Building a data center in the US used to be the obvious play for Big Tech. That calculus is changing fast, and the numbers explain why.
Around 75 projects valued at more than $130B were either blocked or delayed domestically in Q1 2026 alone, largely due to local opposition. A March 2026 Gallup poll found that 70% of Americans oppose new data center construction in their communities. Microsoft, Google, Amazon, and Meta are now doing what any rational capital allocator would do: looking elsewhere.
The cost problem is getting worse
The sticker shock starts with construction. Average new US data center projects now run about $1B, with all-in greenfield development reaching roughly $17.6M per megawatt of capacity.
The raw materials driving those costs aren’t cooperating either. Copper wire prices have climbed 17.9% year-over-year through mid-2026, while steel is up 22.5% over the same period.
Then there’s the cost of waiting. A one-year delay on a typical 100 MW US data center can destroy $500M to $550M in lifecycle value.
Where the money is going instead
India has emerged as one of the most compelling alternatives. Build costs there run $6M to $7M per megawatt, compared to $10M to $14M in the US and Europe.
The power economics are equally attractive. Effective electricity costs in India can drop as low as 6 to 7 US cents per kilowatt-hour, driven by renewable energy availability and open-access power markets.
The Nordic countries and Spain have also drawn attention from hyperscalers. The Nordics offer naturally cool climates, which means less energy spent on cooling. Spain brings a combination of renewable energy infrastructure and relatively streamlined permitting compared to major US markets.
The collective spending tells the story of how urgent this buildout has become. Hyperscalers are projected to pour between $660B and $690B into data centers and AI infrastructure in 2026.
What’s driving the domestic resistance
The 70% opposition figure from Gallup reflects a remarkable shift in public sentiment. Data centers were once sold as clean, high-tech job creators. The reality is that they employ relatively few people once operational while consuming resources at an industrial scale.
Some hyperscalers have pledged to cover incremental power costs that their facilities impose on local ratepayers. But enforcement of those commitments remains uncertain, particularly as utility rates continue climbing and political dynamics shift.
Market implications for tech and beyond
There’s a geopolitical dimension worth watching too. Moving critical AI infrastructure abroad introduces new considerations around data sovereignty, regulatory compliance across multiple jurisdictions, and exposure to foreign policy risks.
For the US specifically, the trend represents a kind of ironic feedback loop. The country that produced the world’s dominant cloud providers and AI companies is making it progressively harder for those companies to build at home. The demand isn’t going away. It’s just finding a new address.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.

43 minutes ago
13









English (US) ·