Capitalist Pressures Limit Worker Choices Over AI Data Center Construction to 'Join the Workforce or Die'

Workers building the AI boom face what critics describe as an increasingly stark choice under capitalist pressures: take the work created by the data-centre build-out or risk being left behind economically.

The argument forms part of a broader critique of how overcapacity, slowing growth and competition can accelerate automation and squeeze workers. It comes as local opposition to AI infrastructure intensifies across the US and communities grapple with the economic and environmental costs of the rapid build-out.

To recall, the past year has seen a sharp surge in grassroots resistance to data centre construction, with at least 25 projects cancelled following local opposition in 2025, according to Heatmap News. Heatmap Pro subsequently recorded at least 20 additional projects cancelled following local pushback during the first quarter of 2026, with later reporting putting the year's total above 50.

Activists cite water and energy use, noise and concerns over electricity costs and local development as reasons for pushing back. In parallel, economists and labour analysts have warned that AI could disrupt employment across industries, prompting renewed debate over policies such as universal basic income as a cushion against the shock.

Worker Dilemmas in AI Data Centre Construction

The core argument advanced by critics is that workers are not simply choosing well-paid jobs on data-centre sites; they are being squeezed by a global economy in which growth rates are slowing and overcapacity is pushing firms to cut costs, automate and consolidate.

Aaron Benanav, in his book Automation and the Future of Work, argues that persistent overcapacity and slowing economic growth have weakened demand for labour across advanced economies. Under his analysis, intensifying competition pushes companies towards cost-cutting and labour-saving technologies rather than automation independently creating mass unemployment.

In this reading, the AI boom looks less like a technological revolution that independently causes job losses and more like another stage in businesses' longstanding pursuit of greater productivity and lower labour costs. Automation is not new; Benanav argues that the economic conditions driving it are crucial to understanding its effects on workers.

On the ground, the picture is more complicated. Construction workers can see an uptick in employment and earnings while projects are being built, and some workers can secure long-term roles managing facilities.

But construction ends, permanent employment at individual data centres can be considerably smaller than the workforce required to build them, and the wider economic benefits depend on factors including investment, taxation and local agreements.

Meta's America's Workforce Academy, a $115 million (£86 million) training programme promising a 'fast track' toconstruction jobs building AI data centres, has been described by left-wing magazine Current Affairs as an 'anti-union' scheme that could weaken organised labour.

Meta presents the programme differently, describing it as a cost-free pathway into skilled trades that provides training, recognised credentials and a job offer through contractors working on its projects. One Reddit user quoted by Current Affairs summed up the criticism, 'Learn coding, teach ai, get replaced by AI, go learn electrical work to build ai's building, get laid off after building complete.'

AI Data Centre Construction and the Limits of Labour Power

US union membership remains historically low, although the extent of organised labour's influence varies considerably by industry and state. Even where workers have leverage, critics argue that broader economic pressure can make it difficult to turn down well-paid work.

German philosopher Søren Mau uses the term 'mute compulsion' to describe how economic dependence can constrain workers without direct coercion. Applied to the AI infrastructure boom, the argument is that workers may have considerably less freedom to reject jobs building technologies they fear could eventually displace other forms of employment.

Local opposition has been effective in some places. Heatmap's research found that roughly 40 per cent of proposed data centres facing sustained local opposition in 2025 were eventually cancelled, although outcomes vary significantly between states.

The number of local restrictions has also increased as municipalities consider moratoriums, zoning restrictions and other measures aimed at controlling data-centre development. Heatmap reported in January that more than 60 towns, cities and counties had enacted some form of moratorium or restrictive ordinance, with researchers expecting the figure to continue rising.

Yet measures intended to make AI infrastructure more expensive or difficult to construct have not stopped the wider expansion. Technology companies continue to announce major investments as electricity demand from data centres rises.

Critics have also questioned whether workers themselves should bear responsibility for constructing the physical infrastructure behind AI. Under the economic argument advanced by Benanav and others, individual workers have limited power to refuse available employment when their livelihoods depend on selling their labour.

For some campaigners, the answer therefore lies in collective rather than individual action. Environmental organisations including Greenpeace USA, Friends of the Earth and Food & Water Watch have called for a national moratorium on new data-centre construction, citing economic, environmental, climate and water concerns.

At the worker level, supporters of stronger organised labour argue that unions could give employees greater influence over how automation is introduced and how its economic benefits are distributed.

As citizens, people can also organise through local activist and environmental groups that apply pressure to proposed developments. Heatmap found that cancellations linked to sustained local opposition quadrupled from six projects in 2024 to 25 in 2025, suggesting that organised community resistance can materially affect whether individual projects proceed.

The larger debate remains unresolved. AI infrastructure is creating skilled-trade jobs and attracting enormous investment even as the technology it supports raises fears about employment displacement, energy demand and environmental costs.

Whether the workers building that infrastructure ultimately benefit from the boom will depend not only on AI itself, but on wages, labour protections and how the economic gains from automation are distributed.