NVIDIA has released freely downloadable AI course materials and outlined new routes to computing resources for European researchers and startups, joining a European Commission initiative intended to make advanced technology more accessible to the scientific community.
The package combines training, open models and software with collaborations involving national AI facilities in France, Spain and Sweden. It addresses practical barriers that NVIDIA identifies for smaller laboratories and early-career scientists: finding computing capacity, assembling research tools and developing the expertise to use them.
The European Commission launched the AI in Science Enablers Initiative at the AI in Science Summit in Dublin on October 15. Connected to RAISE, the Resource for AI Science in Europe, it brings private-sector contributions into a shared framework supporting research.
NVIDIA’s participation draws together existing resources and new activities. Its announcement includes a new courseware hub, developing accelerator programs and access already provided through European supercomputing centers.
“The impact we are aiming for is straightforward: lower the practical barriers, help more teams test good ideas and shorten the path from a scientific question to a result that can make a difference,” Rod Evans, NVIDIA’s EMEA Vice President for Supercomputing & AI Cloud Infrastructure, wrote on LinkedIn.
Course materials that universities can adapt
The NVIDIA Deep Learning Institute’s new open source courseware hub provides hands-on materials covering accelerated computing and deep learning. The resources are openly licensed and available without enrollment, fees or a completion deadline.
Their modular structure allows educators to adapt them for university courses, intensive training at supercomputing centers or developer events. NVIDIA also positions the materials as a route into its wider training and certification programs.
The offer is specifically for downloadable teaching and learning materials. Access to computing infrastructure is being developed through separate programs and partnerships.
“Europe’s ambition in AI in science can only become reality if we bring together the strengths of our research community, industry and other partners,” says Ekaterina Zaharieva, European Commissioner for Startups, Research and Innovation.
“Through the AI in Science Enablers campaign, we are turning this ambition into concrete commitments that give researchers access to the tools, expertise and resources they need to innovate.”
National AI facilities open routes for startups
NVIDIA is working with cloud partners and national AI factories, facilities providing computing infrastructure and related services, to expand access for startups and small and medium-sized enterprises.
Its Inception startup program supplies technical support and access to computing resources. In France, the AI2F facility has been collaborating with Inception for four months. NVIDIA describes the results positively but does not provide performance measures in the announcement.
In Spain, the BSC AI Factory is moving from a pilot to its first full accelerator program. NVIDIA says other European national AI factories have expressed interest in adopting the model in 2027.
Sweden AI Factory has also announced a collaboration to make its infrastructure and expertise available to Inception startups and smaller businesses. Target sectors include life sciences, materials science, autonomous systems and gaming.
For academic teams and startups conducting open research, NVIDIA says European supercomputing centers including CINECA and JSC already provide free access to its graphics processing unit infrastructure and services. NVIDIA supplements that access with technical office hours, bootcamps and hackathons where researchers can adapt models using their own datasets.
A European AI hackathon this month is being co-organized with Austrian Scientific Computing, the Leibniz Supercomputing Centre, the High-Performance Computing Center Stuttgart and The Cyprus Institute, for Pharos-CY.
Open models for specialized scientific work
The research support extends beyond access to machines. NVIDIA says it has released more than 250 open models, 250 open datasets and over 1,400 open source software projects, with models and datasets available through Hugging Face and GitHub.
These resources cover areas including healthcare, robotics, climate science and scientific research. Its Nemotron family includes models, training data and development recipes, while scientific tools include the BioNeMo Agent Toolkit, PhysicsNeMo libraries, Earth-2 models and ALCHEMI NIM microservices.
Together, the tools are intended to support work across biology, molecular design, weather, energy, engineering and materials science. Their usefulness depends on researchers applying them to the requirements of their own disciplines.
At NVIDIA’s Sweden AI Ecosystem Reception in Stockholm, Founder and CEO Jensen Huang emphasized that distinction.
“AI needs domain expertise. General intelligence is not sufficient to come into manufacturing, into drug discovery and become proficient,” he said. “You need to take that general intelligence and apply it for yourself and make domain experts.”
Sweden’s shared supercomputer brings industry into the picture
The European announcements also include Sferical One, a newly launched shared supercomputer built on NVIDIA systems. It is intended to allow researchers in Sweden and the Nordic region to customize and deploy Nemotron models in a secure environment.
The initiative is backed by a consortium anchored by AstraZeneca, Ericsson, Saab, SEB and Wallenberg Investments, with SKF as a strategic partner.
At a Stockholm event, representatives outlined potential applications spanning drug discovery, telecommunications, aerospace research and banking. Those were ambitions for the system, rather than demonstrated outcomes presented in the coverage.
Across Europe, NVIDIA says 35 supercomputers using its technology are currently in development. The company projects that these systems will equip more than 3 million researchers with new infrastructure. That figure describes anticipated reach as the systems are developed, rather than access already delivered.
Building AI agents also means restricting them
Developer events ahead of GTC Berlin have paired practical AI building with discussion of security. At Build-a-Claw sessions in London and Stockholm, participants created specialized AI agents, systems that can use tools and carry out tasks, without needing to write code.
Huang used the events to explain NVIDIA’s OpenShell software, which places agents within a controlled environment and limits what they can do.
“Whatever the task you give to your agent, you then give it minimum rights, minimum privileges, and then you monitor it to make sure it’s compliant with policy,” he told the London audience.
The safety discussion extended to researchers at the London Institute for Healthcare Engineering at King’s College London. Its AI Safety and Security Research Forum examined how individually secure agents can become insecure when working together, and how safeguards can keep pace with growing capabilities.
The forum followed NVIDIA’s previously announced commitment of computing access valued at $250 million over five years for UK research into AI agent safety and security.
NVIDIA’s European events continue at GTC Berlin on October 20–22. The Deep Learning Institute’s open course materials are already available for educators and researchers to download and adapt.
Leave a Reply