The UK is set to experience a data center boom over the next decade – at least that’s the hope of the new Labor government. Prime Minister Sir Keir Starmer is determined to “make the UK irresistibly attractive to AI companies looking to start, expand or grow their businesses,” which means removing barriers to the physical infrastructure on which private sector investment in AI relies. Planning barriers.
But the recent spate of announcements about data center projects has me baffled by the sheer number of new jobs that seem to be involved. The government announced that Blackstone is investing £10 billion in a data center park in Northumberland, which "will create more than 4,000 jobs." Starmer said the three technology companies' separate investments in artificial intelligence infrastructure were worth £14 billion and would "create 13,250 jobs in the UK". hang on. Data centers are filled with machines, not humans. For example, Microsoft says its data center buildings employ about 50 people per building. So where do these big job forecasts come from?
Let's take the example of the Northumberland project, which envisaged building 10 data centers on land originally intended for the (failed) Britishvolt battery factory. The development "is expected to require up to 40 people per data center," according to planning documents. Once all 10 centers are completed (expected to be by 2035), 400 jobs will be created, just one-tenth of the figure announced by the government.
Planning documents say "monthly construction employment peaks are likely to be around 1,200" - these are not permanent positions, but construction work on site can last up to ten years, so they are not transient jobs either. Even so, operations and construction jobs combined account for only 1,600 jobs. Where are the rest of them?
Now we enter the more ambiguous territory of economists' models. It is common practice to estimate the number of "indirect" and "induced" jobs created by large investments. These are the jobs that will be created in the supply chain and as a result of new employees spending money in local pubs, shops and so on. Planning documents for the Northumberland project say one direct job at the data center can support a further five to seven jobs in the local economy and related industries, which is "the equivalent of all 10 data center jobs being added to the local economy." 2,000 to 2,800 jobs.” The data center building is operational." At the midpoint of 2,400, we have reached the government's figure of 4,000.
It is reasonable to assume that there will be some spillover work, but the "multipliers" applied are often derived from models and may vary by industry and country. Blackstone told me the multipliers cited in the planning documents come from a PricewaterhouseCoopers study of the impact of U.S. data centers, commissioned by the Data Center Alliance, the industry’s membership association.
Henry Overman, a professor of economic geography at the London School of Economics, told me that scholars tend to find smaller multipliers when they conduct ex post facto empirical studies of what happens when new jobs arrive in an area. For example, his research into the impact of the BBC's decision to move 1,700 medium and high-skilled jobs to Salford, Greater Manchester, between 2011 and 2012 found that each BBC job created an average of 0.33 additional jobs in the creative industries. jobs, rising to 0.55 jobs will be added by 2017. He and his colleagues found no significant impact on total employment.
That’s not to say the project is bad for Northumberland. But for the government, why invite public disappointment or cynicism by bundling all these estimates into one big number when you can treat people with respect and give them all the details from the start? Another argument might go something like this: "This project will create a number of skilled jobs and ten years' worth of construction work. It may also bring wider benefits to the local area, but it's difficult to give precise figures." "Granted, it may not be dramatic or inspiring, but my sense is that people are in no mood for attention-grabbing nudges right now, anyway.
There are many good arguments for the economic importance of data centers (and some concerns about energy and water use), but "they create a lot of local jobs" is not one of them.
If Starmer wants to revive productivity growth as he says he will, he will have to be comfortable talking about the expansion of industries that don't require many workers to generate a lot of value. The Labor government needs to update its playbook for these new industries that are more capital-intensive than labour-intensive.
sarah.oconnor@ft.com