US energy sector faces unprecedented growth in demand for power

January 2025  |  SPECIAL REPORT: ENERGY & UTILITIES

Financier Worldwide Magazine

January 2025 Issue

The US energy sector is bracing for an increase in demand by orders of magnitude. This is great news for investors in energy infrastructure. For nearly half a century, energy production in the US has generally outpaced consumption. Less than a decade ago, the US electric power sector was grappling with flat, even declining, demand.

The advent of artificial intelligence (AI), cryptocurrency mining, reshoring of US manufacturing and the electrification of industrial processes, buildings and transportation are all contributing to an exponential growth in electricity demand that is rapidly outpacing the ability of new generation and related infrastructure to keep up.

Just how much electricity will be needed? Predictions vary widely, but one of the most cited estimates is an Electric Power Research Institute white paper that reports that data centres running the processing operations for AI, ChatGPT and bitcoin mining represent approximately 4 percent of energy use in the US. By 2030, this figure is expected to grow between 4.6 percent to 9 percent.

To the observer, this may not seem a huge jump. It is, in fact, a staggering figure. By way of illustration, ChatGPT’s energy use each day is about the same as that used in 17,000 households, while the average energy needed for a single bitcoin transaction is approximately equal to the monthly electricity consumption for an average US household.

Multiply this by the numerous, massive data centres, existing and planned, powering AI in operations in thousands of applications, as well as other anticipated demand growth, and the US soon may need up to three times the electric power infrastructure that it currently has. It is the classic ‘be careful what you wish for’ scenario, placing electric utilities and other power suppliers in a position akin to the little dog that caught the bus.

There are, of course, important caveats to the predictions. First and foremost, the energy sector changes notoriously quickly, and forecasts are wrong as often as they are right. Changes in global and domestic policy, natural events, resource availability and rapid technological advancements constantly alter the assumptions underlying forecasts. Recall that in the early 2000s, the US believed that it had less than a 10-year supply of natural gas remaining. Flash forward a few years and a change in technology and this figure was a 100-plus years’ supply.

There is no question that electricity demand in the US is growing, but the rate and duration of growth is dependent on a host of dynamic factors and challenges. Ease of siting and permitting for industrial and commercial facilities, regulatory structures and federal, state and local policy initiatives add layers of opportunity and risk to business decisions on when and where to locate data centres and manufacturing plants. However, availability of power and water are key requirements. A data centre can require up to a gigawatt of power, similar to the aggregate load of a large city, or the amount produced by an entire nuclear unit.

Not every region in the US is experiencing the same level of growth. Northern Virginia, currently the titular ‘data centre capital’ of the world, is a hub of tremendous demand growth, with neighbouring Maryland, Pennsylvania and other mid-Atlantic states vying to attract investment by hyperscalers. Texas, California, Oregon, Arizona and North Dakota are also witnessing growth in energy demand from data centres and re-shored manufacturing. However, as generation and transmission becomes constrained, developers may turn to other regions where those facilities are more available and cheaper, assuming that they can support a massive, inflexible, always-on load.

Data centres are identified as critical infrastructure in the US, increasingly essential to business, finance, defence, and many other sectors and activities. Because of this, policymakers are highly motivated to encourage them to keep their operations in the US. As a result, the growth of data centres, along with the renaissance of other energy-intensive businesses, represent significant opportunities for investors, assuming threshold challenges can be overcome and risks managed.

How to accommodate these new large loads is a primary subject of discussion throughout the US energy sector – regulators, utilities and producers, grid operators, and a broad range of energy market participants and other stakeholders are grappling with potential approaches to meet critical challenges. Key among these are electrical system reliability, environmental priorities, inadequate baseload generation and transmission, cost allocation, and the risk that existing customers may be saddled with costs from upgrades for which they derive little benefit, or stranded costs if estimates of future demand prove to be overstated.

This last risk is especially vexing given the planning and construction horizon for generation and transmission. AI is evolving at an accelerating pace, and may well resolve its own energy demand problem, rendering some new construction unnecessary. It is a chicken and egg problem, however, because the data centres need to be built in order for the algorithms and models to run to find any such solution.

Lengthy generation interconnection queues of up to seven years have caused hyperscalers to look at a range of alternative options, from co-location arrangements to restarting mothballed units to building their own generating facilities, each of which comes with its own challenges. Microsoft recently announced a deal with merchant generator Constellation Energy to restart a nuclear unit at Three Mile Island that was shuttered only five years ago for economic reasons. Meta has also announced plans to develop a 1.5GW natural gas plant with a Louisiana utility to power a data campus in that state.

Meanwhile, the Federal Energy Regulatory Commission (FERC) disrupted an existing co-location arrangement it had previously approved at another Pennsylvania nuclear plant when it rejected a request to increase the power draw from the plant and left the parties uncertain about the future both of the project in question and FERC’s policy on co-location generally.

Meta’s announcement of plans to construct gas-fired generation came as somewhat of a surprise given its sustainability commitment of reaching net-zero emissions by 2030. It may reflect a recognition that natural gas is an essential interim fuel in the energy transition until clean baseload resources are brought to scale.

It also may be indicative of optimism regarding availability of sufficient offsets from clean resources. A FERC order just last month took significant steps to reduce siting and permitting obstacles, with the aim of reducing the current 10 to 17 year horizon for construction of the transmission additions needed to bring remote renewables to markets.

Despite the acknowledged urgency to expand and modernise the grid, transmission has become so difficult to erect that the annual number of miles of high voltage transmission constructed has steadily declined over the past decade, with a mere 55 miles built last year. Much more cooperation and coordination is needed among the states and regional grid operators to bring this to pass, but FERC’s order is an important step in the right direction.

New large loads also raise reliability concerns. A single 1GW load that falls victim to a cyber attack or experiences a sudden outage can have a significant impact on the rest of the grid. Grid operators like the regional transmission organisations, as part of their mandate, do study these impacts prior to bringing a load on line.

To date, they have expressed confidence in the ability of their operations to manage existing loads, including data centre loads. But with the coming proliferation of large loads, along with continued electrification efforts, this is a serious concern that will require substantial investment to manage, including both new builds and deployment of technologies that allow more to be done with existing infrastructure.

In addition to investment in electrical infrastructure, data centres and manufacturing growth will require investment in related infrastructure, including natural gas, nuclear and other energy resources, construction, workforce development and so forth.

The good news is that while there is much work to be done, none of these challenges are unsolvable, and opportunities abound for investors who are patient, creative and willing to try new approaches.

 

Jennifer Morrissey is a counsel at Dentons. She can be contacted on +1 (202) 408 9112 or by email: jennifer.morrissey@dentons.com.

© Financier Worldwide

BY

Jennifer Morrissey

Dentons

©2001-2024 Financier Worldwide Ltd. All rights reserved. Any statements expressed on this website are understood to be general opinions and should not be relied upon as legal, financial or any other form of professional advice. Opinions expressed do not necessarily represent the views of the authors’ current or previous employers, or clients. The publisher, authors and authors' firms are not responsible for any loss third parties may suffer in connection with information or materials presented on this website, or use of any such information or materials by any third parties.