Data Centre Refusal Highlights Strategic Challenges. Again…
The Labour Manifesto commits to an industrial strategy that “supports the development of the Artificial Intelligence sector and removes planning barriers to new datacentres“. A recent decision to refuse permission in the Slough ‘Availability Zone’ shows the limits of a Hyper-local planning system in dealing with Hyper-scale needs.
Back to the Future
The Conservative Government refused permission for a hyperscale[1] data centre scheme on a former landfill site in the London Green Belt in October 2023 (on grounds of harm to Green Belt, harm to the character and appearance of the area and the availability of alternative sites). Our previous analysis dealt with the context of Availability Zones (AZs), demand / need issues and some oddities of the decision.
A slimmed-down version of the West London Technology Park scheme has now been rejected locally. The decision illustrates why a strategic approach is needed to planning.
Smaller is No Better
The revised scheme was reduced by 66% (to 72,000 sqm GEA across two buildings). Notwithstanding the wider strategic significance of the project and the issues, the authority dealt with it through a single officer’s delegated decision.
The decision report is commendably thorough in setting out the competing issues, before dealing – in a single sentence – with the planning balance: green belt and other harm not outweighing the various factors given significant weight. These include:
- The ‘need’ for data centres in the Slough Availability Zone – recognising that “there are a limited number of sites that meet the minimum site requirements in the Slough Availability Zone to accommodate the proposed development, and that even if proposals for data centres on other sites were to come forward, there would still be the need for other sites to be found to seek to meet the identified need in the Slough Availability Zone“.
- The fact that national policy requires decision to “help create the conditions in which businesses can invest, expand and adapt, and that significant weight should be placed on the need to support economic growth and productivity, taking into account both local business needs and wider opportunities for development“.
- The loss of economic and social benefits and potential loss of investment opportunities “to other locations, including other countries“.
In fairness, there were also some locally-significant landscape and amenity-related issues in play.
Like the previous SoS decision, the report notes that the applicant had not assessed the contribution of sites in other London AZs. The Council’s position at the previous Inquiry (accepted by the SoS) was that while there is locational demand, the ‘need’ (i.e. at a London, and UK, level) could be met on sites in AZs outside the Green Belt. Although identified as a relevant factor in the latest refusal, it is not clear how this was applied in terms of the level of lost opportunity ‘harm’ to the London and wider national data economy that was considered likely in practice, if any.
Demand to Need – Not Waving But Drowning
It is tempting to think of ‘need’ as demand the planning system recognises as important to meet in the public interest[2]. National policy after all requires plans to meet needs through strategic policies (including site allocation) unless there are compelling reasons not to.
Enhancing digital connectivity was made a key Government ‘Mission’ in 2022[3]. The annual GVA of each datacentre is estimated as £291-£320m[4]. There is, however, no national policy for data centres or any Government guidance on where they should be located.
The market has instead identified areas of intense demand or optimal potential supply for hyperscale data, Hyperscale data centres need to be developed in clusters with access to high-speed connections and a resilient power network. These Availability Zones are therefore defined by the market according to ‘power, position, and ping’. The Slough AZ itself lies close to digital connections running from London to the south west and across the Atlantic. The previous SoS refusal carefully recognised the significant need for short-term hyperscale data centre capacity in both the specific Slough AZ and wider UK. This is a far broader geography than offered by the patchwork of 262 District and Unitary authorities or the 11 Combined/ Mayoral Authority areas [5].
Mind the Gap
The fact that the shrunk West London Technology Park scheme did not make the grade is unremarkable – Green Belt land is after all serving its own long term, strategic planning purpose (primarily preventing unrelenting, unplanned, urban sprawl). There are also big data centres weaving their way through local planning regimes[6].
But planning means making choices about the future, rather than decisions about the present. How much of the demand needs to be met? Is it long or short term (and when will we (re)check)? Where is the bite point – where the downsides of not meeting the need outweigh the benefits of doing nothing? What are the best places to do it? Where is off limits? Is there a sequence? These are larger than local questions.
The striking thing is that that no-one seems to have had any answer to these strategic questions in making the decisions on either version of the project. Nor any understanding of how digital connectivity ambitions will manifest in physical development at the relevant scale.
Strategy Matters – Where to Play and How to Win?
This gap reflects the fact that there is no spatial framework for these kinds of decisions in England. Nor, since the bonfire of strategic planning on 21 December 2010, has there been any strategic tier for making policy to adopt one[7].
Taking decisions on individual appeal schemes at the highest level is not strategic. Nor is it ‘planning’ – it is just a more centralised, slower, way of reacting to big ticket issues as they arise.
Saying what you will and will not do and where at the higher level is strategic – explaining why and where and how investment should take place. There is then a choice about the process and a need to be wary of sticking anything that is big into a box called NSIPs. Once the higher strategy is set, local decisions can deliver change and focus on the detail without having to get lost in the cloud.
[1] ‘Hyperscaling’ means accommodating distributed computing requirements for ‘mission/business-critical’ functions in a way that can be rapidly scaled up.
[2] Bearing in mind that the Court of Appeal’s comments to the contrary in R (Cherkley Campaign Ltd) v Mole Valley District Council [2014] EWCA Civ 567 @25-35 relate to the interpretation of specific policy wording
[3] Levelling Up White Paper (2022,)
[4] https://www.thestack.technology/content/files/prod/public/32065b6b-67f7-47d8-ad9afaf91498c968/techukuk-data-centre-market-overview-2020.pdf
[5] With the 11 Combined Authority areas providing a better strategic fit and the 21 County areas a less bad second place
[6] Glenigan Data Centre Development Pipeline Continues To Roll Out (10 June 2024)
[7] None of the Combined Authorities or the Mayor of London operate spatial strategies at a scale relevant to AZ geographies. Some of them only just do the job for housing market areas, arguably not including Greater London.
