The rapid expansion of AI data centers is creating a new power-infrastructure challenge: workloads are becoming more dynamic while facilities need reliable backup capacity and increasingly efficient power management. Ampace is positioning advanced lithium-ion battery systems as part of that infrastructure, focusing on modular power architectures and battery-based load smoothing for AI workloads.
Ampace will showcase its battery-storage technology at Data Center World Asia in Singapore, highlighting how energy storage is evolving from a conventional backup component into a more active part of AI data-center power infrastructure.
The company will focus on two areas: integrating batteries into prefabricated Power Train Unit (PTU) architectures and coordinating battery systems with uninterruptible power supplies (UPS) to manage fluctuations associated with AI workloads.
AI data centers are placing different demands on electrical infrastructure than traditional computing environments. High-density accelerators can produce rapidly changing power requirements, creating challenges for facilities and the grids that supply them.
That is pushing data-center operators and infrastructure providers to consider batteries not only as emergency backup systems but also as assets that can participate in short-duration power management.
Batteries Move Beyond Backup
Ampace will discuss the integration of battery systems into prefabricated PTU architectures during a joint session with Eaton.
Prefabricated power systems are increasingly relevant to data centers because they allow electrical infrastructure to be assembled in modular units and expanded as computing capacity grows.
Within that architecture, battery performance becomes connected to broader requirements involving physical footprint, electrical integration, monitoring and scalability.
Rather than treating the battery as an isolated component, the architecture-level approach considers how energy storage interacts with UPS equipment, power conversion systems and monitoring infrastructure.
This becomes particularly important for AI facilities, where additional computing capacity can be deployed in stages and power requirements can change as workloads evolve.
AI Workloads Create New Battery Requirements
Ampace will also demonstrate how batteries can support AI load smoothing.
Traditional data-center batteries are primarily associated with backup power. AI workloads introduce another potential application: frequent charge-discharge cycles designed to reduce short-term fluctuations in power demand.
Ampace says its approach coordinates UPS and battery operation to smooth grid-side demand while retaining backup functionality. The company has evaluated battery performance using electronic-load testing and simulated AI duty cycles.
Its PU100 is designed to support both UPS backup and AI load-smoothing applications, while Ampace will also showcase the PU200 at the event.
The technical challenge is balancing these two functions. Batteries used for frequent cycling face different operating requirements from systems primarily maintained for emergency backup. Thermal management, cycle life, charging behavior and control software consequently become important considerations.
AI Infrastructure Is Becoming an Energy Problem
The development highlights a broader reality of AI infrastructure: scaling compute requires scaling electricity infrastructure alongside it.
Data-center operators are increasingly evaluating power availability, grid constraints, energy efficiency and flexible storage as part of AI deployment planning. Modular power systems can help address deployment speed, while battery storage can provide additional flexibility between the data center and the electrical grid.
For technology companies building AI infrastructure, this makes energy storage increasingly connected to the performance and scalability of the computing environment itself.
Market Landscape
AI infrastructure is expanding beyond GPUs and servers into the electrical systems required to operate them.
Data-center providers are exploring high-density power distribution, liquid cooling, modular electrical infrastructure, UPS modernization, battery storage and grid-interactive technologies as AI workloads increase power demand.
Companies such as Eaton, Schneider Electric, Vertiv and ABB operate across different parts of this infrastructure ecosystem, while battery manufacturers are developing storage systems designed for data-center applications.
The market is also moving toward tighter coordination between computing workloads and energy systems. The longer-term opportunity is not simply providing more backup power, but creating infrastructure capable of responding dynamically to changing AI demand while maintaining resilience.
Top Insights
- Ampace is positioning lithium-ion batteries as an active component of AI data-center infrastructure rather than solely as emergency backup equipment.
- The company is exploring battery integration within prefabricated Power Train Unit architectures for modular data-center deployment.
- AI load smoothing introduces more frequent battery cycling, creating requirements beyond conventional UPS backup applications.
- Coordinating UPS and battery systems could help smooth short-term grid demand while preserving backup capability.
- AI infrastructure expansion is increasingly linking compute scalability with power availability, energy storage and electrical-system design.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
