By Amara Osei

Voters rejected proposed data centres in 44 of 45 local referendums held across the United States during the recent midterm elections. Paul Ormerod, Honorary Professor at the Alliance Business School at the University of Manchester, argues in a new commentary that the AI industry must buy community consent or face a modern Luddite movement.

The economic dimension of the midterms drew less attention than the political horse-race, but Ormerod calls it "important". Democrat candidates campaigned with particular hostility towards both data centres and AI in general. He attributes this to the party's base: liberal arts graduates, a group whose lifestyles and employment are being undermined by the technology.

What happened on the ground?

The US ballots were not the only flashpoint. A coordinated campaign against data centres took place across Europe last weekend. Most protests passed peacefully, but in Amsterdam demonstrators inflicted substantial damage on a Microsoft data centre construction site.

The easy response, Ormerod writes, is to label the voters and protesters as Luddites. The original Luddite movement peaked nationally in Britain in the 1810s, protesting against machinery that slashed the wages of previously high-skilled and, for the time, high-paid workers.

The parallels carry weight. Machines were smashed; the British government made machine-breaking an offence punishable by transportation or death. Despite the penalties, social unrest persisted into the 1840s. The Luddites ultimately lost. The disruption was massive, and many skilled workers were driven into acute poverty, but the overall economic benefits proved too substantial to ignore.

Ormerod's verdict on the current cycle is blunt: the same outcome will eventually occur with AI. He anchors the argument in geopolitics. Countries such as China and India, he writes, are hardly going to pause development simply because members of the middle class in the West are being made downwardly mobile.

Can the transition be managed better?

The 1810s offer a warning, not a template. Ormerod contends that governments and companies "ought to be able to manage the transition better" than during the first Industrial Revolution two centuries ago.

His evidence base is a paper published earlier this year by Daniel Goetz and colleagues at the Brookings Institution in the United States. The paper does not offer a complete solution, Ormerod notes, but it supplies concrete pointers on data centres specifically.

The authors start from an uncomfortable admission: a data centre as such does little for the local economy. It generates some construction jobs, but few permanent positions inside the facility itself. Those permanent roles are well paid, yet in most other respects a data centre resembles a logistics warehouse. It does not, in the Brookings assessment, contribute to transforming the future prosperity of the surrounding area.

A data centre may of course attract additional high-tech companies to locate nearby. That potential clustering effect, however, is speculative. The Brookings report offers more immediate and productive guidelines, drawn from deals already completed or in active discussion with local and state authorities in the United States.

What does the playbook look like?

Ormerod highlights an established precedent: direct payments from resource revenues. Oil- and gas-rich states already channel income through state-operated entities such as the Alaska Permanent Fund and the North Dakota Legacy Fund. More recently, New Mexico created a $7bn fund for universal childcare, financed by endowment revenues from the oil and gas industry.

Similar principles, he argues, can be applied to AI developments. His list of options:

• Direct revenue-sharing funds modelled on the Alaska and North Dakota vehicles, with New Mexico's $7bn childcare fund as the reference case
• Local agendas for enhancing high-tech talent initiatives at local institutions and schools
• Securing abundant computing resources for local academic work from an incoming hyper-scaler (a large cloud operator such as Microsoft or Amazon)
• Partnerships in other forms of R&D at local universities, start-up companies, or entrepreneurial energy companies

These are early days, Ormerod concedes. But in America, at least, companies in the AI sector are beginning to engage in meaningful ways with local communities to generate outcomes which benefit everyone.

The alternative, the 1810s suggest, is two decades of unrest before the technology wins anyway.

Paul Ormerod is an Honorary Professor at the Alliance Business School at the University of Manchester.