Skip to content
Sıfırıncı Dakika
Latest

Leading European financial firms build proprietary intelligence on open models

AINew1 min read
NVIDIA Newsroom

In brief

At GTC Berlin, NVIDIA showcased European banks, payment companies and financial data providers running production-scale AI on open weight models within their own infrastructure. Revolut is rebuilding its technology stack on a single foundation model called PRAGMA, powered by NVIDIA accelerated computing.

  • Open weight models run on organizations' own infrastructure
  • Revolut is rebuilding its stack on a single foundation model, PRAGMA
  • Powered by NVIDIA accelerated computing and cuDF
  • Announced at GTC Berlin

Finance turns to its own models

At its GTC event in Berlin, NVIDIA showed how European banks, payment companies and financial data providers are using AI at production scale. The common thread is running open weight models on their own infrastructure.

An open weight model means the trained model's weights can be downloaded and run inside a company. That lets sensitive financial data stay in-house while the model is still used.

Revolut moves to a single model

The standout example is Revolut. The company is rebuilding its technology stack on a single foundation model it calls PRAGMA. The model runs on NVIDIA accelerated computing, and the source also references NVIDIA's cuDF library.

A foundation model means different business processes are fed from one AI core. The approach aims to give companies a layer of intelligence customized with their own data.

Why it matters

Finance has been cautious about closed, cloud-based models because of data privacy and regulatory pressure. Running open weight models in their own data centers offers a concrete example of how that balance can be struck.

Why it matters

A concrete example for readers curious about the balance between data privacy and AI adoption in finance, showing how open models are positioned in enterprise production.

Sources

Related stories