← All News

Nvidia Unveils AI Blueprint To Help FIs Replace Hundreds Of Models With One

By CU Today Staff —

SAN JOSE, Calif.—Nvidia has unveiled a new blueprint aimed at helping FIs replace sprawling collections of artificial intelligence tools with a single AI model trained on transaction data that can be used for fraud detection, credit underwriting and risk management, according to PYMNTS.

PYMNTS reported the approach is modeled on how large language models learn from text, but instead uses financial transaction behavior as the training data. The move comes as Nvidia’s 2026 State of AI in Financial Services report found 65% of financial institutions are already using AI, while nearly 90% are either deploying or evaluating the technology. According to PYMNTS, the bigger challenge for many institutions is no longer adoption but managing a growing number of disconnected AI systems.

Rather than building separate models for individual tasks, Nvidia’s framework calls for training one model across an FI's transaction history and applying it to multiple use cases. PYMNTS noted that digital bank Revolut has already demonstrated the concept through its PRAGMA model, which was trained on 40 billion transactions across 25 million customers in 111 countries and now supports credit decisions, fraud detection and product recommendations from a single platform.

PYMNTS further reported that Mastercard is pursuing a similar strategy by developing a model trained on billions of anonymized payment-card transactions, while Stripe recently launched a payments foundation model trained on tens of billions of transactions. According to PYMNTS, Stripe said the model boosted detection rates for one common fraud type from 59% to 97% and helped the company block nearly $112 billion in fraudulent activity last year.

The business case for consolidation extends beyond operational efficiency, PYMNTS reported. Maintaining separate AI models for separate functions creates ongoing retraining and maintenance costs, while unified models can apply insights learned in one area across multiple use cases. Nvidia said the blueprint is intended to provide a starting point for institutions that lack the scale of companies such as Revolut, Stripe or Mastercard, with consulting firms including Infosys, EXL and Thoughtworks helping FIs integrate the approach into existing lending, servicing and compliance operations.

Originally reported by CU Today.