Executive summary
- The data centre industry often cites job creation and positive economic impact as a way to get local communities on board.
- Newer research reveals the number of jobs created by a data centre development could be only a quarter of that used in the argument.
- Can we realistically only consider those who will work in the data centre as a ‘job’ or are we right to allow figures to include ‘indirect’ and ‘induced’ too?
There is plenty for data-centre opponents to object to. The electricity. The water. The land. The grid connections. The enormous buildings appearing on the edge of towns and cities.
But there has always been one particularly useful counterargument: It’ll create jobs.
It is a simple argument, and a fairly compelling one, particularly when those jobs are measured in the thousands.
Except now there is a rather awkward question to ask.
What if some of those jobs aren’t quite the jobs we think they are?
Recent research from Verdant – provocatively entitled ‘The mirage of data centres’ has taken aim at the industry’s employment claims, and its findings are just as provocative.
Verdant estimates that existing UK data centres support around 4,400 direct jobs, rather than the 24,300 it says are claimed by industry, and puts the number of direct operational jobs associated with planned developments at around 10,400.
That compares with the 40,200 additional operational jobs techUK projects will ‘supercharge’ the sector by 2035.
It’s quite a difference. And, if Verdant is right, the industry’s favourite answer to its critics may have just become… redundant.
But before we get too carried away, there is a problem…
What exactly is a job?
A proposed £10bn data-centre campus in Northumberland estimated creating 4,400 jobs. Whatever you think about the development itself, 4,400 jobs sounds like a fairly substantial contribution to a local economy.
Within that, according to the Commons Library, around 1,200 are construction jobs, a further 2,700 are indirect and induced jobs, and up to 400 are permanent on-site roles.
Surely nobody could seriously argue that construction jobs don’t count? A £10bn project needs an enormous workforce to build it, from electricians and engineers to project managers and specialist contractors, and those people are being employed as a direct result of the investment. The fact that some of those jobs will disappear when construction finishes doesn’t make them any less real.
But it does make them rather different from the 400 people who are expected to work at the facility once it’s operational.
And then we get to the 2,700 indirect and induced jobs, which are arguably where this whole debate gets particularly messy.
An indirect job created because a data centre needs suppliers, contractors or professional services is still a job. An induced job generated by the spending of people employed by the project is still economic activity. It would be rather convenient for anyone arguing against data centres to simply pretend those jobs don’t exist.
But if you’re a resident being told that a new data centre is going to create 4,400 jobs, it’s also fairly reasonable to assume that you’re being told 4,400 people will eventually be working there.
They won’t.
And that distinction matters.
Then there are the really big numbers
techUK’s 2024 research projected 40,200 additional operational jobs by 2035, alongside 18,200 construction jobs between 2025 and 2035.
Verdant’s much more recent analysis puts the number of direct operational jobs associated with the planned UK data-centre pipeline at around 10,400.
That’s a fairly sizeable gap.
The two organisations are using different methodologies and making different assumptions. And importantly, techUK’s 40,200 is a projection of additional operational jobs rather than a claim that 40,200 people will be sitting inside a particular collection of new data centres.
This is no longer a case of who’s right and who’s wrong – it’s what are we actually counting?
The jobs-per-megawatt problem
Verdant’s research argues that the industry’s employment estimates rely too heavily on jobs-per-megawatt assumptions. And not the comparatively small number of people actually needed to operate increasingly large and automated facilities. A 500MW data centre doesn’t need 500 people sitting next to the servers. That’s true.
But there’s an equally obvious counterargument. The economic value of a data centre doesn’t stop at the front door. It creates construction work, buys equipment, requires engineers and contractors, uses professional services and supports suppliers. The people employed by the project then spend their wages in the wider economy. Those effects are real.
Show us the jobs
If developers want to put employment at the heart of the case for building a new data centre, we need to start breaking down that number.
No more “this project will create 4,000 jobs”, but 400 permanent operational jobs, 1,200 construction jobs and 2,700 indirect and induced jobs.
Then tell us how many apprenticeships there will be, how many people are expected to be recruited locally, and how many of those jobs are likely to remain once the construction teams have packed up and gone home.
And perhaps there’s one more number we should be asking for: how many of those jobs are expected to exist five years after the facility opens?
None of this is to suggest that indirect employment shouldn’t count. Of course it should. Nor does Verdant’s analysis necessarily mean the industry’s wider economic case falls apart. It doesn’t.
The mistake we seem to be making is that we’re using the same phrase to describe permanent employment inside a facility, several years of construction work, jobs supported through the supply chain and wider economic activity.
So now is it our job to make a better-sounding argument?
