Jennifer Harris: AI buildout is crowding out the broader economy

A New York Times opinion piece this week makes an argument worth taking seriously even if you do not buy every word of it. Writing in The New York Times, Jennifer M. Harris argues that the AI buildout has reached a scale where it is starting to crowd out the rest of the economy. Her phrasing: AI is vacuuming up so much of our land, talent, semiconductor chips, and building materials, and above all so much of our money, that it has begun to crowd out everything else.

The framing she leans on is a historical one. She calls what is happening "one of the largest peacetime mobilizations of capital in modern American history," with the buildout reportedly on track to top $1 trillion annually by next year. The part of her case that lands most concretely is the resource-competition argument. Capital, power, and political attention, she writes, are being diverted from other uses, especially in places where data centers compete with housing, manufacturing, or grid upgrades for the same land, labor, and transmission capacity.

Why this matters if you are not following macro debates: this is the first widely read mainstream op-ed I have seen frame the AI spending story as an affordability story rather than a tech story. The same dollars and the same transmission capacity that go into a training cluster do not go into a substation upgrade or a housing development. If that framing catches on with policymakers, the politics around data-center siting and utility rate cases start looking quite different than they did even six months ago.

The honest caveat is that this is opinion writing, single-authored, and the retrieved excerpt does not spell out a policy prescription, attach the trade-off to a dollar figure, or tell you who specifically pays the affordability cost. What the reporting also does not address is whether the implied productivity payoff will arrive on a timeline that justifies the crowding-out. The forward-looking question, and the one worth watching, is whether this affordability framing starts showing up in the next round of state-level data center policy and utility rate filings. If it does, the cost of building large AI infrastructure in the United States goes up.