There has been a lot of ink spilled in recent weeks around AI, data centers and the associated growth in electricity demand. This is indeed a significant change, but it is only part of a larger story. Over the coming years a confluence of transportation electrification, building electrification, onshoring of manufacturing, on top of significant acceleration from data centers is reshaping electricity demand in the United States.
Load growth in the US for the next 10 years is now expected to be materially higher than the effectively flat demand of the last few decades and higher than expected even just a few years ago, signaling a pivotal juncture in the nation’s energy landscape. This will be the first of a series, and we will discuss the consequences of this multi-faceted growth in demand in coming posts.
Background
On January 9, 2024, PJM (the Regional Transmission Authority coordinating power throughout the mid-Atlantic) released their annual 10-year load growth forecast tripling the expected annual growth from the 2023 forecast from 0.8%/year over the next 10 years to 2.4%/year. For comparison, load in the US was flat from 2007 to 2018 and PJM load is still below 2007 peaks.
Source: PJM load forecast 2024; https://www.pjm.com/-/media/library/reports-notices/load-forecast/2024-load-report.ashx[i]
Similarly, an analysis of 2023 FERC filings found that nationwide expectations of 5-year electricity demand growth increased from 2.6%/year to 4.7%/year, adding 38 GW of peak demand in 2028.
Source: Grid Strategies National Load Growth Report, December 2023[ii]
The load forecasting agencies have a long history of over-forecasting load growth- and we approach any top down with caution. But there are reasons to believe this time might be different and that this load forecast may even be an underestimate, as it does not yet include additional load from large grant-driven project developments (e.g. green hydrogen or carbon capture, DAC, blue hydrogen) or fully incorporate potential load from AI-fueled data center growth.
There remains a lot of economic and policy uncertainty, but I think it’s worth diving into the individual drivers of load growth below, as any one of which could change the trajectory from the past decades of flat-ish electricity demand. In particular, the breadth of demand drivers gives us more comfort that load growth will indeed materialize this time.
Individual Sources of Load Growth
Onshoring Manufacturing
The move towards onshoring manufacturing in the U.S. is a significant tailwind for electricity demand. Announced facilities appear to be incorporated into some, but not all, load growth forecasts.
Source: Grid Strategies National Load Growth Report, December 2023
Green Hydrogen and other IRA-driven projects
Green Hydrogen: The Energy Liftoff Report[iii] projects electrolyzers will achieve a low of 46 kWh/kg of hydrogen by 2030, excluding the additional 10 kWh/kg required for transportation. Meanwhile, McKinsey forecasts a global demand for hydrogen ranging from 125-180 million metric tons per year by 2050. This projection, at that efficiency, would require ~5.75 million gWh of electricity per year, which is 1.6 times the total U.S. electricity consumption by 2020.
While these projections may seem unrealistic, even the initial deployment of hydrogen production plans can have a significant impact. According to the DOE, there are already approximately 3.6 GW of peak demand additions from hydrogen plants announced in the U.S.
Carbon Capture: The IRA substantially increases the availability of credits, making it easier for CCUS projects to qualify, provides new ways to monetize credits, and extends the deadline to begin construction to 2033.
McKinsey[iv] reports that more than 200 new facilities were added to the project pipeline in 2022, bringing the global pipeline of CCUS projects to 68 in operation, 39 under development, and 533 in planning stages. Over the past 12 months, the capacity of CCUS facilities under development has grown to over 420 million tons/annum, an increase of more than 40% from 2022. The global CCUS landscape is also diversifying, as cement, blue hydrogen, iron and steel, and power are projected to account for 85% total uptake by 2050.
Building and Transport Electrification
Building electrification, including the transition from gas furnaces and water heaters to electric heat pumps, is expected to drive significant load growth. New England ISO, for example, predicts[v] that electrification from heat pumps and EVs could increase annual electricity consumption by 2% to 13%.
According to the Brattle Group[vi], fully electrified heating and transportation could add up to ~3,000 terawatts to U.S. electricity demand by 2050. This is likely an overestimate, as it does not account for energy efficiency improvements, like demand response.
AI & Data Centers
Investment in data centers is increasing consistently, with Alphabet, Meta, Microsoft, and Amazon investing more than $130B in data centers in 2023. In 2022, America’s 2,700 data centers used ~4% of the nation’s total electricity. McKinsey projections show an increase to 6% in 2026.
AI is driving a significant portion of the demand, as training and running LLMs are energy expensive and often occur in hyperscale data centers. It can be hard to predict the exact power consumption of training an LLM, but researchers estimate that creating GPT3 (175B parameters) consumed 1,287 MWh of electricity and generated 552 tons of carbon dioxide.[vii]
Servicing GenAI queries can be hard to predict as well, as volume and complexity of queries affect consumption.
A simple example of a difference in need is shown below, comparing the energy consumption of an AI-powered search to that of a Google search.
A recent study in Joule[viii] found that by 2027 AI data consumption (as opposed to existing data center usage) could add 85-134 Terawatt hours of annual electric load, or more power than will be generated by ALL of the solar added globally in 2023 (33 GW added 8760 hours/year 20% capacity factor is 57.8 TWh).
There is much more to dive deeper into here, including GPU hardware improvements, domain-specific models, advanced cooling tools, and grid-edge AI – levers that could help drive energy efficiency in the ecosystem.
The Bottom Line
We can’t predict exactly which of these will happen or at what speed, but for the first time in decades, we have multiple sources of significant load growth happening at the same time. There are tons of innovations and solutions, everything from energy storage to DERs, that are popping up to tackle load growth demands and interconnect scarcity, but we’ll save those for another blog.
[iv] https://www.mckinsey.com/industries/oil-and-gas/our-insights/global-energy-perspective-2023-ccus-outlook
[vi] https://www.brattle.com/insights-events/publications/transportation-and-heating-electrification-could-aid-in-increasing-utility-sales-and-reducing-carbon-emissions-according-to-brattle-economists/#:~:text=Brattle%20economists%20released%20a%20whitepaper%20that%20shows%20the,more%20through%20the%20electrification%20of%20transportation%20and%20heating.
[vii] https://www.forbes.com/sites/tiriasresearch/2023/05/12/generative-ai-breaks-the-data-center-data-center-infrastructure-and-operating-costs-projected-to-increase-to-over-76-billion-by-2028/?sh=fe387e67c15e
[viii] https://www.cell.com/joule/abstract/S2542-4351(23)00365-3?_returnURL=https%3A%2F%2Flinkinghub.elsevier.com%2Fretrieve%2Fpii%2FS2542435123003653%3Fshowall%3Dtrue
The views expressed here are those of the individual Keyframe Capital personnel quoted and are not the views of Keyframe Capital or its affiliates. Certain information contained in here has been obtained from third-party sources, including from portfolio companies of funds managed by Keyframe Capital. While taken from sources believed to be reliable, Keyframe has not independently verified such information and makes no representations about the enduring accuracy of the information or its appropriateness for a given situation.
This content is provided for informational purposes only, and should not be relied upon as legal, business, investment, or tax advice. You should consult your own advisers as to those matters. References to any securities or assets are for illustrative purposes only, and do not constitute an investment recommendation or offer to provide investment advisory services. Furthermore, this content is not directed at nor intended for use by any investors or prospective investors, and may not under any circumstances be relied upon when making a decision to invest in any fund managed by Keyframe Capital. Any investments or portfolio companies mentioned, referred to, or described are not representative of all investments in vehicles managed by Keyframe Capital, and there can be no assurance that the investments will be profitable or that other investments made in the future will have similar characteristics or results.
Charts and graphs provided within are for informational purposes solely and should not be relied upon when making any investment decision. Past performance is not indicative of future results. The content speaks only as of the date indicated. Any projections, estimates, forecasts, targets, prospects, and/or opinions expressed in these materials are subject to change without notice and may differ or be contrary to opinions expressed by others.
