How Solid-State Protection Is Unlocking DC at Scale
Brian J. Nelson
Global data center power demand is projected to reach 220 GW by 2030. In the U.S. alone, demand is expected to more than triple, from 25 GW in 2024 to more than 80 GW by 2030.
Data center power demand growth is being driven not only by the development of new facilities, but also by rising rack densities that increase power requirements within each facility. Each new generation of AI chips draws significantly more power than the last. Data centers are also packing more GPUs into each rack to enable faster communication during training and inference, compounding the increase in rack-level power demand.
The conventional path power takes to reach a server requires several conversions between alternating current (AC) and direct current (DC), an approach developed around rack densities below 100 kilowatts (kW). The industry is now experimenting with 400- to 800-kW racks, and many are even designing around 1-megawatt (MW) racks for deployment in 2027. At this scale, efficiency in power delivery becomes increasingly important. Conversion losses that are manageable at legacy rack densities become more consequential when multiplied across large numbers of higher-density racks. Delivering substantially more power to each rack also places greater demands on copper and other electrical infrastructure.
Scaling DC for increasingly power-dense applications will require changes across the power architecture, from how power is converted and distributed to how systems are protected and equipment works together. Solid-state protection addresses one of the fundamental challenges: providing the speed and bidirectional capabilities needed to maintain safety and reliability as DC networks grow in scale and complexity.
The Case for DC Distribution
A more direct DC power path reduces the cumulative losses introduced as power moves through multiple conversion stages. Fewer conversion stages can also mean less conversion equipment and associated copper infrastructure, reducing the floor space required for power delivery.
DC can also simplify the integration of battery storage and renewable generation into the power architecture. Because solar generation and battery storage are natively DC, they can connect to DC distribution systems without an intermediate AC stage. For data centers, integrated battery storage can respond quickly to changing loads, helping buffer rapid fluctuations in demand.
Protecting an Active DC Grid
Scaling DC distribution requires advanced protection capable of detecting and isolating faults, interrupting rapidly rising current, and operating bidirectionally. Alternating current naturally crosses zero each cycle, providing a point at which conventional breakers can interrupt the current. DC has no natural zero crossing, making fault interruption more challenging. DC short circuits can also be dominated by rapid capacitor discharge, causing fault current to rise extremely quickly. Without rapid isolation, a fault can collapse voltage below the system's operating threshold, disconnecting sources and destabilizing the broader network.
The increasingly distributed nature of DC systems adds another challenge. With sources and storage connected throughout the network, current and fault current can flow in either direction, making fault location and isolation more complex.
Engineered for DC: SACE Infinitus
Solid-state circuit breakers address these protection requirements by replacing mechanical switching with semiconductor-based technology capable of detecting and interrupting faults on much shorter timescales.
ABB's SACE Infinitus, the first 1000 V solid-state circuit breaker certified to IEC 60947-2, has a breaking time under 25 microseconds, about 100 times faster than a mechanical circuit breaker. Combined with appropriate system design, this speed can help preserve service continuity by isolating faults within individual protected zones rather than allowing a localized fault to escalate. The SACE Infinitus also provides bidirectional current control.
Building the DC Ecosystem
ABB is helping build an interoperable DC ecosystem through engineering partnerships, industry standards, and investments in technologies that support the transition to DC.
In October 2025, ABB began working with NVIDIA to support its 800 VDC architecture for 1 MW server racks, bringing together medium-voltage UPS, DC distribution, and solid-state electronics. The collaboration has since expanded to bring ABB's power equipment into NVIDIA's Omniverse DSX Blueprint, which enables electrical, thermal, and compute infrastructure to be modeled and validated before construction.
Industry standards are another critical piece. ABB holds board seats in the Open DC Alliance and Current/OS, leading organizations working to align terminology, architectures, and voltage levels so equipment and systems can interoperate across DC environments. Many manufacturers are waiting on clearer standards before committing to DC-specific equipment. Early alignment can accelerate design, certification, and production at scale.
In July 2026, ABB announced plans to acquire Advantics, a France-based specialist in silicon carbide power conversion. The acquisition will expand ABB's DC portfolio with converter modules that deliver up to 99% efficiency, strengthening its capabilities in DC power conversion.
A Shift Across Electrification
Data centers are accelerating development of high-power DC architectures, but the implications extend across electrification. Utility-scale battery storage, renewable microgrids, and EV charging are also bringing together DC-native resources, bidirectional power flows, and increasingly sophisticated power electronics. DC and AC will continue to coexist, with each suited to different applications, but the boundary between them is shifting as more generation, storage, and loads become DC-native.
Making DC work at scale will depend on how effectively conversion, distribution, and protection can work together within an interoperable system.
