Developers are looking to secure locations for global technology companies serving AI and machine learning – the centres are increasingly a critical part of the digital-economy infrastructure.

AI is changing how technology is used in work, leisure and almost every aspect of life, as organisations and individuals access systems that enable computers to think like humans. Machine learning, a type of AI, uses algorithms to gather and store information from data, helping processors identify patterns and make decisions.

Hyperscaler demand

Property entrepreneurs are searching for Scottish locations to build campus-style projects on between 30 and 120 acres to accommodate up to 1.2 million sq ft of data centre space. These sites can require upwards of 100 megawatts per 30 acres of land or more, depending upon intensity of use.

Large amounts of energy are needed to train and run AI models as well as to cool the data centres that house these models. The largest power requirement we have seen is just over 500 MW.

The ‘hyperscalers’ driving AI development such as Google, Amazon, Microsoft and Meta – require guaranteed electricity, via a so-called  ‘firm connection’. For continuity, sites must be connected to the grid and the availability of  renewable energy sources such as wind can be a benefit to the operators.  

In a process of grid reform aimed at modernising Britain’s electricity supply, the window for granting applications for new connections is closing, adding pressure to demand for sites.

The Scottish Government is keen to get behind green datacentres and AI infrastructure, though could do more to catch up, while Scottish Enterprise sees an opportunity to repurpose a number of sites.

There is competition among developers to gain representation by way of option agreements to enable them to negotiate separate occupational arrangements with hyperscalers and other users.

What developers are looking for

Land values for data centre development are important though not critical drivers. The offers we have seen range from reasonable to very low, albeit based on multiples of standard agricultural values. Offers are likely to be attractive some farmers who are approached., but this must be balanced with the potential success of planning applications and the availability of quality employment.

Developers look for relatively flat land, capable of achieving planning permission without delay due to environmental or similar considerations, ideally located in the Central Belt of Scotland or in reach of major population centres, to ensure that they can attract the right level of employment.

Each data centre could employ up to 200 to over 500  and there are long-term requirements for education and skills development. Employment grades are generally high quality, with salaries approximately 1.2 times the average equivalent.

The total development cost of a data centre is significantly higher than for say a traditional industrial of business use.

The potential for using excess heat generated as a ‘community benefit’ or alternative uses to nearby population centres could aid planning applications.

Future-minded developers will be examining technology trends closely. Currently we understand that under 20% of data is held ‘in the cloud’, that is, in data centres, leaving significant opportunities for large-scale developments, assuming coming generations adopt AI and machine learning technologies.