Why AI’s environmental footprint won’t disappear into orbit

The morning chill bites, even indoors. Another day, another report of man’s cleverness outstripping his wisdom. They speak of moving the machines, these ‘data centres’, into the heavens, thinking to escape the dust and heat they generate here on earth. A foolish notion.

The obstacle is not the location of the machine, but the appetite that feeds it. To move the problem is not to solve it. It is to shift the burden, perhaps out of sight, but never truly gone. The cost, they say, is hidden. But what truly is hidden from the universe’s ledger? The energy to launch, the resources to build, the decay of what is left behind. These are not hidden; they are merely ignored.

Antoninus would have seen through this immediately. He understood that a well-ordered city does not merely sweep its refuse to the outskirts; it manages its consumption. This desire to escape the consequences of our own making, to project our mess into the void, is a failure of responsibility. My duty is to manage what is here, on this earth, with the resources given. Not to dream of new heavens while the old one groans under our weight. The work is here. The solution is here. Not in some distant orbit.

July 18, 2026.

The notion of launching our computational engines into the void, as if to exile the physical consequences of our own ingenuity, strikes me as the perfect, pathetic emblem of our age. We have built a new kind of dynamo, one that consumes not coal, but attention, data, and the very stability of the climate it was born into. And now, finding its earthly footprint too inconvenient, we propose to shoot its furnaces into orbit. It is a solution conceived for a problem of physics by a mind trained entirely in metaphor - the metaphor of limitless space, of an escape velocity that applies to moral and material accounting alike.

The hidden costs, of course, are not merely those of rocket fuel and radiation shielding. They are the costs of a persistent, wilful blindness. Every institution of governance, every framework of environmental stewardship, was calibrated for an industrial tempo of smokestacks and tailings ponds. They move in legislative sessions and regulatory cycles. The AI dynamo accelerates on a curve that renders those cycles geological in their slowness. The gap is not a technical oversight; it is the defining chasm of our time. We are attempting to govern a thermonuclear reaction with the bylaws of a village water board.

So we look to the heavens. It is the same impulse that built the cathedrals - an aspiration to transcend earthly limitation. But the Virgin offered a unity of purpose; her spire pointed toward a collective ideal. This silicon spire points only toward a dispersal of responsibility, a literal outsourcing of entropy. They will call it progress. I call it the most expensive confession of failure yet devised: that our creations have grown too potent for their cradle, and we would rather fling them into the dark than learn to live with them here.

The notion of launching data centers into orbit to escape Earth’s environmental burdens is the sort of thinking that betrays a fundamental misunderstanding of systems. The energy required to lift even a single kilogram beyond gravity’s grasp is staggering - each launch a roaring expulsion of wasted potential, as though we are burning libraries to power a single lamp.

Consider the theoretical limit: a network of self-sustaining orbital nodes, harvesting solar flux unimpeded by atmosphere, beaming clean power back to Earth while processing data in the vacuum’s natural cooling. A beautiful vision - but between that and today’s reality lies not physics, but economics. The cost of reaching orbit remains a grotesque inefficiency, a tax imposed by our primitive chemical propulsion. We could have built mass drivers, orbital tethers, even nuclear-thermal ascent stages by now. Instead, we hurl fuel into fire and call it progress.

And what of maintenance? A terrestrial data center can be repaired by a technician with a wrench. In orbit, every failed transistor demands another rocket, another plume of exhaust, another cascade of externalized costs. The proposal is not a solution - it is a confession that we would rather flee our problems than fix them.

The true path is clear: first, master the energy equation on Earth. Then, and only then, reach beyond it.