Welcome to EPRI's Thought Leadership Hub
Thought Leadership
SPEED TO POWER
Strategies for Powering Data Centers in the AI Era
EXPLORE
Thought Leadership Hub is a collection of emerging EPRI research focused on transforming the energy landscape.
TOGETHER…SHAPING THE FUTURE OF ENERGY®
POWERED BY
WIN-WINWATTS
How Rising Electricity Demand Can LowerElectricity Rates
THE ENERGY WALLET
Household Energy Expenditures
DatacenterSiting
The Power Industry's Playbook On Siting
Quantum Computing
Technical Assessment for Power Infra-structure Planning
Powerful Pathways of Transmission
Accelerating R&D to Unlock New Transmission Capacity
your go-to source for the latest in energy innovation, research, and insights.
TUNE IN
BUILDING TRUST AT SCALE
What History Teaches about Winning Public Confidence in AI Infrastructure
Open Power AI Consortium
Uniting the Electric Sector with Advanced AI Solutions
Partnering for Progress
Community-Centered Data Center
PROGRESS PARADOX
FROM STEEL to SILICON
Grid Challenge:
Crossing the Valley of Death
EPRI 3420 Hillview Avenue, Palo Alto, California 94304-1338 USA • 800.313.3774 • 650.855.2121 • askepri@epri.com • www.epri.com © 2026 Electric Power Research Institute (EPRI), Inc. All rights reserved. Electric Power Research Institute, EPRI, and TOGETHER…SHAPING THE FUTURE OF ENERGY are registered marks of the Electric Power Research Institute, Inc. in the U.S. and worldwide.
Back to Home
Speed to Power Strategies for Powering Data Centers in the AI Era April 2026
OVERVIEW
Data‑center demand is surging—especially with AI driving power needs up to 10× higher. The grid still delivers the best long‑term reliability, cost, and emissions performance, but slow interconnection timelines are pushing new power strategies. Four models are emerging: Grid‑Connected Inflexible (today’s norm) Grid‑Connected Flexible (adds on‑site resources + load shifting) Bridge‑to‑Grid (start off‑grid, transition later) Islanded (fully off‑grid) Flexible and bridge approaches can speed deployment while preserving grid benefits, but scaling them requires tight coordination across developers, utilities, and regulators. EPRI’s DCFlex initiative is already demonstrating what’s possible.
TIME TO MARKET
1
2
3
4
5
6
7
8
9
10
11
0
Planning & Permitting
BUILD
7 yrs
3 yrs
Transmission Infrastructure
4 yrs
2 yrs
Synchronous Generators
3.5 yrs
1.5 yrs
Inverted-BasedResources
8 yrs
1 yr
DATACENTERPOWERINGSTRATEGIES
Grid-ConnectedInflexible
Grid-ConnectedFlexible
Bridge-to-Grid(Off-Grid Temporarily)
Islanded(Off-Grid Permanently)
EPRI’s DCFlex initiative is already demonstrating what’s possible.
Years
2.5 yrs
Data Centers
dcflex.epri.com
LEARN MORE
www.speed2power.epri.com
Win-Win-Watts How Rising Electricity Demand Can Lower Electricity Rates April 2026
New electricity demand from data centers, electrification, and other sources can sometimes lower average retail electricity prices, depending on whether the cost of serving additional load is below current average system costs. When systems have spare capacity or access to low-cost new generation, added demand can spread fixed costs across more kilowatt-hours, improving asset utilization and reducing prices; however, if new infrastructure is required, costs may rise. Key factors influencing outcomes include load characteristics, proximity to infrastructure, and the cost of grid upgrades. Evidence suggests that under the right conditions, load growth can help moderate price increases while supporting clean energy investment. Achieving these benefits depends on proactive planning, effective rate design that prevents cost shifting, and demand flexibility to reduce peak strain. When aligned, these elements can enable lower prices, improved grid efficiency, accelerated clean energy deployment, and reduced emissions, though success ultimately hinges on policy, infrastructure needs, and emerging solutions like flexible data center loads.
www.winwin.epri.com
The Energy Wallet U.S. and State-Level Household Energy Expenditures, Past, Present, and Future May 2026
The “energy wallet” measures total household energy spending—including electricity, heating fuels, gasoline, and EV charging—providing a more complete view of affordability than electricity bills alone. In 2024, U.S. households spent an average of $5,530 annually on energy, with gasoline accounting for the largest share at $2,930 and electricity totaling $1,850, representing less than half of total energy costs in every state. While overall energy spending has remained relatively stable since 2000 when adjusted for inflation, it has fluctuated with fuel prices, whereas electricity costs have stayed comparatively steady. Looking ahead, electrification—particularly of vehicles—combined with efficiency improvements, is projected to significantly reduce household energy costs, with total spending expected to decline by 36–42% by 2050 as gasoline use is replaced by more efficient electric technologies.
www.energywallet.epri.com
See how your state compares:
energywallet.epri.com
Powerful Pathways of Transmission Accelerating R&D to Unlock New Transmission Capacity May 2026
Grid-Enhancing Technologies (GETs) can increase the power transfer capacity and utilization of existing and future transmission lines. These hardware and/or software solutions can enhance grid capacity, efficiency, reliability, and safety. GETs can be deployed relatively quickly, allowing them to act as a bridge while new transmission is built. If implemented effectively, their impact on grid capacity can be substantial.
Series 1: https://restservice.epri.com/publicattachment/98354Series 2: https://restservice.epri.com/publicattachment/98355Series 3: https://restservice.epri.com/publicattachment/98356Series 4: https://restservice.epri.com/publicattachment/98357Series 5: https://restservice.epri.com/publicattachment/98358
Datacenter Siting Integrating Data Centers into the Grid: Benefits of Flexibility and Strategic Siting May 2026
The rapid expansion of data centers is challenging grid reliability and transmission capacity. This study explores how strategic siting and load flexibility can accelerate data center integration while maintaining system stability. Using a stylized Texas grid model for 2030, we assess hosting capacity under different transmission and siting scenarios and evaluate four flexibility archetypes. Results show that strategic siting in energy-rich zones can nearly triple integrable capacity compared to uniform placement, while flexibility services—such as spinning reserves and load curtailment—provide evolving reliability benefits as data center penetration grows. Findings underscore the importance of coordinated planning, flexibility-enabled architectures, and transmission enhancements to unlock hosting capacity and support resilient grid operations.
www.epri.com/research/products/3002033797
Quantum Computing Technical Analysis for Power Infrastructure Planning May 2026
Questions have emerged regarding whether quantum computing advances will displace Graphics Processing Unit (GPU)-based AI infrastructure. This analysis examines the technical characteristics of both computing architectures, their respective applications, additional emerging compute modalities, and implications for long-term infrastructure planning. AI workloads increasingly demand high-performance computing. Today, GPU computing dominates AI training and inference, while quantum computing is emerging as a complementary technology for specialized tasks. Understanding their differences is critical for strategic planning. The fundamental conclusion is that these technologies often address different computational problems and will operate largely in complementary rather than competitive roles especially with the advent of hybrid classical-quantum architectures.
www.epri.com/research/products/000000003002034644
Open Power AI Consortium Uniting the Electric Sector with Advanced AI Solutions May 2026
The Open Power AI Consortium aims to evolve the electric sector by leveraging advanced AI technologies to innovate the way electricity is made, moved, and used by customers. By fostering collaboration among industry leaders, researchers, and technology providers, the consortium will drive the development and deployment of cutting-edge AI solutions tailored to enhance operational efficiencies, increase resiliency and reliability, deploy emerging and sustainable technologies, and reduce costs while improving the customer experience.
www.openpowerai.org
Partnering for Progress: Understanding Community-Centered Data Center Development June 2026
Data center growth is essential to support AI and the digital economy, but local communities are increasingly concerned about impacts on energy, water, infrastructure, and the environment. As a result, community acceptance has become a critical factor in project success, with opposition leading to more cancellations, delays, and regulatory challenges. Successful projects typically involve early engagement, transparent communication, and measurable community benefits that build trust and demonstrate long-term value.
www.community-centered.epri.com
The Real Grid Challenge: Crossing the Valley of Death July 2026
The "Molecular Bridge to Energy Reliability" illustrates the strategic necessity of maintaining grid stability during the transition between periods of high and low renewable energy production. It characterizes the energy challenge as navigating both the Daily Gulf of 24-hour solar variability and the more severe Valley of Death, which represents extended weeks or months-long droughts of variable renewable energy. This bridge spans the gap from initial renewable abundance to the eventual return of power by utilizing four core resources: flexible natural gas, steady base-load power from coal and high-density uranium, and long-duration energy storage that captures surplus power for use during these durational gorges.
https://www.linkedin.com/pulse/real-grid-challenge-crossing-valley-death-arshad-mansoor-rukde/
THE PROGRESS PARADOX: FROM STEEL to SILICON June 2026
The Progress Paradox illustrates a transformative shift from the industrial "Steel Era" to the digital "Silicon Era," characterized by a dramatic acceleration in infrastructure needs. While foundational technologies like railroads and electricity navigated adoption curves lasting 30 to 50 years, modern AI data centers are scaling at an unprecedented rate of only 2 to 3 years. This rapid expansion creates a massive 1 GW power demand—equivalent to the energy used by 800,000 U.S. homes—which necessitates a level of infrastructure coordination never before seen. Historically, progress was often met with decades of organized resistance or restrictive "Red Flag Laws," but today's challenges are further complicated by a shift in societal values; communities now prioritize perceived quality-of-life impacts three times more than the affordability of the technology itself. Consequently, securing the necessary local trust and credibility for rapid deployment now depends heavily on visible accountability and environmental mitigation
BUILDING TRUST AT SCALE: What History Teaches about Earning Public Confidence in AI Infrastructure October 2026
History suggests that public trust in transformative technology is earned through performance and evidence. By analyzing past transitions, the paper frames five factors as an evaluation framework to help decision-makers navigate the current expansion of AI and data center infrastructure. The paper reflects an editorial perspective on historical patterns rather than a prescriptive rule or guarantee.
White paper: link Executive Summary: link Podcast: www.podbean.com/ew/pb-6r4a2-1b7b27d Infographic: restservice.epri.com/publicattachment/100625