Back to articles

What Microgrids Deliver for AI Edge Data Centre Developers

Category
AVK thinking
Topic
innovation
Date
27 January 2026
Author
AVK
Read Time
10 min

How microgrids are solving the disconnect between AI Edge data centre power demand and grid reinforcement timelines in cities across Europe.

AVK microgrid for independent power generation for data centres

At a time when demand for electricity is set to surge from data centres, transport and heating the power capacity outlook for Europe’s cities looks challenging for generation companies, transmission and distribution systems operators (TSOs, DSOs).

A power system built on large turbine halls in far away places means cities tend to be demand centres with limited large scale generation close by. Attempts to secure new capacity at scale in or close proximity to cities means navigating a complex mix of regulation, constraints and regional disparities. At the same time data centre developments, especially of AI Edge data centres, are faced with potential capacity shortfalls and extended grid connection timelines.

Andrew Jay, Head of EMEA Data Centre Solutions, Advisory & Transaction Services at property firm CBRE, recently wrote about the impact of AI on powering Europe’s data centres saying: “The rise in consumption is significant, and for many cities, particularly within the FLAPD cluster, local grids are now saturated. Securing power has become the defining constraint, as much as land availability or planning permission.”

Grid changes across Europe

An examination of power grid capacity and power demand forecast to 2030 for cities such as London, Manchester, Paris, Lyon, Amsterdam, Berlin, Munich and Barcelona reveals some worrying trends.

A breakdown of what is happening shows some common issues around demand potentially outstripping capacity. The overall picture is one of uncertainty. Almost all metros will face constraints by 2030 if renewable expansion or flexible gas capacity is delayed.

This is leading to real supply risk with multi-year connection queues to access new generation in almost every large metro. Common factors include increased demand due to competition for connections and the rapid move to variable renewables with offshore wind and solar on the network set to rise materially by 2030.

Another commonality for AI Edge data centre developers is navigating reform of grid connection processes. The UK’s connection queue will move from a “first come, first connected” process to a “first ready and needed, first connected” system.

Ofgem says: “Industries of the future, ranging from data centres and AI to wind and solar projects, will be pushed to the front of the queue for grid connections.” (Source: Latham Watkins.) While this could be positive for some data centres, the competition will remain fierce.

In Amsterdam amid a strong push for more renewables and storage, analysts warn of supply risks after 2030. Here, the TSO’s 2030 plans for national grid upgrades and storage are crucial to avoid local curtailment or constrained connections which are a risk factor for data centres.

The situation is not so different in Frankfurt, Berlin and Munich where grid access is a bottleneck for data centre developers. In some German cities, grid connection requests are thought to exceed existing capacity by several gigawatts. Frankfurt’s “first come, first served” model is being reconsidered.

According to a Federal Network Agency report (2025), Germany faces potential stress points by 2030 under slower build-out scenarios with worst-case modelling suggesting tens of GW of additional capacity might be required to avoid disruption.

Berlin and Munich will benefit from national decarbonisation but may face local reinforcement needs and must rely on demand-side flexibility and storage to avoid local bottlenecks. Here as well as increased electricity demand, moves to renewables are creating the need for controllable capacity through on site generation, storage, demand response and dispatchable power.

Barcelona’s metropolitan demand growth requires distribution upgrades and grid reinforcements for large load demands of data centres and industry. The city has planned connection and grid enhancement projects through to 2030 but these require large transmission or substation works which take multiple years to build.

Even in cities such as Paris and Lyon which benefit from the firm base of France’s nuclear capacity, grid reinforcement can take years as the regional expansion of renewables grows rapidly.

Europe’s power industry is undergoing its greatest transformation since the world electrified in the late 19th and early 20th centuries, however it’s dependent on historic national energy strategies, grid reform, connection processes and decarbonisation – all moving at different speeds.

What “clean-fuel microgrids” mean in the context of AI Edge data centres

Data centres require certainty, energy security and speed. This has prompted developers to seek alternatives to traditional engagement with power companies. The main change is the deployment of dispatchable power capacity in the form of flexible, on-site, low carbon generation and energy storage.

It is here AVK’s expertise in design, implementation and maintenance ensures long term flexible power tightly coupled to the day-to-day and second-to-second operational realities of Edge AI data centres. A clean-fuel microgrid is a primary local system that can operate either connected to or independent from the national grid, using lower-carbon sources such as HVO and battery energy storage to provide backup.

For decades, grid independent operation has been the norm for industries such as mining where the remoteness of sites made grid connection impossible or un-economical.

But now for the first time, microgrids are providing urban-based data centre developers with options of full grid independence in parallel to traditional grids without sacrificing resilience, security, reliability, responsiveness, and scalability. Developers can also choose to couple with local grids to provide peak shaving and load shedding support and open up revenue opportunities for demand response power provision.

How microgrids help with power and planning consent for AI data centres

As well as securing power, developers must cross through complex planning systems. There is growing evidence that planning officers, who mostly operate at a local level, are looking more favourably on zoning and permitting where developers can avoid straining local grids through microgrid dispatchable power generation and storage.

Planning statements should explain how the microgrid will contribute to local energy autonomy for public benefit. In sustainability terms, microgrid components such as HVO-fuelled generators improve scores on Environmental Impact Assessments, and lifecycle CO2 savings versus grid power, can align with net-zero policies. 

Sustainability: AI Edge data centre heat and cooling re-use opportunities

One advantage of AI Edge data centres is the realisation of district heating and cooling benefits where higher water temperatures combined with shorter distances make district heating systems viable. Cooling of densely packed GPUs will result in water temperatures in excess of 25°C meaning rising rack densities to 100s of kWs. The need for direct-to-chip liquid cooling will generate medium grade heat which can be transferred efficiently to neighbouring commercial, public, industrial and residential buildings.

Microgrids can integrate combined heat and power (CHP) or heat pumps that supply local district heating, turning a data centre from an energy consumer into an energy partner. Metropolitan planning authorities in Amsterdam, London and elsewhere will require or reward heat reuse from data centres. The upcoming EU Energy Efficiency Directive will require data centres with total rated energy input >1 MW to utilise their waste heat or waste heat recovery applications unless not technically or economically feasible.

Microgrids and modularity: Alignment of Edge data centre and microgrid designs

Both AI Edge data centres and modern microgrids are built around modularity and scalability.

As a new data centre category, AI Edge data centres have seen the founding of companies such as Metrobloks, a data centre developer intent on building modular AI Edge data centres in cities across Europe. There are also a growing number of specialist companies who design and factory build prefabricated modular data centre components. Such modularity makes them naturally compatible with microgrids with rapid deployment, control, lifecycle economics, scalable capacity, and footprint efficiency being just some of the benefits.

Microgrids and renewables

AI Edge data centres powered through the integration of microgrids with renewable power generation represent a major opportunity for developers and cities to realise the benefits of proximity-based inference AI services operating on a decarbonised energy system.

Data centre dispatchable capacity: A major opportunity for Europe’s energy transition – An extract

In our ground-breaking research paper in partnership with Wärtsilä, we explored how data centre microgrids can transform Europe’s energy system with reliable power, lower costs and lower emissions. Below is an extract showing the four AI data power solution modelling configurations.

Microgrid models

In this data centre-focused scenario, we evaluate four possible power solution configurations, in terms of cost effectiveness and reliability, to assess the most viable options available to an 80 MW AI data centre located in Northern Europe operating under real-world conditions. For all models, the capacity of each technology is sized to meet the full potential power demand while minimising total system costs.

The options considered in the model include the following configurations and rationales:

  • Model 1 Renewables + batteries

    In this option only renewable energy, comprising onshore wind and solar photovoltaic solutions, together with battery energy storage systems (BESS) are deployed.

  • Model 2 Combined cycle gas turbines + batteries

    In this option, only combined cycle gas turbines (CCGTs) and battery energy storage systems (BESS) are deployed. The deployment of renewable energy is not considered, reflecting scenarios where constraints such as limited land availability prevent the installation of renewable capacity.

  • Model 3 Internal combustion engines + batteries

    This includes build options for medium-speed gas internal combustion engines (ICE) and battery energy storage (BESS). As in Model 2, the deployment of renewable energy is not considered.

  • Model 4 Renewables + internal combustion engines + batteries

    In this model, the internal combustion engine (ICE) and battery energy storage (BESS) are sized to align with those in Model 3. However, unlike Model 3, the deployment of renewable energy is permitted. The level of renewable capacity is determined based on system cost minimisation.

The findings

The operational emissions of Model 2 are comparable to those of natural gas power generation, which is 450 g/kWh. With the integration of renewables, as demonstrated in Model 4, carbon intensity can be reduced by over 60%, reaching approximately 180 g/kWh, or 200 g/kWh including embodied emissions. As the indicative line shows, this is lower than the 2024 grid generation emission levels of Germany, Ireland, the Netherlands or the UK.

Once the internal combustion engine (ICE) power plant is connected to the grid and begins providing flexibility and ancillary services, emissions would decrease even further. According to the “Crossroads to Net Zero” report, 100 MW of flexible balancing capacity can support up to 1 GW of renewable energy. Assuming that the ICE power plant is then needed for only 440 hours (5%) of the year to back up renewables, the combination of renewables, ICE, and battery energy storage could reduce emissions to below 10 g/kWh. This is without the use of sustainable fuels, which would reduce it further.

Analysis: Modelling impacts for data centre microgrids

Clearly, balancing power plants with dispatchable power – particularly medium speed engine-based solutions – offers the most cost-effective approach for powering AI data centres in off-grid settings. These systems provide high reliability and operational flexibility.

Read our full research paper here.

In conclusion

AI has been described as a global economic arms race. It has sparked the development of a new category of AI Edge data centres which must find many 10s of MWs of power in already strained metro power markets.

For city-based AI Edge data centres to become a reality, they will require microgrids powered through technologies such as internal combustion engines, batteries and renewables running in parallel to and in support of traditional grids as they are reinforced and upgraded. Microgrids and AI Edge data centres share many design and build characteristics such as fast modular deployment, and sustainable and flexible operation (low carbon power, heat reuse).

For cities, they turn AI Edge data centres from energy liabilities into energy partners, accelerating the path to resilient, low-carbon urban power systems. For Edge AI data centre operators, clean-fuel microgrids provide local, dispatchable power systems integrating HVO-fuelled generation, batteries, renewables, and CHP – meaning developers have a way to bypass grid bottlenecks by quickly securing reliable capacity.