Denmark just launched the Gefion AI Supercomputer (GAIS), a significant leap into the world of artificial intelligence and quantum computing. This makes Denmark the front-runner among Nordic countries, revolutionizing how businesses and industries leverage AI.
The GAIS is built on the Nvidia DGX SuperPOD platform, packing an impressive 1,528 Nvidia H100 Tensor Core GPUs, all linked via cutting-edge Nvidia Quantum-2 InfiniBand networking. This powerhouse is designed to tackle high-performance computing workloads for various applications.
The Danish Centre for AI Innovation (DCAI) operates GAIS, funded primarily by two major partners: the state-controlled Export and Investment Fund of Denmark (EIFO) and the Novo Nordisk Foundation (NNF). With NNF holding an 85% stake and EIFO the remaining 15%, this collaboration has pledges worth €95 million for developing GAIS in its initial phase. NNF, with nearly €163 billion in assets, contributed €81 million last year, they distributed over €1.3 billion in grants to support research and innovation.
Nadia Carlsten, CEO of DCAI, aims to have GAIS ready for researchers and industries by June 2025. “We completed this complex project in just six months, and it’s now ready for customers eager to explore new possibilities,” she stated, highlighting the anticipation surrounding GAIS.
An exciting aspect of the project is its revenue model. All earnings generated will go back into the DCAI-managed initiative, and GAIS will operate entirely on renewable energy.
Eviden, connected to the French IT giant Atos, constructed this supercomputer. During its pilot phase, GAIS will cater to selected clients using AI for diverse applications, from quantum computing to green energy advancement. The Danish Meteorological Institute (DMI) is one of the first agencies granted access, as they aim to develop a more accurate weather prediction model based on 33 years of data.
Morten Bødskov, Denmark’s minister of business and financial affairs, talked about the immense potential of GAIS. “This supercomputer will not only boost innovation but also help Danish companies craft tailored solutions and enhance their competitiveness on a global scale,” he said.
Before GAIS, Denmark lacked access to GPU-accelerated supercomputers, hindering the growth of its AI research scene. Now, Peder Lundquist, CEO at EIFO, believes GAIS will usher in new energy for AI-led research and innovation, stating, “We are already significant in AI development, and this technology will take us even further.”
The collaboration between EIFO, NNF, and DCAI is timely, as Danish agencies seek to ramp up their investment in deep tech, focusing on areas like AI, quantum computing, and life sciences. Notably, GAIS has earned a spot as the 21st most powerful supercomputer in the world, raising Denmark’s profile in the global AI landscape.
Carlsten remarked, “Being ranked 21st is a milestone for us, showcasing what we can achieve in scientific research and innovation previously thought impossible here.”
In preparation for GAIS, the DCAI ran a competition to discover innovative AI projects, leading to six chosen initiatives for the pilot phase. For example, the University of Copenhagen plans to explore quantum algorithms for understanding molecular interactions, while the Technical University of Denmark will investigate CO2 reduction methods using machine learning. The DMI will employ GAIS to enhance atmospheric predictions through sophisticated environmental models.
