ABB Partners with NVIDIA to Power Next-Gen AI Data Centers
ABB Partners with NVIDIA to Supercharge AI Data Centers with Next-Gen 800VDC Power Systems
A groundbreaking ABB–NVIDIA partnership aims to reshape data center power for AI, increasing efficiency and enabling 1MW GPU racks worldwide.
ABB and NVIDIA partnership
Swiss engineering giant ABB has announced a strategic collaboration with semiconductor leader NVIDIA to develop gigawatt-scale, next-generation artificial intelligence (AI) data centers. The partnership aims to revolutionize power infrastructure for AI by introducing advanced, high-efficiency 800 VDC power systems designed to meet the surging global demand for AI computing.
AI power solutions
Global data center power demand is projected to soar from 80 GW in 2024 to 220 GW by 2030, driven primarily by the rapid growth of AI workloads. These workloads are expected to account for nearly 70% of the total increase in data center capacity over the next five years.
Following the announcement, ABB shares jumped as much as 1.9% in Zurich trading before settling with gains of 0.9%, reflecting strong investor confidence in ABB’s leadership role in next-generation data infrastructure.
Revolutionary Power Architecture for AI Factories
At the heart of the collaboration is the development of NVIDIA’s 800 VDC power architecture for its upcoming Kyber rack system, which will support 1-megawatt (MW) server racks capable of housing up to 576 NVIDIA Rubin Ultra GPUs.
This marks a fundamental shift from traditional 415 or 480 VAC three-phase systems to direct current (DC) architectures, enabling higher efficiency, lower losses, and simplified power design.
“ABB is leading the development of the key new power distribution technologies that will create the next generation of data centers,”
said Giampiero Frisio, President of ABB Electrification.
“We have been an early investor in the cutting-edge UPS, DC, and solid-state electronics that will enable data centers to stay ahead of AI’s growing power demands.”
800V DC power architecture
The 800 VDC system provides several breakthrough advantages over traditional AC infrastructure:
- 157% more power capacity through the same wire gauge compared to 415 VAC systems
- Reduced copper usage thanks to fewer conductors and smaller connectors
- Elimination of multiple AC-to-DC conversions, improving overall efficiency
- End-to-end efficiency above 90%, cutting operational costs and carbon footprint
By adopting this architecture, ABB and NVIDIA aim to redefine how large-scale AI factories are powered and cooled — critical for maintaining performance at megawatt power densities.
Meeting Unprecedented AI Power Demands
Dion Harris, Senior Director of HPC, Cloud, and AI Infrastructure at NVIDIA, emphasized the urgency of the collaboration:
“As AI demands continue to grow around the world, data centers require new approaches to power distribution that improve efficiency and simplify designs. Through our collaboration, NVIDIA and ABB are supporting the industry in advancing toward 800-volt architectures that will enable the high-density AI infrastructure needed to fuel the next generation of AI.”
ABB’s latest innovations — such as the HiPerGuard solid-state medium-voltage UPS and the SACE Infinitus solid-state circuit breaker — are engineered specifically to deliver unmatched reliability, safety, and scalability for these AI-powered data centers.
Strategic Vision and Global Impact
With global data center capital expenditure projected to exceed $1 trillion by 2030, ABB is positioning itself at the core of the world’s digital transformation. Notably, 40% of ABB’s electrification R&D is now dedicated to next-generation data center technologies, including DC power systems, solid-state protection, and smart grid integration.
This partnership with NVIDIA represents a critical milestone in building the gigawatt-scale AI factories that will define the infrastructure backbone of tomorrow’s artificial intelligence economy.
Frequently Asked Questions (FAQ)
1. What is the ABB and NVIDIA partnership about?
ABB and NVIDIA have partnered to develop gigawatt-scale, next-generation AI data centers. The collaboration focuses on introducing 800VDC power architecture to support 1-megawatt server racks, enabling higher efficiency and scalability for AI workloads worldwide.
2. Why is the 800VDC architecture important for AI data centers?
The 800VDC system marks a major shift from traditional 415 or 480 VAC systems. It can carry 157% more power through the same wire gauge, use fewer conductors, reduce energy loss, and increase overall efficiency above 90%, making it ideal for high-density AI applications.
3. How will this partnership impact global data center growth?
With global data center power demand expected to rise from 80 GW in 2024 to 220 GW by 2030, the ABB–NVIDIA collaboration will help meet these massive energy requirements more efficiently, while reducing environmental and operational costs.
4. What technologies will ABB contribute to this partnership?
ABB brings its expertise in direct current (DC) systems, solid-state electronics, UPS solutions, and advanced circuit breakers — such as the HiPerGuard medium-voltage UPS and SACE Infinitus solid-state breaker — to ensure reliability and high efficiency in next-generation AI power systems.
5. What is NVIDIA’s role in this collaboration?
NVIDIA is integrating ABB’s power solutions into its Kyber rack architecture, which supports 576 Rubin Ultra GPUs per rack, consuming over 1 megawatt of power. NVIDIA’s goal is to create AI factories capable of handling the massive computational demands of future AI models.
6. How does this partnership support sustainable energy goals?
By moving to 800VDC power and reducing multiple AC/DC conversion stages, the new design minimizes energy waste, heat generation, and carbon emissions — contributing to more sustainable and eco-efficient AI data centers.
7. When will the ABB–NVIDIA powered data centers be available?
The collaboration is currently in the development phase (2025), with pilot projects expected to begin soon. Full-scale gigawatt-level AI data centers are anticipated toward the late 2020s, as part of NVIDIA’s global AI infrastructure expansion.
8. How does this partnership benefit the AI industry?
The ABB–NVIDIA partnership provides the foundation for high-density, energy-efficient AI infrastructure, enabling faster model training, reduced downtime, and scalable design — essential for powering future AI innovations, from large language models to autonomous systems.
[…] AI Infrastructure Takes Center Stage […]