When American manufacturing was truly coming into its own in the early 20th century, workers sometimes labored twelve-hour days for seven days a week and risked life and limb working near dangerous machinery, while factories blackened the skies of American cities and poisoned waterways.
Yet, some of the same people who insist that the American economy needs to re-industrialize to return to a version of those halcyon days can’t bear the thought of the construction of industrial-scale computing facilities.
They complain that data centers are unsightly and use lots of energy and water. But what do they think major factories are like?
If today’s standards applied, we almost certainly never would have built manufacturing facilities at scale, for fear of the aesthetic consequences and worry about the collateral damages to communities, which were much more severe than anything we are talking about with data centers.
Industrial buildings look pretty in Edward Hopper cityscapes, and, say, the Ford Motor Company River Rouge complex had a certain majesty captured by photographers and painters. Otherwise, factories were generally closer — especially when clustered together — to William Blake’s dark Satanic mills than anything the average local zoning commission would want in the neighborhood.
No one ever said, “God, that smoke stack looks gorgeous — I wish I lived closer to it!”
Data centers have caused outrage because they look like big warehouses, but imagine the vapors everyone would have if they spewed soot over streets and houses, created slag heaps, divided neighborhoods with rail lines and other supporting infrastructure, and packed workers into nearby subpar housing.
AOC thought she had a trump card in the anti-data-center case when she brandished two jars of dirty water from a house near a data center (the house’s well was perhaps disturbed during construction). That’s nothing compared to the Cuyahoga River in Cleveland repeatedly catching fire from 1868 to 1969, as a by-product of the sort of industry that populists on the left and right look back at fondly.
The early 20th century gave us the worst of these sorts of phenomena; they were alleviated in the postwar years. Still, even then, factories inevitably were ugly, used a lot of water and energy, and depended on workers willing to take jobs much more hazardous than service or office work.
The claims about data centers using ungodly qualities of water and energy are often exaggerated, but — whatever the underlying merits — these sorts of complaints would have utterly doomed old-style industry.
If we can’t tolerate operations using prodigious amounts of water and energy — key inputs of the Industrial Revolution — we never could have had, for instance, integrated steel mills. They sucked up an unbelievable amount of water to circulate through furnaces and the like and, at their height, accounted for something like 5 percent of the entire country’s energy usage.
Where do the Great Lakes go to get all their water back? (Actually, much of it was returned in the course of the operations — the same is true of closed-loop data centers.)
Paper mills would have been a complete nonstarter, given the quantity of water needed to wash and bleach pulp, dilute the fiber, and generate steam, etc., not to mention the energy necessary for chipping and pulping wood and drying finished sheet.
Aluminum smelters, the largest of which used as much electricity as a city? Forget about it.
Chemical plants? Oil refineries? Or electric power plants, a different category of operation, but one that can use great quantities of water? No, no, and no.
We hear much about the arrogance of the titans of Big Tech and how it explains or justifies opposition to data centers. How does their high-handedness compare to the industrial titans of yore?
Well, we haven’t heard of any tech companies locking out the workers building their data centers, then sending in Pinkerton guards for a gun battle with the resisting electricians and plumbers, until the National Guard finally occupies the jurisdiction and re-establishes order.
That would be what Carnegie Steel did during the 1892 Homestead Strike near Pittsburgh. Labor strife was significantly reduced in the postwar era, but such unrest was part and parcel of the wave of industrialization that created the Golden Age of manufacturing in the U.S.
If the standard is that we won’t embrace industrial-scale projects if the people pushing for them are a**holes, presumably we would have rejected many manufacturing operations that provided blue-collar workers reliable, relatively well-compensated employment in the 20th century.
Now, data centers don’t provide the regular employment of old-school factories. The blue-collar workforce needed to build them and to maintain and update them is substantial, though, and driving up wages in exactly the fashion everyone has long said they wanted.
It is a symptom of the perversity of the current moment that we’ve imposed huge tariffs on the rest of the world in order to — at the margins and perhaps only in theory — create more blue-collar jobs, at the same time we are preparing to turn our backs on private-industry-driven investments to make such jobs more lucrative and create more of them.
If the tariffs are driven, in large part, by nostalgia, the data centers are meant to power the next wave of innovation. It doesn’t speak well of us if the backward-looking vision is more appealing than the future-oriented one (although the tariffs are unpopular).
One of the advantages of the mass manufacturing revolution at the time is that it was at the technological and economic frontier, where large-scale productivity gains and national power are forged. The AI revolution is at the frontier now, and it should be considered a minor miracle that an economic phenomenon all about computing power is driving a blue-collar construction boom. What’s not to like?
We never would have become the country we are now if the response to the physical assets necessary for economic growth and good blue-collar work had been, “Not in my backyard.”
© 2026 by King Features Syndicate
