- Electrical infrastructure partnership delivers innovative solution for AI’s power demands at a greenfield 400 MW data center campus in North Dakota, US
- New medium voltage power architecture with ABB’s HiPerGuard UPS helps data centers increase efficiency and reliability
- ABB’s supply chain scale and production capacity facilitate AI-ready data centers
Credit: ABB
Applied Digital, a builder and operator of next-generation data centers, has launched an infrastructure partnership with ABB at the company’s greenfield 400 MW campus in North Dakota, United States. The collaboration will innovate fast, reliable, energy-efficient solutions to meet the needs of rapidly advancing artificial intelligence (AI) capabilities. As part of this long-term partnership, the first orders were booked in Q4 2024 and Q1 2025. Further financial details of the partnership were not disclosed.
Central to the partnership is a new medium voltage power architecture using ABB’s HiPerGuard Medium Voltage Static Uninterruptible Power Supply (UPS), an industry-first technology. HiPerGuard solutions help AI data centers increase power density while compressing electrical plant footprints and increasing energy efficiency. The partnership covers the complete design and development of the site’s electrical infrastructure, optimized to improve the efficiency and resilience of major AI facilities.
Conventional data centers use low voltage power distribution and UPS systems. Switching to medium voltage for the UPS system and power distribution enables data centers to scale more efficiently, expanding in 25 MW blocks. With fewer parts and cabling, installation is simpler and reliability is higher. Running operations with higher voltages, lower conversion losses and less heat generation supports greater energy efficiency.
"What we are building with ABB is going to completely change the way our industry designs large-scale data centers. By leveraging ABB's cutting-edge technology and global reach, we are able to redefine the electrical infrastructure landscape and meet the growing demands to support large scale AI Factories,” said Todd Gale, Chief Development Officer, from Applied Digital.
Massimiliano Cifalitti, Smart Power President, ABB Electrification, said: "Partnerships like this are accelerating the timeline for AI-ready data centers with more competitive, resilient power infrastructures. ABB’s technology is designed for peak performance with dedicated capacity across our global production network and industry-leading innovation. ABB teams are ready to deliver at scale and at speed.”
ABB’s early investment in next-generation medium voltage UPS technology is paying off. Early adopters of ABB's HiPerGuard solution include large-scale data centers and research facilities in North America and Europe that want more energy efficiency and higher power density. With AI’s power densities driving an overhaul of power architectures, ABB’s medium voltage UPS solution is unlocking a more efficient future for data centers.
ABB is a global technology leader in electrification and automation, enabling a more sustainable and resource-efficient future. By connecting its engineering and digitalization expertise, ABB helps industries run at high performance, while becoming more efficient, productive and sustainable so they outperform. At ABB, we call this ‘Engineered to Outrun’. The company has over 140 years of history and around 110,000 employees worldwide. ABB’s shares are listed on the SIX Swiss Exchange (ABBN) and Nasdaq Stockholm (ABB).www.abb.com
Applied Digital (Nasdaq: APLD) develops, builds and operates next-generation data centers and infrastructure. Different by design, Applied Digital’s purpose-built facilities are engineered to unleash the power of accelerated compute and deliver secure, scalable and sustainable digital hosting, enabling CSaaS and GPU-as-a-Service solutions. Backed by deep hyperscale expertise and a robust pipeline of available power, Applied Digital accommodates AI Factories and beyond to support AI/ML, blockchain and high-performance computing (HPC) workloads.