Grid and planning reforms can keep the UK and Europe competitive in the data centre race

Data centres are now a mainstream infrastructure asset and are rapidly expanding in the UK and Europe, but investors, developers and operators face structural obstacles that could hold back this growth. Demand, capacity, risk, financing structures and policy are shaping the investment, development and operation of data centres in the UK and Europe as well as opportunities into 2027.

The UK now has around 450 data centres, up from 52 in 2000, with £59 billion of investment committed for 52 new sites over the next nine years, with 50 of those expected to be delivered within five. In London, colocation take-up reached 150 megawatts (MW) in 2024, almost three times 2019 levels, as enterprises move to hybrid and multi-cloud environments.

Yet senior figures from across finance, development and real estate, in discussion at a recent Osborne Clarke roundtable event, pointed to structural barriers holding back expansion in the UK and Europe amid a growing imbalance between what the data centre market conceptually requires and what can practically be delivered on the ground. 

Growth barriers

Many data centre operators and consumers, shaped by the US blueprint, where power connections are expected within two to three years of application, are arriving in Europe and discovering that these easy wins do not exist.

FLAP-D (Frankfurt, London, Amsterdam, Paris, and Dublin) markets are saturated and will remain so well into the 2030s. The alternative markets in the Nordics and Baltics may have capacity, but developers and investors face the task of navigating entirely different consenting regimes and local requirements before capital can be secured and deployed.

In the UK, much of the focus is on pre-2030 projects, particularly in central or Greater London where power and fibre connections already exists. Nevertheless, due to a number of constraints, including the practical delivery of the power connections secured which in turn is slowing the release of funding, many of those projects are unlikely to advance before 2030.

Power and grid constraints

The UK grid is old, insufficiently resilient, requires extensive and costly upgrades, and is facing unprecedented levels of connection demand that continues to grow at pace year on year. While many viable sites have secured grid connections, these are not deliverable by the grid until  2030 and beyond. In Slough and Docklands in the UK, for example, key substation and network upgrades that will release further capacity are not deliverable until at least 2029.

The core challenge is to secure power that is reliable, affordable and sustainable enough to satisfy both customers and financiers. Going off-grid and operating without connecting to the network, is widely discussed in the industry but in practice the options are limited and are largely deployed as a temporary measure to enable funding draw down whilst a connection to the grid is awaited.

Private gas turbines require an upgraded gas supply and conflict with the sustainability commitments that many operators and their investors maintain. Heat recycling, including models such as Deep Green’s and water cooling are also becoming increasingly important. Regulatory requirements for waste heat reuse, as in Germany, are beginning to shape site selection and design, as data centres seek to integrate into existing or planned heat networks.

Small modular reactors (SMR), which have been around for some time, offer a potential long-term option in an industry that needs short-term solutions. However, some industry experts consider central European countries more likely to adopt SMRs than the UK. Some large cloud providers in the US already have test facilities connected to SMRs. As for other options, the US and China are adopting large-scale nuclear capacity at pace, fuelling concerns that the UK and Europe will be left behind.

Legislative overhaul

Electricity distribution network operators are sitting on 75 gigawatts of grid connection applications, a large proportion of which are speculative. Their ability to process, filter and deliver upgrades is severely constrained. Equally, without meaningful legislative reform, they are unlikely to be able to reinforce the grid and deliver the connections required, particularly at pace.

Proposed legislative changes have, so far, fallen short in this area. The government should learn from the extensive changes made and successes achieved in the telecoms sector. Large-scale reform provided greater rights to telecom operators and ultimately enabled the swifter roll out of digital infrastructure. The data centre industry is increasingly calling for a similar legislative overhaul.

Building at pace

Even if the power issues were to be resolved, the sector faces the question of whether it can build projects at pace.

Demand is seeing development timelines being compressed from 24 months to 18. However, a traditional build for a pre-2030 data centre could take four years. Modular construction, where components are manufactured off-site and assembled on location, is increasingly a credible option. Buy-in to modular construction heavily depends on data centre operators and consumers accepting the module approach, which constrains the delivery of their often bespoke requirements.

Punchy delivery timescales may move the dial on this. A 50 MW data centre in Norway was delivered in approximately 15 months using a fully modular approach. A life science development that would have taken four years via a traditional build programme, achieved first handover in nine months using modular methods while meeting some of the most demanding regulatory requirements in any sector.

At peak construction, a traditional build might require around 1,300 people on site. A modular equivalent requires as few as 250, which is better aligned with the available workforce. The skills shortage in this sector is acute, spanning not just construction trades but engineering more broadly. Modular does not solve that problem, but it makes it more manageable.

The contractor landscape, however, is a more serious obstacle. Major engineering, procurement and construction (EPC) contractors do not yet fully understand data centres and are unwilling to accept the risk associated with them. They will not provide the EPC wrap that banks and lenders need to underwrite facilities running to over a billion euros in capital expenditure. Developers are therefore managing supply chains directly. This can be done, but it is more complex and adds a layer of risk that investors and lenders are having to weigh carefully.

Capital and cover

There is a growing appetite for capital deployment in the European data centre market. A substantial wall of bank lending and refinancing is already in place and institutional investors are seeking longer-term positions beyond the bank market. Preferred equity and mezzanine structures have emerged to bridge the gap between construction start and stabilised operation phase; appetite spans the full capital stack.

Insurance cover for data centres differ from that for general infrastructure assets. In the US, banks are requiring full reinstatement coverage for data centres, not the "maximum foreseeable loss" basis that applies to conventional infrastructure. The cost of maintaining that cover is, in some cases, prohibitive. As insurance costs escalate with the growing concentration of hyperscale capacity, who bears that burden becomes a critical negotiating point, particularly in markets where lease structures do not easily allow for recovery of those costs.

Technology also presents risks to investors. Fund managers running 25-year infrastructure vehicles are finding it increasingly hard to apply their traditional investment approach to an asset class where rack density is shifting rapidly, the lifecycles of graphics processing units are measured in months, and end users are still establishing what business models will generate returns. Fund structures themselves may need to evolve and shorten in tenure to match the pace of a fast-moving sector. Alternatively, a model that separates the long-lived structural asset from the technology within it may emerge as the preferred approach.

Getting the deal done

The window for monetising powered land positions may be narrowing . If hyperscalers successfully deploy self-powered SMR facilities at scale, their dependence on the conventional powered land market diminishes materially.

Location could become less relevant if sites are self-powered, with a knock on effect for sites with existing grid connections that have already been assembled and consented. These could become less strategically relevant, putting pressure on current opportunities to be closed quickly. 

Osborne Clarke comment

The data centre sector remains entrepreneurial and fast-moving but is complex and fragmented. There are roughly 50 active data centre operators in Europe alone, from hyperscalers to edge and inference players, each with fundamentally different power requirements, design preferences and customer relationships. What connects them all is the same underlying infrastructure and supply chain constraints which are not keeping pace.

While the sector is exciting and can offer clear benefits, caution and preparation are critical for success. As with other asset classes, the type of end-user and the balance between standardised and bespoke design are important variables for developers to consider. The industry as a whole would benefit from collective government lobbying to drive more expansive legislative reform, increased entrepreneurship when exploring power options, greater collaboration and transparency between all players in the market and the maturing of funder knowledge  of the market to encourage more investors to be comfortable investing in the assets.

This Insight draws on the discussions and views expressed at a recent Osborne Clarke client roundtable on data centre investment and development in the UK and European market