The pin factory, reconsidered

When I described the pin factory in The Wealth of Nations, my point was not that specialization is inherently good — it is that specialization, coordinated by exchange, lowers cost and raises output for everyone who trades. The data center, as reported by the Washington Examiner, is a modern instance of exactly that logic: concentrated, specialized infrastructure enabling a vast division of cognitive labor across the economy. So far, so promising.

But the Washington Examiner's lead immediately flags the matter I cannot pass by: the backlash in northern Virginia began over utility prices. That is the consumer speaking. The consumer is the sovereign of any honest market, and when she complains that a new class of industrial user is raising her electricity bill, we are not in the presence of a mere aesthetic dispute — we are in the presence of a classic problem of shared infrastructure and uncompensated cost.

The mercantile writer of every age has a ready answer: the new industry creates jobs, attracts capital, and will eventually lower prices through its own productivity. I do not dismiss this. But I observed, and I observe still, that the merchant is very eloquent about the benefits he creates and very quiet about the costs he imposes on others. The hyperscaler's lobbyist is no different from the clothier who assured Parliament that the Navigation Acts protected English labor, when they principally protected English profit at the consumer's expense.

The institutional question — which I take to be the only question worth settling — is this: who bears the cost of the data center's power draw, and who decides? If the cost is socialized onto ordinary ratepayers through regulated utility pricing, while the profit is privatized by the hyperscaler's shareholders, we have arranged a subsidy by another name. I would call it mercantile preference dressed in the language of innovation. Inference, not recollection, leads me to suspect this is precisely what the northern Virginia backlash detected, even if it could not fully articulate it.

The proper remedy is not to ban data centers, as some in the backlash appear to demand — that is merely replacing one monopoly of judgment with another. The remedy is the institutional one: transparent cost allocation, competitive procurement of power, and a regulatory body genuinely independent of the industry it oversees. A regulator who has been captured by the firm she regulates is not a regulator at all; she is a chartered monopolist's secretary. I wrote at length about the East India Company for precisely this reason.

Public goods — roads, harbors, basic education — require public investment because the private return is insufficient to call them forth. The grid that powers the data center is, in important respects, a public good of this kind. It would be a peculiar arrangement indeed if the private firm extracted the public infrastructure's value without contributing to its maintenance. The consumer in northern Virginia is not wrong to notice. The question is whether the institutions of democratic government have the clarity of purpose to make the hyperscaler pay her fair share — and whether the political argument the hyperscalers are currently winning is an argument about the public interest, or merely about their own.