Energy Storage for Data Centers

As AI drives unprecedented demand for data center power, energy storage is evolving from a backup technology into a core component of project development and investment strategy. This session will examine how batteries, microgrids and behind-the-meter storage are reshaping data center site selection, power procurement, project finance, contracting, and M&A. Attendees will gain a framework for evaluating storage as both critical infrastructure and a strategic asset.

This presentation peels away the layers of batteries for data centers, and identifies which segments of the data center market, which battery types and which suppliers will lead and which will lag. Based on the likely trajectory of renewable power, data center construction, and EV production, and considering politics and productivity, it offers viable strategies for battery OEMs and service providers to data centers.

The massive power consumption of AI data centers demands rigorous efficiency tracking for sustainability and ROI. To optimize energy use, we must understand efficiency of AC/DC conversion to point-of-load regulation. This session examines power architectures—including 800 Vdc systems that reduce losses—and how to measure them. Static efficiency alone is insufficient; accurate characterization requires evaluating performance under steady state conditions, during dynamic events dynamic events such as load transients and startup sequences, and across varying workloads and temperatures.

AI data centers require energy storage to do more than provide backup power. This talk explores how choices at the cell level shape system performance, thermal management, safety, reliability, and use of limited grid capacity. Connecting battery chemistry to system architecture, it examines storage across dynamic AI workloads and facility requirements, showing how simplified systems can reduce operating complexity while enabling resilient, efficient power infrastructure.

Engineering interfaces remain one of the central challenges in realizing the full potential of all-solid-state batteries. This presentation describes a collaborative approach between SandboxAQ and Solid Power, combining advanced atomistic modeling, machine-learned interatomic potentials, and materials design to engineer chemically stable, ionically-conductive interfaces/interphases. We present AQVolt26 as a foundation for accelerating interface optimization to support commercially ready solid-state electrolyte architectures.

IDTechEx forecasts that data centers will represent a market of US$950 million for supercapacitors by 2037, with opportunities for both EDLCs and hybrid capacitors. Growth is tied to AI data centers, which have characteristic power demand profiles, featuring rapid, large-scale power fluctuations that can lead to grid strain and blackouts. This presentation explores the potential for supercapacitors in data centers, covering supercapacitor technologies, applications, players, and supply relationships.

Data center backup power remains dominated by lead-acid, largely because lithium's thermal runaway risk imposes siting, spacing, and fire-protection costs inside the building. Urban Electric Power's rechargeable aqueous zinc chemistry is UL 1973 listed, and UL 9540A testing found no thermal runaway initiated by any method. This session presents the resulting safety and ventilation design basis, and integration experience from multiple MWh systems built using commercially available power conversion equipment.

AI training & inference clusters impose a power profile unlike any prior data-center load: synchronized GPUs drive megawatt-scale swings within milliseconds, stressing both upstream grid infrastructure and the ESS systems meant to protect the IT load. Conventional energy storage, optimized for rare, one-directional backup events, is poorly matched to the continuous, symmetric, high-rate charge and discharge these loads demand. Echion’s XNO niobium-anode cells are being deployed into UPS and rack-level (BBU/sidecar) energy-storage systems to address this. This XNO battery technology delivers symmetric high-C-rate operation, ultra-long cycle life, and high safety, enabling a solution that not only rides through outages but actively buffers AI load volatility and supports grid-facing services. Because Echion offers multiple high-power cell formats that can be easily integrated into existing OEM designs, it offers a de-risked, near-term path to deployment for UPS OEMs and hyperscale operators.

AI data centers are transforming power architecture, driving demand for higher-voltage DC systems, faster transient response, and greater energy resilience. This presentation examines how Asia’s battery industry is preparing for the emerging AIDC and HVDC landscape, covering high-power lithium-ion, sodium-ion, battery backup units, UPS, and grid-scale energy storage. It highlights technology trends, system requirements, and strategic implications for next-generation data center power infrastructure.

The buildout of data centers to support AI growth is highly controversial. They provide limited job growth; require massive amounts of energy and water, taxing regional supplies; and generate noise pollution. In the U.S., they take over prime land to avoid complicated environmental assessments and permitting. Like with lithium-ion battery plants, the community backlash has been intense. This talk will provide a blueprint for companies to successfully navigate public consultation through successful global case studies that can be adapted for many high tech and renewable energy infrastructure projects.