09/01/2026
I Went Down a Rabbit Hole on Data Centers…
I've been trying to better understand all the controversy surrounding data centers—the electricity they consume, water use, utility costs, and what they may mean for our communities. Well, I went down a rabbit hole, and I discovered that I've been thinking about data centers too narrowly. The AI boom isn't simply a data-center story, it's an enormous infrastructure story. I had wondered whether some of the answers might be to put more infrastructure underground, or even in space. As it turns out, there's some truth to both, just not quite the way I thought.
Starlink and others are developing "direct-to-cell" technology that allows satellites in low-Earth orbit to communicate directly with ordinary cell phones. In simple terms, we're beginning to put something resembling cell towers in space. Pretty amazing, but here's the distinction I hadn't appreciated: Moving communications into space doesn't move the data centers, or their enormous appetite for electricity, off the ground. In fact, AI is creating a tremendous need for more infrastructure here on Earth: power generation, transmission lines, substations, transformers, and miles of cable.
So why not put all those power lines underground? Turns out that's not quite the answer either. Underground can make sense in certain places and can protect critical infrastructure, but burying the entire electrical grid would be extraordinarily expensive and impractical. The more likely future is much more interesting: more communications capability in space, while simultaneously building a much bigger, stronger and smarter electrical grid on the ground, with selective undergrounding where it makes sense.
This changed the data-center debate for me. Maybe the question isn't simply, "Do we want data centers?" Maybe we should also be asking, “Where will all this new electricity come from? Who will build the infrastructure to deliver it? Who will pay for it? And how do we make sure ordinary utility customers aren't left unfairly footing the bill?”
And then, inevitably, my investor brain kicked in. Wall Street is already well aware of this story, so I'm apparently a little late to that party, but it gave me a different way to think about AI. I've spent so much time hearing about who will win the AI race that I haven’t stopped to think about everything all those competitors are going to need, regardless of who wins. Somebody must generate the electricity, move it, build the substations and transformers, manufacture the cable, and increasingly, provide communications infrastructure from space.
That's the rabbit hole I found particularly fascinating. Sometimes the more interesting question isn't who will win the race. It's what everyone in the race is going to have to buy.
From Lynn's Desk
Send a message to learn more
