Data center flexibility: Foreword + intent

First part of impactECI's framework

Foreword

It’s no secret that America’s digital ambitions are outpacing its energy reality. AI, cloud computing, and advanced analytics are driving the biggest wave of data center growth in history, and its power demand is set to more than double in just a few years. But while servers and software scale rapidly, the grid currently doesn’t. Connecting new load now takes years, not quarters, and the queue keeps growing.

At impactECI, we’ve spent the past few years working on this challenge, helping hyperscalers connect GWs of new capacity, advising developers on siting and grid strategy, and supporting utilities and transmission operators as they plan how to connect increasing load. Unlike many who talk about data centers from the sidelines, we’ve been in the room, translating between tech companies’ need-for-speed and utilities’ need for reliability and fair processes to find common ground.

This paper reflects what we’ve learned. Data centers don’t have to be grid liabilities, they can be powerful partners in flexibility, reliability, resilience, and decarbonization. Through smarter coordination and flexible interconnection, existing grid infrastructure could support far more capacity than most believe possible.

The moment to act is now. Over the next year, following the emergence of a new data center flexibility industry in 2025, we’ll see data centers and the grid evolve together. The technology is ready. The economics work. What’s needed is leadership and collaboration across utilities, operators, regulators, and tech companies.

As you read this, imagine a future where digital and energy systems grow side by side faster, cheaper, cleaner, and stronger. That future is within reach if we choose to move together.

Peter Hans Hirschboeck

Founder & Managing Director

impactECI

November 2025

Document intent and purpose statement

Purpose and intent

Recent research shows that data center flexibility has real value across the energy grid. MIT’s GenX modeling shows that shifting data center operations in time and location can cut total system costs.¹ Emerald AI and Digital Realty are taking this concept into the real world. Their pilot 96 MW Aurora AI Factory, in conjunction NVIDIA, EPRI, and PJM, is designed to prove what flexible data centers can actually do at scale.² The project aims to remain fully SLA-compliant for customers yet shifts its power use in real time to help the grid stay balanced.

Based on these lessons, our framework, detailed in the coming series of articles, provides a path for America to better coordinate its digital and energy systems. We’ve developed technical, economic, regulatory, and coordination steps needed to change data centers from simple electricity users to active grid participants. We identify the bottlenecks that remain and must be overcome to achieve this vision. Our proposed framework will speed up infrastructure build out and help meet grid modernization and decarbonization goals.

Strategic purpose

The U.S. power grid faces a coordination problem on a scale not seen before. According to Wood Mackenzie, as of June 2025, over 134 GW of data center capacity are waiting to be connected, with wait time reaching up to eight years in some major data center markets.³⁴ Not every project will be built, but the scale alone illustrates the urgency for new planning, interconnection, and coordination models. Traditional grid expansion processes, built for predictable demographic and economic growth, are no longer compatible with today’s rapid AI and cloud infrastructure boom.

This framework aims to fix that problem by using the ability of data centers to flex their operations as a short-term solution; a matter of hours, when needed, for a few dozen days per year. This can give the grid support today, buying time for low-cost and low-carbon energy options to come online, like new nuclear power, long-duration energy storage paired with renewables, and enhanced/advanced geothermal, over the next ten years.

This adaptation is urgent. Ireland offers a clear example. In 2022, grid operator EirGrid paused new data center connections, especially around Dublin, after power demand surged beyond what the grid could handle.⁵ The freeze has stalled projects and redirected billions in investment to other countries. In contrast, the Southwest Power Pool’s 90-day connection trial shows that planned changes can work.⁶ Texas is taking a bolder step. Senate Bill 06 (SB06), passed in June 2025, introduced a stricter requirement on data centers, including cost sharing for grid upgrade, mandatory information sharing about behind the meter assets, and load curtailment during emergency events.⁷ Recently, impactECI has also co-sponsored interconnection reform proposals for ERCOT led by Arushi Sharma Frank, a leading expert in energy system flexibility. We are also excited to support the DOE goals and FERC efforts in interconnection reform for large loads. Together, we’ve focused on operational and policy innovation to accelerate large load integration into the power grid. The U.S. still has a window to act, but it’s narrowing fast.

Target audience and use cases

Primary stakeholders

• Utility leaders and grid managers who need ways to handle the quick rise in connection requests without hurting reliability or costing too much

• Data center operators and large cloud service providers looking for ways to connect to the grid sooner and create new income by being able to change how they operate

• State and national regulators who are balancing new ideas and reliability as they make rules for the quickly changing energy field

• Technology companies creating systems, tools, and hardware to allow flexible involvement in the grid

• Investors and lenders considering new markets for flexible services and improvements to the grid

Practical applications

• Regulatory documents: Models for adding flexibility into rates and connection rules

• Utility planning: Ways to include flexible demand into plans for resources and transmission

• Data center investment: Tools to judge design and operation plans based on flexibility

• Policy creation: Good practices from top international markets that can be changed to fit

• Technology matching: Common rules for coordination and performance verification

Stakeholder Value

Across utilities, operators, and communities, it creates measurable value by cutting system costs, accelerating investment, and boosting economic growth.

• For utilities, studies by Gridlab and NV Energy find that 1-2 GW curtailment can cut system costs by $300-400+M in NPV primarily through avoiding capacity additions and upgrading transmission infrastructure.⁸ For utilities, it’s a way to meet rising demand without overbuilding the grid. A small amount of curtailment, 0.25-1% can unlock about 76-126 GW of grid capacity according to Tyler Norris, now at Google but formerly at Duke University’s Nicholas Institute, and supported by our own research, enabling faster growth while maintaining reliability.⁹

• For data centers, flexibility speeds interconnection, lowers operating expenses, and adds new revenue streams through participation in grid services. Flexibility allows projects to come online years earlier and pulls forward billions in revenue for operators and developers. Beyond the economics, it reshapes perception: data centers become active contributors to a resilient grid rather than a burden shared by ratepayers.

• At community level: As data centers connect earlier, they bring forward construction activity, tax revenue, and infrastructure improvements while enabling a more stable rate. Each 100 MW data center drives ~$243.5M in local economic activities, creates around 150 permanent jobs and brings $7.8M in annual wages during operation.¹⁰ Flexibility may also help avoid the construction of new transmission lines and substations, which are often not welcomed by communities. At some data center campuses, it might even prevent the need for on-site primary and/or secondary power fossil generation plants which carry the cost of local air pollutants and CO₂e emissions.

Unique value proposition

This framework represents what we believe to be the first comprehensive, cross-sector blueprint for building the data center flexibility ecosystem at scale. It goes beyond pilot studies or isolated technical discussions, delivering a full architecture for how utilities, hyperscalers, regulators, and investors can coordinate to unlock shared value.

• Systematic integration: A four-pillar framework (compute, energy resource, infrastructure, and contractual flexibility) that links operational capability to commercial structure

• Stakeholder coordination: A six-tier coordination model that enables collaboration while preserving market competition and balanced risk allocation

Differentiation from existing approaches

• More than just demand response: This goes past just cutting load to include workload shifting, geographic balancing, and active market participation. This keeps up-time while giving clear system benefits

• From one-on-one to group-wide: This goes past just deals between utilities and customers to include many groups across different regulatory and market structures

• A way to get there, not the end: Flexibility is a way to get past today’s limits and move toward a cleaner, more reliable generation mix in the future, avoiding new construction of unnecessary fossil fuel generation facilities

• Global learning, local use: International examples, best practices, and grid programs, tailored to U.S. market designs and realities

Intended outcomes

We proposed a three-phase deployment roadmap, Foundation (next 18 months), Commercial scale (2027–2028), Ecosystem growth (2029–2030), offering a realistic path to adoption:

Foundation (next 18 months)

• Regulatory plans in at least five states to establish organized flexibility programs

• Industry-standard rules to ensure interoperability and competitive vendor ecosystems

• Federal guidance through FERC Advance Notice of Proposed Rulemaking (ANOPR) and DOE comment tracks, formally recognizing flexibility load in wholesale markets, aligned with ERCOT, SPP and other RTOs/ISOs emerging “connect-and-manage” interconnection models

• At least 10 utility-data center pilots demonstrate commercial, operational, reliability and coordination benefits

Commercial Scale (2027–2028)

• Expansion across major utilities

• Verified system cost savings and grid reliability improvements

• Regional coordination frameworks enabling multi-state scaling

• Competitive market and tariff models for flexible services

Ecosystem Growth (2029–2030)

• Smooth integration between grid and data center operations

• Flexible loads as a core reliability asset supporting firm clean-tech deployment

• U.S. leadership in digital infrastructure - energy infrastructure nexus

Measuring success

• Technical: Proven, reliable flexibility without compromising uptime

• Financial: Documented cost savings and shared economic value

• Collaborative: Established frameworks for continuous innovation and transparent regulation.

This framework calls for data centers to play an active role in supporting grid reliability and resilience. Flexibility gives them the power to support the grid, not strain it: accelerating interconnections, lowering costs, and improving resilience as demand grows. Collaboration among utilities, operators, regulators, and investors will be critical to realizing these benefits. With rapid adoption and innovation, data centers can become essential partners for a cleaner and more resilient energy future.

Schedule of Future Chapter Releases

1. https://www.prnewswire.com/news-releases/epris-data-center-flexibility-and-grid-reliability-initiative-expands-to-europe-302382800.html

2. https://emeraldai.co/blog/launching-the-first-power-flexible-ai-factory-with-nvidia

3. https://www.woodmac.com/press-releases/with-the-us-data-center-pipeline-surging-regulated-utilities-are-proving-advantaged-in-meeting-demand/

4. https://www.camus.energy/blog/how-flexible-interconnections-can-help-data-centers-connect-faster-without-overloading-the-grid

5. https://www.datacenterdynamics.com/en/news/eirgrid-says-no-new-applications-for-data-centers-in-dublin-till-2028/

6. https://www.spp.org/news-list/southwest-power-pool-board-approves-accelerated-pathway-for-large-load-connection/

7. https://capitol.texas.gov/BillLookup/History.aspx?LegSess=89R&Bill=SB6

8. https://gridlab.org/portfolio-item/data-center-flexibility-nv-energy-case-study-report/

9. https://nicholasinstitute.duke.edu/sites/default/files/publications/rethinking-load-growth.pdf

10. https://www.uschamber.com/assets/documents/ctec_datacenterrpt_lowres.pdf