Nvidia backs public-sector AI projects across Europe

NVIDIA and its partners are rolling out a series of public-sector artificial intelligence projects across Europe, spanning the European Commission and national initiatives in Poland, the UK, Germany, France and Greece.

The projects focus on open models, locally controlled infrastructure and synthetic datasets for public administration. They cover healthcare, employment services, multilingual document analysis, collaborative software and systems for classified information.

Commission platform

In Brussels, Berlin-based deepset is expanding its Haystack Enterprise Platform, which underpins AI@EC, the European Commission's shared platform for building and running AI applications on commission-managed infrastructure.

The update brings agent development, execution, optimisation and governance into one system. It also adds tools to compare operating costs, assess answer quality and trace faults.

NVIDIA Triton Inference Server supports the shared services used to find and rank relevant information. The platform reflects requirements seen in deployments such as AI@EC, where reusable components, role-based access, audit trails and guardrails support applications including document summarisation, knowledge management and multilingual policy analysis.

Poland focus

In Poland, Deloitte is involved in projects covering healthcare and employment services. The country's eHealth Centre is building an Intelligent Services Platform for medical imaging using NVIDIA DGX BasePOD and the MONAI framework.

The system is intended to help physicians analyse X-rays, CT scans and MRI images, detect abnormalities and prioritise urgent cases, with the aim of cutting waiting times for results.

Separately, Poland's Ministry of Family, Labour and Social Policy is developing what it describes as an agentic factory on NVIDIA-accelerated computing infrastructure. It is designed to support more than 400 labour offices and run models including PLLuM, GPT OSS and Inkling for document search, summarisation and digital assistants.

NVIDIA is also working with SpeakLeash in Poland and the University of Ljubljana in Slovenia on country-specific additions to its synthetic personas programme. Localised datasets are expected to follow.

UK dataset

In the UK, London-based Locai Labs has released Nemotron-Personas-UK, a dataset of 1 million synthetic personas. It is intended to reflect the country's demographic and geographic diversity while reducing privacy risk.

The dataset is aimed at developers testing whether public-service models produce consistent results across different communities. Potential uses include healthcare evaluation and bias testing across age groups, education levels, and rural and urban populations.

Germany projects

Germany hosts several of the initiatives. Materna is introducing Sinfonie, an on-premises platform that combines generative AI, voice interaction and 3D avatars, with NVIDIA Audio2Face available for animation.

The platform stems from a project for a German state administration. It is designed to help employees access institutional knowledge through spoken questions while keeping data on their own infrastructure.

A separate effort targets organisations handling classified information. Secunet, NVIDIA, Red Hat and deepset are developing a sovereign cloud offering based on secunet's SINA Cloud foundation, which includes security components authorised for use with classified information.

The proposed service combines NVIDIA computing and Nemotron open models, Red Hat OpenShift AI for deployment and management, deepset's Haystack software for AI agent orchestration, and infrastructure run by secunet in German data centres.

Germany is also part of a research effort known as the Sovereign Open Source Foundation Models initiative, or Soofi, which is developing AI models for German industry and the public sector. This includes Nemotron-Personas-Germany, an open dataset of 1 million records, each containing six synthetic personas grounded in the 2022 census across all 16 federal states and 400 districts.

Created with NVIDIA NeMo Data Designer without using real people's private data, the dataset is designed to help developers test whether public-service models perform consistently across regions, age groups and communities. It can also be used with NeMo User Sim to evaluate and train models against simulated users based on synthetic German personas.

France and Greece

In France, CANUT, the public purchasing body for digital services, has selected LINAGORA and Docaposte to provide a sovereign collaborative workspace built on Twake.ai. More than 2,500 eligible organisations, including municipalities, schools and healthcare institutions, can use the framework without running their own tender.

The workspace includes email, document-sharing and collaboration tools for public employees. NVIDIA technology is being used for speech transcription, and the infrastructure is hosted in France.

In Greece, KIEFER is developing an agentic factory built on NVIDIA DGX SuperPOD with DGX GB200, alongside Sophea open-weight models and Greek-language tools for knowledge retrieval, speech and digital assistants. The project is intended to make institutional information easier to find and use for Greek-speaking public-sector teams, while allowing organisations to control where agents run and how data is handled.

Wider push

The wider European effort also includes work on NVIDIA's Nemotron-Personas collection, described as partner-grounded datasets behind NeMo User Sim, the company's open-source library for population-scale user simulation in local languages. The aim is to let governments evaluate models, adapt them to local public services, apply their own safeguards and keep sensitive data inside controlled environments.

Alongside the government projects, NVIDIA said its Open Models Codefest had identified five leading teams working on public-sector challenges including medical reasoning, AI tutoring and model governance: Pleias, Zehnlab.ai, Disseqt, Algoritmica and ABA Technology. Deutsche Telekom and T-Systems provided the computing infrastructure for the teams to train and deploy large language models.