Public opposition to data center expansion is stalling an estimated $119 billion in AI infrastructure investments across the U.S. and Europe, forcing a reckoning over who bears the cost of scaling compute capacity for artificial intelligence.
Research from STL Partners found that approximately $42 billion in European data center investments have experienced delays or cancellations due to local opposition. The U.S. figure stands at roughly $77 billion in affected projects.
Europe is facing the heaviest friction. More than 70 data center projects were rejected or restricted between January and April, exceeding the total number of such actions in all of 2025, according to the European Data Center Monitor. Concerns center on water depletion, grid strain, power costs, and land use.
Scotland paused planning approvals for new hyperscale data centers after observing Ireland's experience: data center power demand triggered a national moratorium. Denmark, which had marketed itself to investors on cheap land and renewable energy, enacted an emergency law that can deprioritize data centers on its electrical grid. Spain proposed rules requiring new facilities to source 80 percent of electricity from renewables. U.K. projects have stalled following community pushback.
Olivier Darmouni, an energy transition specialist at HEC Paris, said the backlash reflects a fundamental economic mismatch: "The gains from AI are very diffused," while data centers concentrate resource consumption locally. He noted that Europe's population density and dominance of U.S. operators likely intensify resistance there.
The friction extends to Asia. South Korea's government designated AI data centers as a priority investment alongside semiconductors in June, yet opposition is mounting near residential areas and local governments are drafting tighter restrictions. There is no consensus on permanent job creation or how to calculate economic return per megawatt of capacity, muddying investment justification at the local level.
The economics matter. AI labs need massive compute capacity overseas to diversify from U.S. power grids and reduce latency to Asian markets. But if Europe and Asia block growth, U.S. operators face a constrained playbook: either accept higher electricity costs and denser grid competition at home, or delay international expansion. Neither solves the core problem: demand for AI compute is outpacing grid capacity worldwide, and someone has to pay for the infrastructure.
