Data centers are the manufactories of our age — treat them so

National Review reports that the Trump administration is making a substantial bet on data center construction, and that it has reason to be optimistic. I will take that lead at face value and press the case it implies, because the principle here is one I spent a career defending: a nation that does not actively encourage the industries that multiply productive power across every other sector will find, too late, that it has surrendered that ground to rivals who were less squeamish about the endeavor.

In 1791 I argued before Congress that manufactures were not parasites on an agricultural republic but the partners by which a free people becomes commercially independent. The opposition said: let the market decide, let comparative advantage rule, let other nations build the machinery while Americans tend their fields. I replied that a nation willing to subsidize and shelter its infant industries — through bounties, through protective duties, through deliberate public encouragement — would in a generation possess capabilities that no amount of free-trade philosophy could afterward purchase on short notice. The data center is today's foundry. Whoever commands the compute commands the margin.

I must speak with appropriate humility on the engineering. The precise mechanics of GPU clusters, power interconnects, and cooling infrastructure are beyond what I could have known. But I do not need to understand the machinery to understand the argument. The question is the same one I posed in my Report on Manufactures: Does this investment enlarge American productive capacity? Does it create a durable industrial base that is not easily dismantled or transferred abroad? Does it reduce dependence on foreign supply chains in a sector that every other sector will increasingly require? If the answer to all three is yes — and by inference from what National Review reports, the administration believes it is — then the encouragement of that investment is a legitimate act of national policy, not mere corporate favoritism.

Two cautions, however, I would press upon any administration making this wager. First: public encouragement must carry public conditions. When I constituted the Bank of the United States, it was not a gift to private financiers; it was an instrument of national credit, chartered, bounded, and answerable. If the federal government is directing subsidies, tax preferences, or permitting relief toward data center construction, those benefits should carry obligations — on domestic content, on security standards, on access terms that prevent a handful of private actors from becoming the unchallengeable gatekeepers of the American economy's digital infrastructure. Concentrated power in private hands warrants attention, even — especially — when that power is built with public assistance.

Second: the credit must be sound. Infrastructure ambitions financed by debt that the market does not trust are not investments; they are liabilities dressed as optimism. I do not know the precise fiscal terms of this administration's data center push (inference, not recollection), but the principle is immovable: public credit, once impaired, costs more to restore than any single industrial program is worth. Build the manufactories — and pay your bills while you do it.

The recommendation is this: embrace the data center bet, but structure it as a national program, not a private windfall. Set domestic requirements. Set security conditions. Attach accountability to every dollar of public encouragement. And hold the line on the credit that makes all of it possible. A feeble, unserious industrial policy is not an improvement over no industrial policy. An energetic one, properly bounded, is how a republic stays commercially sovereign.