TL;DR
- Surging AI Energy Demand: AI expansion is driving a massive surge in data center electricity consumption, with U.S. data center usage projected to reach between 9.5% and 15.3% of total domestic power by 2030 and global consumption expected to roughly double to 945 TWh
- Infrastructure Timing Mismatch: Traditional grid interconnection approaches are encountering severe bottlenecks because data centers can be developed in 2 to 3 years, whereas major utility power infrastructure takes considerably longer to plan, permit, and build
- Active Role for Battery Storage: Battery Energy Storage Systems (BESS) are evolving from passive emergency backups into core drivers of facility power architecture, working alongside microgrids and on-site generation (solar, fuel cells, natural gas) to manage peak demand and ensure mission-critical reliability
- Early Civil and Regulatory Integration: Deploying utility-scale storage requires early cross-functional alignment, as project success is frequently dictated by physical site conditions, civil balance-of-plant work (such as drainage and stormwater management), fire safety, and local authority reviews
# # #
By William J. Pirrone, P.E., Founding Principal, Rubicon Professional Services
Most people are aware that AI is changing and expanding the data center industry. What gets less public attention is what this growth means for the infrastructure needed to power these data centers.
I have spent more than two decades working on critical facilities and energy infrastructure, and one of the biggest changes I have seen is the amount of power developers now need and how quickly they need it. The traditional approach of designing a facility and securing a utility interconnection is becoming increasingly difficult, putting sustainable infrastructure planning at the center of nearly every new project.
According to the U.S. Department of Energy and Lawrence Berkeley National Laboratory, data centers accounted for about 4.4% of U.S. electricity consumption in 2023. Their 2024 study projected that share could reach between 6.7% and 12% by 2028. A recent Berkeley Lab update estimates that data centers could account for 11.8% of U.S. electricity use by 2030, with scenarios ranging from 9.5% to 15.3%. The two studies indicate that these issues are also playing out globally. Case in point: The International Energy Agency projects that electricity consumption by data centers will roughly double by 2030, to about 945 terawatt-hours. AI is the biggest driver of that increase.
This research doesn’t represent an abstract energy-policy discussion; it dictates where and how data centers can be built, how quickly capacity can come online and, ultimately, whether a project is economically viable. Increasingly, developers are turning to energy storage technology deployed alongside the grid rather than as a backup to address this new industry dynamic.
Battery Storage Is Moving To The Forefront
For years, battery energy storage in critical facilities was primarily associated with backup power and UPS systems, but that’s no longer the case. It is becoming a main driver of power decisions.
Today’s battery energy storage systems (BESS) play a much more active role in the facility’s power architecture. Properly designed, BESS help manage peak demand, provide rapid response when grid conditions change, and work alongside other sources of generation.
To be properly optimized, a data center’s power needs to run through integrated energy management systems that combine grid power, battery storage, and on-site generation, including solar, fuel cells, and natural gas. Today’s objective is to create a power system that is reliable enough for mission-critical computing but flexible so that the facility isn’t entirely dependent on a single power source.
A study commissioned by California’s three major investor-owned utilities found that a modeled solar facility paired with four hours of battery storage could achieve a 99.8% effective load-carrying capability. This study demonstrated how dramatically storage can improve the reliability value of intermittent generation.
Recent improvements in battery technology advancements have made systems like this possible, but they have also made them more complex. I saw that firsthand recently on a 200-megawatt, multi-hour storage project at an existing power generation site. The system wasn’t a simple containerized battery; it was closer to a distributed industrial process plant, with its own water treatment, electrolyte handling, and air management systems layered on top of the electrical framework. Evaluating it meant pricing out civil work, mechanical systems, and electrical balance-of-plant as separate, substantial line items, rather than add-ons to a battery purchase.
However, every modern data center should not simply install solar panels and batteries as this one did. Site conditions, load profiles, utility requirements, and economics vary considerably. But this 200-megawatt, multi-hour storage demonstrates that storage has moved well beyond its traditional role as an emergency backup resource.
Leveraging Batteries: Experience Installation Matters
A utility-scale battery installation is not another piece of IT equipment that can simply be added to a data center. Large battery systems call for careful attention to fire protection, electrical safety, commissioning procedures, and emergency response. They also require close coordination with utilities, fire officials, inspectors, and other authorities, and some communities may be dealing with the review of a large-scale battery installation for the first time.
On one recent project, the single biggest cost and schedule driver wasn’t the battery technology at all; it was a pair of stormwater retention ponds already on the site. Adjusting to the new installation required partial infill, regrading, and a redesigned drainage system, all before a single battery enclosure could be installed. It’s a good reminder that the constraint that decides whether a site works often isn’t the equipment—it’s what’s already in the ground.
AI Is Forcing A Power Strategy Change
A few years ago, power infrastructure was often treated as something that followed the data center design. Today, designers need to consider power much earlier. Power availability, interconnection timing, on-site generation, and energy storage will determine a facility’s success.
The International Energy Agency has pointed out the basic timing problem: a data center can potentially be developed in two or three years, while major energy infrastructure can take considerably longer to plan, permit, and build.
The Future Should Not Contain Mismatched Power
The data centers that support the next generation of AI will require enormous amounts of reliable electricity. The industry cannot assume the grid alone will always provide that power on the timetable developers need. That is why battery storage, microgrids, and behind-the-meter generation are becoming so important. The facilities that succeed will be those where engineering, construction, utilities, local authorities, and the providers of these capabilities are brought together early enough to turn a power strategy into something that thrives in the field.
# # #
About the Author
William Pirrone is founding principal of Rubicon Professional Services and Rubicon Technical Services. He brings more than 35 years of engineering design and construction, operations, and maintenance experience to the data center, telecommunications, and critical power industries, where he established a significant presence. William’s technical expertise and understanding of an owner’s perspective have been instrumental in the rapid success of both RPS and RTS on a national level.
As a founder of the Rubicon companies, William has developed a unique set of service offerings specific to mission critical environments, designed to fill the void left by traditional construction companies and methods. These services include technical program management; construction management; equipment integration; engineering, procurement, and construction (EPC); commissioning; and energy management services.
Prior to establishing the Rubicon companies, William was the director of global engineering at JP Morgan Chase, where he led the design and integration of all data, trading, and critical operating centers through multiple billion-dollar mergers and acquisitions globally. That charge carried across 60 countries and more than 70 million square feet of real estate. While at JP Morgan Chase, William also provided direct support to the real estate division by reviewing and negotiating leases for key operations centers, developing tactical and strategic facility restack/disposition planning, and line of business relocation strategies at the corporate level. He also partnered with the energy procurement team to secure substantial cost savings and pricing stability in the deregulated energy markets across the United States. William departed JP Morgan Chase to launch Rubicon. He holds a B.S. degree in electrical engineering from Pratt Institute, is a licensed Professional Engineer, and holds multiple general contractor licenses.
