The University of Southern Denmark (SDU) is building an advanced, energy‑efficient data center in Sønderborg that will serve as Denmark’s new flagship hub for artificial intelligence and high‑performance computing (HPC), in a three‑way collaboration with Danfoss and Hewlett Packard Enterprise (HPE). The facility is designed not only to turbo‑charge Danish research in AI and data‑intensive science, but also to act as a real‑world demonstrator of how next‑generation data centers can slash energy use and emissions through cutting‑edge cooling and heat recovery.
The project builds on the SDU eScience Center’s role as a national HPC provider via the DeiC Interactive HPC service based on the UCloud platform, which has delivered cloud‑style supercomputing to universities across Denmark since 2019. With the new Sønderborg facility, SDU and its partners aim to show that sovereign, open‑source cloud technologies can coexist with some of the world’s most advanced accelerators for AI workloads—without defaulting to energy‑hungry, imported hyperscale designs.
Nvidia‑powered AI Mod POD anchors the compute platform
At the heart of the new data center is an HPE Data Center Services – AI Mod POD, a modular, factory‑integrated data hall optimised for AI and HPC that will host a dense cluster of next‑generation Nvidia hardware. The system will comprise 128 Nvidia Blackwell GPUs alongside high‑performance HPE Cray XD225v and HPE ProLiant XD685 servers, backed by HPE ProLiant DL325 and DL365 systems for supporting workloads.
High‑speed interconnect is provided by Nvidia Quantum‑2 InfiniBand networking, while HPE Networking CX6300 switches tie the platform into SDU’s wider infrastructure. The entire environment is tightly integrated with direct liquid cooling, a prerequisite for running such high‑density AI hardware within realistic energy and space envelopes. For researchers across SDU and other Danish universities, the new system promises a step‑change in capability for training large AI models, running complex simulations, and handling massive, potentially sensitive datasets.
Danfoss turns waste heat into a local energy asset
Danfoss’ role is to supply and continuously refine an advanced cooling and heat‑recovery ecosystem that wraps around the HPC installation, effectively transforming the data center into a controllable energy node within Sønderborg’s wider system. Using its expertise in HVAC, refrigeration and district‑energy technology, Danfoss is deploying direct liquid cooling and heat‑reuse modules that allow nearly all surplus heat from the data center to be captured and fed back into local heating networks.
Operational data from the live facility will be fed back to Danfoss engineers, who will use real‑time performance metrics to optimise cooling efficiency, heat‑recovery yield, and control strategies over time. In parallel, SDU’s compute specialists will adapt the UCloud platform and workload scheduling to favour energy‑saving usage patterns, aligning job placement and timing with the thermal and grid conditions that maximise overall system efficiency.
The initiative is supported by ProjectZero, a public‑private partnership in Sønderborg aiming to make the municipality’s energy system carbon‑neutral by 2029, and is framed around a three‑step strategy of reducing, reusing and renewing energy. By fully reusing waste heat, the data center is expected to become a net positive contributor to local energy resilience, rather than a passive consumer.
Sovereign, open‑source cloud underpins secure AI research
Crucially, the new facility will run on Danish‑developed open‑source cloud technology, UCloud, reinforcing what SDU calls Denmark’s “digital sovereignty” in AI and data‑driven research. UCloud, co‑developed over the past eight years by SDU’s eScience Center with partners at Aalborg University and Aarhus University, already serves around 20,000 users across all Danish universities, several public institutions and private companies.
By fully integrating the new AI and HPC resources into UCloud, SDU aims to give researchers from every discipline self‑service access to top‑tier accelerators in a secure, sovereign environment, including for projects that involve sensitive personal data. The platform’s user‑friendly interface is intended to lower barriers for humanities, social‑science and health researchers who may not be traditional HPC specialists but increasingly rely on machine learning and large‑scale data analysis.
“We are proud that with the new data center, we can combine research, sustainability, and technological innovation in one unified project,” said Thomas Buchvald Vind, University Director at SDU, calling it both an investment in future research capacity and a “concrete example” of how SDU supports Denmark’s green and digital transition.
A template for sustainable, sovereign digital infrastructure
For Danfoss, the project is a high‑profile test bed to prove that green data center technologies are commercially and technically viable at scale. “This collaboration allows us to demonstrate the power of cross‑industry collaboration in driving sustainable solutions and proves that sustainability, technological advancement, and commercial attractiveness can go hand in hand,” said Sune T. Baastrup, Chief Information Officer at Danfoss.
HPE, meanwhile, positions the initiative as a showcase for its modular AI and supercomputing portfolio in sovereign, research‑led settings. “We look forward to the scientific discoveries Danish researchers will make as a result of this collaborative effort,” said Carsten Nielsen, Vice President and Managing Director for HPE’s Nordic cluster, highlighting the blend of world‑class performance and sustainable operation.
Industry observers see the Sønderborg project as a marker of maturity for Denmark’s data center ecosystem, demonstrating how universities, technology vendors, and infrastructure players can jointly design digital capacity that is simultaneously high‑performance, energy‑aware, and nationally controlled. If successful, the partners hope the architecture can serve as an international reference model for energy‑efficient, sovereign AI data centers, aligning the rapid growth of computational power with the equally urgent imperative to decarbonise.