Communities Resist Forced Datacentres Over Energy Strain

We are told that the datacentre is now the fixed price of admission to the age of artificial intelligence - that Scotland’s glens and India’s cities must yield their power lines and their water tables to it, or forfeit the wealth the technology promises, as if compute could be conjured without concrete and copper. But notice the instrument that has actually arrived this summer, in the fields of Scotland and the districts of India where communities have taken to the streets against facilities they say were forced upon them: it is not merely a building, it is a purpose-built apparatus for training and running models at a scale no laboratory, no university cluster, no prior arrangement of transistors could sustain. That is a genuine capability, and it should be named honestly before it is defended or attacked. The question worth asking is not whether the datacentre is holy or wicked but what, precisely, it does that could not be done five years ago, and what that doing costs in the specific grid, in the specific summer, in the specific village whose taps run dry when the cooling towers spin up.

Both sides in this dispute commit the same error, though from opposite ends. Big tech companies assert, as though it required no proof, that the datacentre is essential to cashing in on artificial intelligence - a phrase that smuggles in the assumption that value flows automatically from capacity, that a rack of processors is as good as a factory floor. This is the idol of the market dressed as engineering. The communities in Scotland and India, for their part, warn that these facilities threaten energy supplies and the climate - and here too the claim, however sincerely felt, is offered as settled fact rather than measured harm. A warning is not a demonstration. If the grid strain at a particular Scottish substation this summer produced brownouts, name the substation and the hour. If a district in India saw its aquifer drop by a measurable margin because a cooling system drew from it, produce the well readings. Bacon’s method does not ask which side to trust; it asks which side has done the measuring.

What makes this dispute genuinely new, rather than a rehash of every prior fight over factories and pylons, is that the instrument in question is not fixed in its hunger. Unlike a blast furnace, whose thirst for coal is more or less what it is, a datacentre’s demand for power and water is a function of architecture still being rewritten - in the chips chosen, in the cooling method installed, in the software efficiency of the models it serves. The governments now caught between the two camps, granting permits with one hand and fielding petitions with the other, are treating siting as though it were the only lever, when the lever that actually reorganises the field is the design of the instrument itself. A datacentre cooled by air in a temperate Scottish summer draws differently from one drawing from an Indian river in monsoon season; a model served efficiently draws differently from one running bloated because no one bothered to compress it. These are not settled facts of nature. They are engineering choices, and engineering choices can be tested and improved, which is more than can be said for a grievance or a press release left unexamined.

So the experiment worth running is not a referendum on whether datacentres are good or bad but a public accounting: measured draw against grid capacity, measured water use against local supply, published before construction and audited after, in Scotland, in India, and everywhere else this contest recurs this summer and beyond. Let the company that claims necessity show its efficiency curve. Let the community that claims harm show its meter readings. Whichever side refuses the measurement has already told you which of its claims is the cobweb - the inherited certainty dressed up as knowledge, unwilling to risk the test that might unmake it.

I picture the villager standing at a dry standpipe within sight of a cooling tower’s steam, waiting not for an apology but for the number that would tell her whether the steam and the dry tap are actually connected. That number is the whole argument, and neither the boardroom drafting its next capacity announcement nor the picket line outside the perimeter fence has yet produced it.