CALGARY, Alberta (March 11, 2025) — Enverus Intelligence® Research (EIR), a subsidiary of Enverus, the most trusted energy-dedicated SaaS company that leverages generative AI across its solutions, is releasing a series of reports analyzing long-term load forecast, providing an update on load growth and total system load, and the unprecedented capital expenditures cycle in the technology sector. Driven by the focus on scaling AI capabilities, the reports examine the implications for U.S. electricity markets, particularly in terms of data center capacity growth and maintenance.
EIR’s modeling considers not only historical drivers of power demand across the Lower 48, but also variable load drivers it believes will impact the future load. The organization has also revised its outlook on residential solar and storage, data center projections — including hyperscalers — and reshored manufacturing and the effects that these new exponential load drivers will have on EIR’s power demand forecasts from 2025-2050.
“Total power demand is set to climb 30% by 2050, driven primarily by artificial intelligence,” said Carson Kearl, analyst at EIR. “In the short term, EIR predicts a 15.83% rise in total annual energy consumption through 2035, stemming from a rapid data center buildout and rising electric vehicle adoption.”
“Reshoring manufacturing also has a modest impact on our load forecast, increasing average load by less than 3% by 2050. However, it will have a more pronounced regional effect in the Pennsylvania-New Jersey-Maryland (PJM) and the Southeast Power Pool ISOs,” Kearl said.
“If you are a believer in the gigawatt scale data center story, you should have long exposure to data center supply chain equities where ~$30B/GW will flow into high margin businesses, compared to ~$2B/GW in energy. If you are skeptical of hyperscaler growth beyond this initial rollout, both industries are overbought after the last 12 months of multiple expansion,” Kearl said.
Key takeaways from the report series:
- Total load is forecast to grow 30% by 2050, driven by increased data center and baseload demand and overall electrification trends, partially offset by residential solar and storage expansion.
- EIR predicts a 15.83% rise in total annual energy consumption through 2035, stemming from a rapid data center buildout and rising electric vehicle adoption.
- Overall, we anticipate significant load growth in Electric Reliability Council of Texas (ERCOT), PJM and WEST regions by 2050. Daily load profile shapes are expected to evolve with greater adoption of residential solar and storage, as well as industrial load growth.
- Reshoring manufacturing contributes minimally to our load forecast, lifting average load less than 3% by 2050. It will have more regional impacts on PJM and Southeast regions.
- The technology sector is entering an unprecedented capital expenditures cycle focused on scaling AI capabilities, with hyperscalers’ annual U.S. AI spend expected to exceed $50 billion each by the end of the decade.
- Capital is allocated between growing and maintaining datacenter capacity, with chips on a five-year replacement cycle costing ~$27 billion per GW. The ratio of growth to maintenance capex will shift significantly post-2030, making it harder for hyperscalers to expand their capacity stacks.
- PJM will see the largest share of data center growth, followed by West, ERCOT and California Independent System Operator (CAISO). ERCOT will host the first official Stargate facility, totaling 2.2 GW by 2027, one of the largest data center campuses in the world.
- Power capacity will nearly triple by 2035, but efficiency gains will result in a 31-fold increase in total compute. Leading large language models now outperform human experts on the Massive Multitask Language Understanding benchmark, with significant cost reductions in model output.
EIR’s analysis pulls from a variety of Enverus products including Enverus Foundations® | Power & Renewables and Enverus Foundations® | Carbon Innovation.
You must be an Enverus Intelligence® subscriber to access these reports.
About Enverus Intelligence® Research
Enverus Intelligence ® | Research, Inc. (EIR) is a subsidiary of Enverus that publishes energy-sector research focused on the oil, natural gas, power and renewable industries. EIR publishes reports including asset and company valuations, resource assessments, technical evaluations and macro-economic forecasts; and helps make intelligent connections for energy industry participants, service companies and capital providers worldwide. Enverus is the most trusted, energy-dedicated SaaS company, with a platform built to create value from generative AI, offering real-time access to analytics, insights and benchmark cost and revenue data sourced from our partnerships to 95% of U.S. energy producers, and more than 40,000 suppliers. Learn more at Enverus.com.
