The Invisible Trillion · How a Neighborhood Dies

The Alternative

You can’t go back. Refrigeration, air conditioning, the internet, AI — none of it is being given up, and the infrastructure underneath it will keep getting built.

But it doesn’t have to be built next to people. A facility the size of a small town does not need to sit beside anyone’s house. It can go twenty miles out, with its own access road and substation and water, where the land is cheaper, the legal exposure is lower, and the impact on a community is close to nothing. The workforce can commute; the fiber can be run. Nothing about a data center’s actual requirements demands a residential neighbor.

The reason so many land near homes is the externality logic again: the remote site costs the developer more, while the site near an existing exurb is cheaper precisely because some of its costs fall on the residents instead. The highway parallel holds here too. The interstates did not have to run through East Tremont and Black Bottom and Overtown and Rondo and Tremé. Other routes existed — a little more expensive, a little more complex — and the choice to take the cheaper one at the residents’ expense is a choice we now broadly agree was wrong. Those neighborhoods weren’t lost to physics.

Sixty years on, that recognition is slowly producing the will to undo a little of the damage. The Embarcadero Freeway in San Francisco came down in 1991, the Park East in Milwaukee in 2003, the eastern leg of Rochester’s Inner Loop was filled back in to street level beginning in 2014. Detroit now has federal money to turn I-375 — the mile of sunken freeway that finished off Black Bottom — back into a surface boulevard, with construction staged for the late 2020s; New Orleans is pursuing the same for the Claiborne Expressway that gutted Tremé. The work is slow, expensive, and incomplete, and the neighborhoods that existed before the concrete are not coming back. But the wounds are at least being treated.

The data center industry is making the same kind of decision right now. Some of it will be undone by the lawsuits in progress and the bills being introduced; some won’t; and the ones that stick will shape those places for decades. What would change the outcome is knowable and not exotic — it’s the same category of change that eventually slowed urban highway construction in the 1970s.

Site the facilities away from homes. Put the grid-expansion costs on the parties that cause them. Measure low-frequency noise honestly. Enforce the air rules already on the books. Protect local zoning from pre-emption. Give the residents on the wrong side of a jurisdictional line some standing in a decision that lands on them. None of it is technically hard. All of it is politically hard, for one reason: each piece asks the buildout to absorb a cost it currently passes along, and the tools in the last chapter are exactly what have kept those costs passable.

That’s the heart of it, and it’s where I’d rather leave the question than tie a bow on it. The current pattern is not a law of nature. It’s what a particular set of rules and incentives produces, and a different set would produce a different pattern — more of the cost carried by the people who choose the buildout, less by the people who only live near it.

Whether the rules change isn’t something a book can predict, and I won’t pretend otherwise. It’s one of the genuinely open questions the buildout leaves on the table — and how it resolves will say something about the era: whether a society that could build a technology this powerful could also share out its costs in a way its own citizens would recognize as fair.

All 19 chapters of The Invisible Trillion · Library · All courses