A total of 15,128 NVIDIA H100 Tensor Core GPUs (pictured here) will be used in the Denmark AI Innovation Center. The hardware can support the development of advanced AI applications, from protein structure prediction to quantum computing research.
A collaboration between the Novo Nordisk Foundation, the Export and Investment Fund of Denmark (EIFO), and NVIDIA will establish a national AI Innovation Centre in Denmark focused on accelerating research and innovation in fields including healthcare, life science, and quantum computing. The initiative is led on the Danish side by the Novo Nordisk Foundation, which has committed roughly DKK 600 million (around $90 million) toward the initial costs of the center, and the Export and Investment Fund of Denmark (EIFO), which has contributed another DKK 100 million.
In a press briefing, Kimberly Powell, NVIDIA’s VP of healthcare, highlighted Denmark’s “incredibly rich digital healthcare data set” as a key asset for the Center’s work in drug discovery and precision medicine. Denmark’s existing strengths in healthcare data and life sciences research position the country well for success in these areas.
“In the current geopolitical climate, it is important that we strengthen our strategic positions,” said Morten Bødskov, Danish Minister of Industry, Business and Financial Affairs, in a press release. The government aims to elevate Denmark’s global standing in cutting-edge technology, Bødskov said. “I am pleased to see both public and private sectors in Denmark investing in transformative technologies like supercomputers, which will undoubtedly boost our competitive edge on the world stage.”
NVIDIA’s generative AI platforms to power Denmark’s Innovation Center
“Through our collaboration agreement, we’re going to be working with their collaborators, bringing all of our generative AI platforms to their sovereign AI computer and application development, including NVIDIA BioNeMo, NVIDIA CUDA Quantum,” Powell said. “We’re going to be working on really important application areas like accelerating protein-based drug discovery and design.”
CUDA Quantum is an open-source platform for integrating and programming quantum processing units (QPUs), GPUs, and CPUs in a single system for hybrid quantum-classical computing. BioNeMo is a framework for building AI models in biology and chemistry, with a strong focus on protein structure prediction and drug discovery. These NVIDIA technologies are expected to accelerate research in areas like protein structure prediction, drug design, and hybrid quantum-classical computing.
“This collaboration with NVIDIA and the resulting national AI innovation center can help Denmark’s brilliant researchers and innovators rise to the next level,” said Mads Krogsgaard Thomsen, CEO of the Novo Nordisk Foundation, in a press release. “The transformative potential of this initiative aligns perfectly with our strategic priorities in AI and healthcare.”
The GEFION, a potential top 25 supercomputer
At the heart of the center will be an AI supercomputer named GEFION, which is poised to be a “top 25 supercomputer,” Powell said. Inspired by the Norse goddess of foresight and abundance, GEFION will feature 15,128 H100 Tensor Core GPUs and Quantum-2 InfiniBand networking, The NVIDIA H100 Tensor Core GPU is a processor model designed for large-scale AI and high-performance computing (HPC) workloads that can deliver up to 6x throughput improvement over the previous generation A100 GPU. The considerable number of GPUs will enable the Denmark AI Innovation Center to tackle computation-intensive tasks, from protein structure prediction to quantum computing research, with speed and efficiency.
Eviden, an Atos Group company and European leader in advanced computing, will deliver, install, and configure the Gefion supercomputer. The data center hosting Gefion will be provided by Digital Realty and is designed for sustainability, using 100% renewable energy.
Complementing the raw processing power of the H100 GPUs, NVIDIA’s Quantum-2 InfiniBand networking will provide the high-speed, low-latency interconnect needed to accommodate the processing horsepower of Gefion’s massive GPU array. Quantum-2 is one of the most advanced end-to-end networking platform ever built, offering extreme performance for large-scale data processing tasks in data centers and research institutions. including for scientific computing applications.
The supercomputer will have the highest level of security and support Danish data sovereignty, allowing it to handle sensitive data. Differential payment models are being developed to balance commercial and academic use of the center, with the aim of financial sustainability.
Digital Realty’s AI-ready data center in Denmark will host the supercomputer. The facility prioritizes sustainability with a design built around renewable energy sources and Gefion will run on sustainable energy.
Gefion is expected to be ready for pilot projects before the end of 2024, with the center fully operational in early 2025.
The most recent NVIDIA BioNeMo foundation models can help drug discovery groups design drug molecules virtually. [Credit: NVIDIA]
Denmark’s AI commitment
Denmark’s commitment to tapping AI for societal benefit is not new. In 2019, a McKinsey report titled “An AI Nation? Harnessing the opportunity of artificial intelligence in Denmark” explored the potential of AI in the country, noting strengths, such as a digitally advanced public sector and world-class public data.