[Part 2 of 2]
Summary:
As the demand for AI grows, the power infrastructure for data centers is changing quickly. Utilities, commercial operations, and tech providers are re-evaluating their strategies for planning, storing, and managing energy. The data center industry faces many challenges, including changes in battery technology, the need for extensive field testing, and the increased importance of AI and local production. This requires a new approach to ensure grid reliability.
This article is a recap of the panel discussion hosted by Stryten Energy at CES 2026. Stryten Energy brought together experts from policy, research, utilities, and battery technology to discuss how to support the evolving power needs of data center infrastructures.
Key topics:
- Coordinating grid planning
- Selecting suitable storage for AI workloads
- The coexistence of batteries and gensets
- Using AI and real-world data to enhance battery performance and grid reliability
Meet the panelists:
- Moderator: Scott Childers, Vice President, Essential Power, Stryten Energy
- John A. Howes, Principal, Redland Energy Group
- Frank Sharp, Technical Director, EPRI
- Travis Torrey, Chief Technical Officer, Storion Energy
- Erik Spoerke, Senior Analytics Advisor, Energy Storage Division, DOE
Highlights from the Discussion:
What regulatory changes would help utilities and data centers coordinate better?
Regulatory changes should aim for quicker decision‑making, clearer jurisdiction, and more flexible ownership models for on‑site energy resources. The current approval processes are too slow to keep up with the rapid, high‑magnitude load growth driven by AI and hyperscale data centers.
Streamlined regulatory timelines would allow utilities to plan and build reliably. Regulators also need to resolve the growing conflict between state commissions and FERC over who has authority to site generation and infrastructure for co‑located loads. A consistent national framework would reduce delays and uncertainty.
Updated rules that allow utilities or data centers to own and operate dispatchable assets such as backup generation or energy storage based on what best supports grid reliability would help stabilize variable demand. Together, these changes would create a more predictable environment for both utilities and data centers while improving overall grid resilience.
Should storage for AI data centers be technology‑agnostic?
While AI data centers face rapidly evolving power and reliability demands, storage should not only be technology‑agnostic but also aligned with the battery architecture that best fits the application’s durability needs and performance requirements.
Chemistries will continue to change just as lithium applications evolved from NMC to LFP. What matters most is selecting the right architecture for the specific operational profile of workloads.
Today, flow batteries, particularly vanadium flow systems, offer strong durability and high‑cycle performance that match the continual cycling and long-duration support that AI facilities require.
Over the long term, additional flow technologies and advanced chemistries will emerge, but the key is choosing purpose-built solutions for data centers rather than repurposing technologies designed for other applications.
What’s the leading storage technology today, and will that change?
Today, lithium-based technologies dominate the energy storage landscape mostly because of their availability at scale. Lead batteries, however, remain a viable option in suitable applications. But there is no single battery chemistry “champion” for an energy storage solution.
Taking a systems-based approach to address the needs of a project based on location, performance, requirements, and budget will help you find the solution. As more technologies emerge, matching capabilities to real‑world requirements will matter far more than championing any single chemistry.
Why are more demos and tests so important right now?
Right now, many new energy storage systems are being tested in real-world settings. Much of this work is happening at data centers, where teams evaluate performance, safety, lifespan and cost under actual operating conditions—not just in controlled lab environments. Because batteries can outlast computer hardware, these tests help organizations think long‑term about which chemistries make sense for future‑proofed designs.
Field testing is important to learn what different batteries can actually do, not by what the spec sheet says, but by running them through real conditions. Efforts from groups like EPRI and DOE are helping reveal where each technology truly fits.
At the same time, there’s growing attention on testing at scale through a domestic supply‑chain lens. Most large U.S. storage systems still rely on non‑U.S. lithium cells, underscoring how far the country is from having a secure, reliable, homegrown supply chain for mission‑critical assets. That gap is both a major challenge and a strategic opportunity.
This is where trusted partners like the Department of Energy come in. DOE‑supported programs and national‑lab collaborations provide the large‑scale testing, validation and R&D needed to understand what storage technologies can truly deliver in the field and to help expand domestic manufacturing capacity. These programs are valuable today, but they need to grow to match the scale and urgency of the industry’s needs.
Will batteries replace diesel generator sets as reliable backup power for data centers?
Batteries and diesel generator sets are going to coexist, as each serves a different purpose. Diesel and other spinning assets can run as long as they’re refueled, while batteries are valuable for multi‑hour events, load‑shaving, and bridging power transitions.
In practice, most data centers will rely on a mix of both, using batteries to handle shorter disruptions or to extend runtime when paired with intermittent energy sources, and keeping generators for long‑duration reliability. The key is integration and operating each asset in a way that maintains power quality and doesn’t compromise the data center’s performance.
Ultimately, these technologies are complementary, not interchangeable, and the future is about combining them intelligently rather than expecting one to replace the other.
Can AI improve battery performance and the grid that supports it?
Yes, AI can improve both battery performance and the grid, but its impact depends on access to large, high‑quality datasets. The more assets deployed in the field, the more real‑world data AI can learn from, creating an iterative cycle where better data leads to better models and, ultimately, better storage and grid performance. For AI to be effective, the industry needs more open, shared data rather than proprietary silos.
AI is already being used in manufacturing to reduce cost and improve processes, and it can also accelerate R&D and system‑level optimization across the grid. As data sets expand, AI can help refine product design, operational strategies, and overall grid architecture.
The U.S. is well positioned to lead this space, both because of its significant investment in data centers, which supports AI development, and its long‑established domestic battery industry, particularly in lead‑acid. Together, these give the country a strong foundation for advancing AI‑driven improvements in energy storage.
Key Takeaway/ Conclusion
The future of powering AI and data centers will depend on flexible, application‑specific energy systems, not any single technology, with real‑world data, intelligent regulation, and AI‑driven optimization guiding the way.
Across storage technologies, grid planning, backup strategies, and manufacturing, one thing is evident: progress comes from integrating the right mix of solutions, validating them through field data, and enabling AI to continually improve performance across the entire energy ecosystem.
