From the newsletter (sign-up if you want it sooner): America’s AI infrastructure boom is creating an unexpected bottleneck: the people needed to build it. Data centers are competing with broadband for scarce technicians and trades, threatening to deepen the digital divide.
We could write here about Lumen, but we have written a lot about them, and it’s a lot (of good stuff, about trebling fiber miles, building global fabrics, and smashing the competition) to rehash in 200 words for the top of a newsletter. We could write as well about drone sensing with Lockheed Martin and Verizon in Miami, or AST SpaceMobile’s latest results – which are both worth searching out, but have been written about thematically in this section just recently. Plus we could write about any number of other things. But we will take a look at Matt’s piece about a dogfight in the AI jobs market, between juggernaut data centers and rural broadband providers in US back waters. This is not about models or robots, but a fight for human expertise – for a shortage of fiber technicians.
The US has two infrastructure priorities converging on one scarce resource: people who know how to build networks. AI data centers are sucking skilled workers into a construction boom, just as the long-delayed BEAD initiative finally starts moving into deployment. And actually, it is not just about the fiber technicians; the same labor pool includes electricians, HVAC specialists, surveyors, trenchers, engineers, and other trades needed to build broadband networks. The numbers are already uncomfortable. The industry was estimated in 2024 to be short around 58,000 workers to meet BEAD’s goals. AT&T plans to add 3,000 technicians this year. Meta has its own fiber-training programme. Meanwhile, the US telco workforce has fallen from 1.4 million in 2001 to fewer than 598,000 in 2025.
There is a stark economic imbalance here, of course. A hyperscaler or major carrier building AI infrastructure can afford recruitment, training, and wages that smaller community-based broadband operators cannot. Every worker pulled toward a lucrative data-center project is potentially one less available for a rural fiber build. And this is before the AI construction boom really peaks. Economist Anirban Basu expects labor pressure to worsen in 2027 as AI infrastructure spending expands. So the irony is clear: America’s AI infrastructure boom could make it harder, slower, and more expensive to deliver the connectivity needed by Americans still waiting for broadband.
We wrote at the start of the year about a K-shaped digital divide, where AI drives the haves on an exponential boom cycle and the have-nots into a downward spiral. Well, here it is, hitting the AI builders themselves.
