Paving the Way for Faster Access to Power
Irene Danti Lopez is EPRI senior team lead.
Rapid growth in artificial intelligence, cloud computing, electrification, and advanced manufacturing is driving growth in electricity demand faster than new infrastructure can be developed to meet it. In many regions, access to power — rather than capital, land, or technology — has become the primary constraint on the deployment of large loads.
As regulators, utilities, and system operators grapple with this challenge, one powerful opportunity for data centers is becoming increasingly clear: large load flexibility. Modern data centers, as well as other large loads, can be designed and operated to support grid reliability, manage congestion, and provide support during periods of system stress. Despite this technical potential, leveraging flexibility capabilities to accelerate time to power has been challenging in practice.
The reason is not technology. It is language.
The Flexibility Gap in Today’s Interconnection Process
Today, large loads lack clear signals regarding which flexibility capabilities are most valuable to the grid during their design process. Additionally, flexibility may mean different things in different regions: requirements vary by utility, market, and jurisdiction. Without common definitions or performance expectations, developers can struggle to design facilities that align with grid needs, making it difficult to plan for flexibility.
Electric utilities and system operators, in turn, face fragmented processes for interpreting large load flexibility. Often, each large load project may be modeled as a custom case, increasing study time and uncertainty regarding how flexibility will be delivered. Also, conservative assumptions at the planning phase can overlook the existing grid headroom that could be unlocked with flexible assets.
The result is longer interconnection timelines, reduced confidence, and missed opportunities to maximize the usage of existing grid assets to unlock available capacity.
Introducing a Unified Framework: DCFlex and Flex MOSAIC™
To address this gap, EPRI’s DCFlex initiative launched the Flex MOSAIC™ classification system in March. The framework provides a technology neutral, uniform structure for describing and evaluating large load flexibility.
At its core, the framework establishes a shared language. It defines flexibility capabilities using practical, grid-relevant performance attributes — such as magnitude, timing, duration, frequency, ramp rate, and advance notice requirements — and organizes them into a limited set of uniform flexibility classes. These classes translate technical capability into terms that grid planners and operators can consistently interpret and apply.
Rather than prescribing specific technologies or operating models, the framework focuses on outcomes: what flexibility large loads can provide, under what conditions, and with what degree of reliability. It covers flexibility needs driven by both short-term and longer-duration events on the grid, as well as the option for large loads to be fully interactive with the grid.
NERC President Jim Robb remarked, “The framework is meant to provide a technical foundation that jurisdictions and market participants can adapt to their local needs. As large, flexible loads play a growing role in the power system, having clear, technically grounded definitions of flexibility are critical for reliability. A common framework like this can help system operators and planners speak the same language, essential for maintaining a reliable grid.”
NARUC President Ann Rendahl added, “As demand from data centers accelerates, state regulators are focused on ensuring customers are not burdened by the costs of serving new, large loads, as well as maintaining grid reliability.
NARUC looks forward to engaging with EPRI and others on how a voluntary, standardized framework like Flex MOSAIC™ can create a common language and shared understanding of flexibility and provide benefits to state regulators when evaluating data center integration, without shifting costs to customers or compromising grid reliability.”
NARUC and NERC join 45+ participants in Flex MOSAIC™, including utilities, hyperscalers, equipment manufacturers, and data center developers and operators, with more joining each week.
From Concept to Actionable Planning Tool
One of the most important attributes of the DCFlex framework is its immediacy. Flex MOSAIC™ is not a future market design or a regulatory mandate. It is a practical planning and communication tool that can be applied today — within existing interconnection studies, hosting capacity analyses, and reliability discussions.
By increasing transparency and trust, utilities and system operators gain new tools to manage grid stress while integrating large loads with confidence. For developers, it creates a pathway to design facilities with flexibility from the outset, improving predictability and expanding viable siting options.
Flex MOSAIC™ provides a common technical foundation that jurisdictions, markets, and system operators can tailor to regional requirements. This shared understanding can shorten study timelines and enable faster and confident planning decisions, without sacrificing reliability or affordability for electric customers.
Aligning Large Loads with Real Grid Needs
Uniform flexibility classes do more than standardize definitions; they align large loads with the specific challenges grid operators face every day. Different classes map to different system needs, including peak reduction, congestion management, emergency response, fast balancing, and longer duration capacity support.
By clearly linking flexibility capabilities to operational needs, the framework helps planners move beyond abstract concepts and toward actionable decisions. In many cases, these benefits can be realized without new infrastructure or major policy changes, offering near-term relief as demand continues to accelerate.
Proof Through Demonstration
The framework is informed by a growing portfolio of DCFlex demonstration projects that test flexibility in real-world and simulated environments. These projects show that data center compute loads, both AI and nonAI, can be modulated across a range of durations and response times, with varying levels of advance notice. Other demonstrations explore geospatial load shifting, grid interactive controls, and cleaner backup generation fuels.
Collectively, these efforts provide empirical grounding for the framework and help translate theoretical flexibility into operational confidence.
A Foundation Built for Collaboration and Speed to Power
As large, flexible loads assume a central role in electricity demand growth, the stakes are rising. Without coordination, the energy industry risks a fractured landscape of definitions and requirements, slowing deployment precisely when speed matters most.
Flex MOSAIC™ is intentionally designed to be adaptable — supporting pilots, voluntary programs, and regional implementation with a shared and consistent foundation. By establishing a unified, credible language for largeload flexibility, Flex MOSAIC™ can pave the way for faster access to power, more confident planning, and a more resilient grid that can support the growth of an AI-driven economy.
For more information and to join participants in advancing the framework, visit Flex MOSAIC™ | DCFlex.
