A common language for data centre flexibility

As demand for digital infrastructure surges, data centres and other large-scale loads are being connected to power systems faster and at greater scale than ever before. Eamonn Lannoye, Director of Research & Development at EPRI Europe, examines the challenges and opportunities this rapid growth presents for grid planning, reliability and the future of electrification.

In Europe, this trend is stress testing existing planning timelines for energy assets. For example, new data centre campuses can come online much faster than supporting infrastructure such as substations, transmission networks and generation capacity can be built. As a result, 'time to power' has become a critical constraint, even in regions that appear well supplied on paper. 

Grid planners are typically cautious. Interconnection studies and planning processes are built to protect reliability, so they lean on worst case assumptions about load behaviour. That conservatism is a safeguard, but it can also create an unintended bottleneck. When every large load is modelled as a static peak at the system’s most constrained hour, it can slow interconnection queue progression, increase uncertainty and force each project to rely on customised modelling. 

The flexibility bottleneck

A major obstacle to integrating large loads is the lack of a unified understanding of flexibility. Data centre developers often lack clear signals about which grid supporting capabilities are most valuable during design, and flexibility is interpreted differently depending on the utility, market or jurisdiction. Without standard definitions and transparent expectations, planning for flexibility or designing facilities that meet system needs can become challenging. 

Utilities and system operators also struggle without a common framework, often resorting to fragmented, project specific approaches that require each request to be studied individually. This can increase modelling effort, uncertainty and cost.  

Flex MOSAIC™: A unified language for data centre flexibility  

EPRI’s DCFlex initiative demonstrates how data centres can support and stabilise the electric grid while boosting interconnection and efficiency. However, the industry lacks a shared, credible way to define and trust flexibility. There are inconsistencies across regions and markets that make it difficult for large load customers to understand what flexibility characteristics are most valuable to the grid.  Similarly, utilities and system operators face inconsistent and fragmented processes for interpreting and incorporating flexibility into planning and operation.

To address this challenge, the voluntary Flex MOSAIC™ framework was developed, offering a clear, technology neutral method to describe and measure flexibility in large electrical loads.  

Flex MOSAIC organises key performance factors such as load size, timing, duration, frequency, speed of response and advance notice into straightforward categories, creating a unified framework for flexibility.  

These unified standards help planners, operators, regulators and developers efficiently assess solutions and support early design integration. Its adaptable technical details suit different regions and markets, streamlining planning while preserving reliability and affordability. It also connects flexibility categories to practical grid needs such as peak reduction, congestion management and capacity support, enabling clear, actionable strategies for grid improvement. 

Aligning large loads with real grid needs  

Uniform flexibility classes do more than standardise terminology; they connect large loads directly to the operational challenges grid operators face every day. Each class in the Flex MOSAIC framework maps specific system needs such as peak reduction, congestion management, emergency response, fast balancing or longer duration capacity support. 

As large, flexible electrical loads become more common, they are changing how electricity is used and managed. In Europe, this shift makes coordination even more critical because limited investments and long wait times for grid connections are slowing the pace of infrastructure development. 

Flex MOSAIC offers a practical, adaptable foundation for large load flexibility that provides consistent definitions planners and developers can trust across regions. By establishing a clear, shared language for flexibility, the framework can streamline planning and interconnection, enabling more predictable outcomes, smarter use of existing grid capacity, and phased connection strategies that allow earlier energisation without compromising reliability.  

Greater transparency and a consistent framework benefit all parties. Utilities and system operators gain greater confidence in integrating large, flexible loads while maintaining reliability. Meanwhile, developers can proactively design facilities with flexibility in mind, unlocking faster and more predictable grid connections, and expanding siting options for projects. Notably, several leading companies, including Google, Honeywell, ING, Jenbacher (INNIO Group), KPMG, Meta, NVIDIA, Octopus Energy and Siemens, are already collaborating with EPRI on the Flex MOSAIC™ initiative.

EPRI is inviting utilities, operators, data centre developers, hyperscalers and policymakers to co-develop, pilot and adopt Flex MOSAIC. 

For more information and to join participants in advancing the framework, visit  Flex MOSAIC™ | DCFlex.

About the author