Technology

Danish supercomputer to drive innovation


The Gefion AI Supercomputer (GAIS) mission, which delivers Denmark’s first synthetic intelligence (AI) turbo-charged supercomputer, has positioned Denmark as essentially the most superior of the Nordic area’s quantum computing investing nations.

It additionally serves to speed up using AI to drive innovation throughout Denmark’s enterprise and industrial sectors.

Constructed on the Nvidia DGX SuperPOD AI supercomputer, GAIS is powered by 1,528 Nvidia H100 Tensor Core graphics processing items (GPUs) and interconnected utilizing Nvidia Quantum-2 InfiniBand networking.

GAIS might be used to sort out high-performance computing (HPC) workloads utilizing a full-stack datacentre platform.

The Danish Centre for AI Innovation (DCAI), the corporate established to function Gefion, is drawing a big a part of its funding from two principal traders and house owners, the state-controlled Export and Funding Fund of Denmark (EIFO); and Danish philanthropic establishment Novo Nordisk Basis (NNF). The NNF acquired an 85% fairness holding in DCAI and the EIFO owns the remaining 15%.

This groundbreaking private-public partnership has dedicated €95m in startup improvement capital to the GAIS mission over its preliminary stage. The NNF is contributing €81m and the EIFO the rest. The previous, a charitable basis, has a web value of just about €163bn. Final 12 months, it distributed grants of over €1.3bn to qualifying analysis and innovation initiatives.

The DCAI’s aim is to have the GAIS useful resource typically obtainable to researchers, trade and different chosen customers by June 2025, stated Nadia Carlsten, the corporate’s CEO. “Gefion’s launch is a exceptional achievement given it took solely six months from disclosure of the mission to completion for what’s a posh state-of-the-art supercomputer,” she stated. “We now have a number one functionality that’s prepared for purchasers to begin testing new use circumstances on. Gefion is the game-changer that many innovators in
Denmark
had been ready for.”

Reinvesting revenues

Below an settlement reached between the house owners and stakeholders, all revenues generated from GAIS might be reinvested within the DCAI managed mission. The turbo-charged laptop might be operated on 100% renewable vitality sources.

The GAIS was constructed by Eviden, the high-performance computing and methods integration subsidiary of French IT agency Atos.

The pilot part of GAIS will give attention to making it ready-to-use for chosen clients which might be utilizing AI to speed up innovation in a broad vary of areas; together with quantum computing, drug discovery, societal challenges, deep studying and the transition to inexperienced vitality.

The Danish Meteorological Institute (DMI) is among the many first scientific companies chosen to have entry to GAIS in the course of the pilot check stage. The DMI’s meteorologists are utilizing GAIS to develop a Danish climate mannequin, based mostly on 33 years of recorded knowledge, that may predict climate and local weather occasions with larger accuracy.

The “superpower” elements in GAIS will current Danish enterprises and organisations with solely new alternatives, stated Morten Bødskov, the Danish trade, enterprise and monetary affairs minister.

“Gefion has the potential to drive innovation and advance the inexperienced transition,” he stated. “It’s going to allow tailored options, and strengthen the aggressive standing of Danish corporations in international markets. At a time of heightened worldwide competitors, this AI supercomputer, and its groundbreaking know-how, can propel Danish companies into the long run. Considerably, Gefion will assist Danish corporations and researchers develop progressive options to future challenges.”

Denmark had no GPU-accelerated supercomputers in use earlier than the launch of GAIS. This lack of entry to supercharged computing energy was broadly regarded by Danish legislators and know-how sector leaders as a big roadblock to additional bolstering the nation’s AI-based analysis ecosystem.

GAIS will present recent momentum to drive future AI-based analysis, innovation and progress alternatives for enterprises of all sizes in Denmark, stated Peder Lundquist, CEO at EIFO.


Denmark is already an necessary participant in AI-based analysis and enterprise improvement,” he stated. “The facility and innovation behind Gefion, and the chances it brings, can speed up improvement many occasions over. EIFO’s funding in DCAI will help Danish companies in key areas like inexperienced transition and life science. It’s going to additionally speed up the commercialisation of analysis.”

Deep tech innovation

Collaboration between the EIFO, NNF and the DCAI is going on at a time when the state company is additional scaling-up its partnership-based investments in progressive deep tech initiatives, notably inside its strategic focus areas, AI; quantum computing; the inexperienced transition; life sciences; and the commercialisation of analysis.

The Gefion AI supercomputer is formally ranked twenty first on the High 500 checklist of the world’s strongest supercomputers. It’s going to allow analysis and innovation at ranges beforehand unattainable in
Denmark
, stated Carlsten.

“Gefion’s number-21 rating on the High 500 checklist is a milestone for DCAI and Denmark’s international standing in AI and supercomputing,” she stated. “This recognition shouldn’t be solely proof of the superior know-how we’re in a position to present to our clients, nevertheless it additionally displays our ambitions to drive vanguard scientific analysis and innovation that was beforehand not doable to do in Denmark.”

As a part of the lead-in to the central mission, the DCAI ran a contest in 2024 to determine progressive AI use circumstances that had the aptitude to completely harness the potential of GAIS. The DCAI chosen six out of a complete of fifty proposals obtained to run in the course of the pilot part.

Among the many accredited pilot initiatives is a proposal from the College of Copenhagen to check large-scale distributed simulation of quantum algorithms for quantifying molecular recognition processes.

The Technical College of Denmark will use GAIS to discover CO2 discount strategies in non-metal formate dehydrogenase utilizing machine-learned power fields. As well as, the Danish Meteorological Institute will make use of GAIS to develop proficient atmospheric prediction with clever environmental community fashions.