In a strategic move to address the escalating power demands of artificial intelligence, industrial technology leader ABB has entered a formal partnership with chipmaker Nvidia. The collaboration will focus on developing and deploying next-generation power infrastructure specifically designed for the immense energy densities required by modern AI data centers. By combining ABB’s deep expertise in electrification and power management with Nvidia’s leadership in high-performance computing, the two companies aim to create a new blueprint for data center power distribution capable of supporting compute-intensive workloads at an unprecedented gigawatt scale.
The core of the initiative is a fundamental rethinking of how electricity is delivered from the grid to the server rack. As AI models and high-performance computing tasks become more complex, they require server racks that consume dramatically more power than traditional hardware, pushing legacy alternating current (AC) power systems to their absolute limits. This partnership seeks to accelerate the industry’s transition to a more efficient and robust direct current (DC) power architecture. The collaboration will directly support Nvidia’s planned rollout of an 800 VDC power architecture for 1-megawatt server racks, a design that promises to significantly reduce energy conversion losses, simplify cooling requirements, and improve the overall reliability of the power chain for mission-critical AI applications.
Surging Energy Demands of AI
The rapid integration of artificial intelligence across all sectors of the global economy has triggered an exponential increase in the demand for data processing, which in turn requires a massive expansion of data center capacity. This growth is creating an energy challenge of historic proportions. According to market analysis from Dell’Oro Group, global data center power demand is projected to surge from 80 gigawatts in 2024 to approximately 220 gigawatts by 2030. The research highlights that AI workloads are the primary driver of this expansion, expected to account for nearly 70% of this growth. This unprecedented demand places immense strain on both public power grids and the internal electrical infrastructure of the data centers themselves.
The infrastructure platforms developed by Nvidia, which power the most advanced AI and high-performance computing systems, are at the forefront of this trend, constantly testing the limits of energy density. A single server rack equipped with the latest GPUs for AI training can consume as much power as dozens of traditional server racks from just a decade ago. This concentration of power in a small physical footprint generates enormous heat and creates complex power delivery challenges. Simply scaling up existing AC-based power systems is becoming increasingly inefficient and unsustainable. The multiple conversion steps required to change high-voltage AC from the grid into low-voltage DC for the server components lead to significant energy losses, primarily as wasted heat. This inefficiency not only drives up operational costs but also increases the carbon footprint of the facility, running counter to the sustainability goals of many operators.
A Fundamental Shift to DC Power
To overcome the limitations of legacy systems, the ABB and Nvidia partnership champions a decisive shift toward high-voltage DC power distribution within the data center. The collaboration is centered on developing and standardizing an 800 VDC architecture that delivers power more directly and efficiently from the source to the processors. By minimizing the number of AC-to-DC and DC-to-DC conversion stages, this approach can substantially reduce energy losses, improve thermal performance, and increase the overall power-per-square-foot capacity of a facility. This transition represents one of the most significant evolutions in data center power design in decades.
In this new architecture, power would enter the data center at a medium voltage, be converted to 800 VDC via a highly efficient uninterruptible power supply (UPS), and then be distributed directly to server racks. This eliminates multiple intermediate conversion steps common in traditional AC “dual-conversion” UPS systems. Dion Harris, Senior Director of HPC, Cloud, and AI Infrastructure at Nvidia, emphasized the need for this evolution. “As AI demands continue to grow around the world, data centres require new approaches to power distribution that improve efficiency and simplify designs,” he stated. The simplified design not only saves energy but can also reduce the amount of copper wiring and electrical switchgear required, freeing up valuable space for more computing hardware and potentially lowering capital expenditure on new builds.
ABB’s Technological Contributions
ABB is bringing a portfolio of advanced technologies and deep domain expertise in power electronics to the partnership, providing the foundational components needed to make large-scale DC power distribution a practical reality. Approximately 40% of ABB’s research and development in its Electrification division is already focused on solutions for next-generation data centers, including electrical protection for DC systems and energy-efficient cooling. The company’s contributions are centered on making the high-voltage DC architecture not only efficient but also safe, reliable, and scalable for mission-critical environments.
Solid-State Power Electronics
A cornerstone of ABB’s offering is its development of solid-state electronics, which are far faster and more reliable than traditional mechanical systems. One key innovation is the SACE Infinitus, the world’s first IEC-certified solid-state circuit breaker. Unlike mechanical breakers that involve physical moving parts, a solid-state breaker can interrupt a fault current in microseconds. This incredible speed and precise control are essential for protecting sensitive server electronics in a high-voltage DC environment, making the architecture viable and safe for large-scale deployment. Another critical technology is ABB’s HiPerGuard, a solid-state medium-voltage UPS. This system is specifically designed for AI and hyperscale facilities, offering superior power density and energy efficiency while occupying a minimal physical footprint, which is a critical consideration in modern data center design.
Integrated Power Systems
The partnership’s vision involves integrating these advanced components into a seamless power chain. The future power architecture anticipates combining the HiPerGuard MV UPS with direct DC distribution that runs straight into the server rooms. This entire system will be managed and protected by advanced solid-state power electronics like the SACE Infinitus. This tightly integrated approach allows data center operators to achieve higher overall efficiency, reduce the potential points of failure associated with complex conversion losses, and ultimately improve the resilience of their most critical infrastructure. ABB complements this hardware with intelligent power distribution systems and digital monitoring tools that provide operators with real-time visibility and control over their electrical environment.
Building Sustainable Digital Infrastructure
The collaboration between ABB and Nvidia extends beyond pure performance, aiming to establish a new standard for sustainable digital infrastructure. By creating more energy-efficient data centers, the partnership directly addresses the growing environmental concerns associated with the AI industry’s massive power consumption. Reducing energy waste at the source lowers the operational carbon footprint and decreases the strain on public utilities, enabling a more responsible scaling of AI capacity. This forward-looking approach is critical for ensuring that the continued growth of AI is compatible with global sustainability objectives.
Giampiero Frisio, President of ABB Electrification, underscored this long-term vision and commitment. “ABB is leading the development of the key new power distribution technologies that will create the next generation of data centres,” he said. “We have been an early investor in the cutting-edge UPS, DC and solid-state electronics that will enable data centres to stay ahead of AI’s growing power demands.” By combining ABB’s leadership in high-efficiency power systems with Nvidia’s pioneering work in accelerated computing, the two companies are working to build a foundation for future-ready, gigawatt-scale data centers. The ultimate goal is an infrastructure that not only meets the relentless performance targets of the AI era but does so in a manner that is both economically viable and environmentally responsible for generations to come.
