Do London’s AI companies face the prospect of a data centre squeeze?

Serverfarm

London is in the midst of an AI boom. 2024 investment levels into London’s AI companies could outstrip the £15.2bn of capital raised in 2023. The UK tech sector boasts 171 unicorns and a market valuation of $1.1 trillion. It is the leading tech ecosystem in Europe with the majority of tech companies located in London and the south-East of the country.

London is currently Europe’s major centre for AI investment (more than Paris or New York). The city is home to an estimated 1,800 AI companies. Among the most-high profile is Wayve, a maker of AI software for autonomous vehicles, which in May 2024 raised over $1bn. Microsoft and NVIDIA took part in the investment round.

A booming AI sector means tough competition to attract investment. It also means competition to find the right data centre partner. It is a fact that for many companies seeking AI ready data centre could struggle as suitable space, power and cooling is unlikely to be available off the shelf.

Those brightest and best who secure early stage, second and third round funding could find themselves in a race to secure suitable data centre capacity.

Early stage, fast growth AI companies who are looking to scale would do well to engage with data centre operators where bespoke services through proven expertise in resource management, energy efficiency and workload management (which are key to successful model training and applied AI inference workloads) are part of the day one discussion.

It is critical that AI companies find the right data centre partner to secure suitable near term capacity for rapid accelerated growth.

Of course in any growing AI ecosystem such as that in London many companies will be cloud platform based (and as such will already be hosted in colocation data centres such as Serverfarm facilities). But others will require bespoke services, tailored sustainable power and cooling along with clear capacity expansion roadmaps to match their specific workload needs and that of their chosen IT hardware environment. According to Beauhurst, an investment consultancy, the top ten UK investments in AI companies span healthcare, fintech, lawtech, cyber, autonomous vehicle software, and AI semiconductor development.

This points to a development ecosystem where one size will not fit all. Whatever the sector served and whatever the AI workload, customers will benefit from hands on engagement to meet specific needs.

New builds versus near term capacity

Since the UK’s July 2024 General Election there have been much publicised data centre investment announcements into new builds in the UK. Not least from the new UK Government.

As part of the “Invest 2035: the UK’s Modern Industrial Strategy” The Department of Business and Trade announced $6.3bn of foreign investment from US technology firms including CoreWeave, and Servicenow. It said: “Major tech firms based in the US have committed to the UK as the place to invest in data centres, fuelling Britain’s economic growth and spurring on AI development.” This is most welcome. But in the same statement the Government acknowledges that future planned data centre developments are “subject to planning permission and that “the projects should be operational by Q4 2028.”

Welcome as they are the ‘come to market’ timelines for future data centre builds across the UK stretch out from 18mths for planning permission with another 2-3 years for construction.

The question is: Can London’s AI start ups and scale ups and their customers in transitioning businesses who are looking to use AI to drive growth afford to wait?

The AI sector is evolving rapidly. It would be ironic if it found itself hampered by a scramble for data centre capacity.

As the market awaits new builds, companies requiring more immediate capacity are faced with falling vacancy rates. The numbers tell the story. According to CBRE’s Q224 market report colocation data centre vacancy rates have fallen below 10% in London and other major European markets. UK plc and London’s fast growth AI companies (and their investors) cannot afford the risk of being caught in a data centre squeeze.