40 billion transactions across 25 million customers in 111 countries. One system now handles credit decisions, fraud detection and product recommendations that previously required separate ones.
Tadas KriΕ‘ΔiΕ«nas, Revolut's head of group credit data science, said the initiative cut the time needed to set up a new use case from weeks or months to "no time required for it at all."
Mastercard is building toward the same outcome at larger scale. The payments network is developing a model trained on billions of anonymized card transactions, including fraud, chargebacks, merchant data and loyalty activity. Personal identifiers are stripped before training begins.
Stripe launched its own payments foundation model trained on tens of billions of transactions. The model raised detection rates for one common type of payment fraud on large businesses from 59% to 97%. Stripe blocked close to $112 billion in fraud last year and cut average fraud rates by 38%.
The Competitive Advantage of One Model
Banks that maintain separate AI models for separate problems pay a compounding cost. Each new market requires retraining from scratch. Each new use case adds another system to maintain. None of those systems can use what the others have learned.
Adyen, which processes $1 trillion in payments annually, said even a 0.1% improvement in the rate at which payments successfully clear translates to significant incremental revenue for merchants. Shared intelligence makes those gains possible across the board, not just in one product line.
A bank that consolidates its AI into one model trained on its full transaction history can move faster, make better decisions and extend those gains to new problems without starting over.
Bringing the Model to Smaller Institutions
Not every institution has Revolut's data or Stripe's engineering capacity. Nvidia's blueprint is designed to give smaller institutions a starting point, letting teams build on their own transaction data without rebuilding systems from scratch.
Services firms including Infosys, EXL and Thoughtworks are helping banks integrate the approach into existing credit, servicing and compliance environments.
What Revolut, Stripe, Mastercard and Adyen share is an asset competitors can't replicate: years of their own customers' transaction history. That data advantage compounds as the model learns from more transactions.
For finance professionals evaluating AI strategy, understanding AI for Finance applications and their implementation is essential. Those in leadership roles may also benefit from an AI Learning Path for CFOs to assess how consolidated models affect organizational strategy.
Your membership also unlocks:
