Energy · AI · Infrastructure

We're Complaining About the Wrong Thing

AI's energy demand is being treated as a crisis. It might be the best gift America has received in fifty years — if we're willing to see it that way.

Think about the last time you read a headline about AI and energy. Probably something like: AI is straining the power grid. Or: Data centers are drinking our water. Or, my favorite genre, the breathless calculation of how many homes a single chat message could power.

The concern is real. I'm not dismissing it. But I've been sitting with a different question, one that I can't shake: what if we're looking at this completely backwards?

What if AI's energy demand isn't a crisis to manage — but a forcing function that finally justifies building the energy infrastructure America should have built decades ago?

We've been here before

In 1869, the transcontinental railroad was called a boondoggle. In 1883, critics thought the Brooklyn Bridge was dangerous fantasy — ferries worked fine, why risk it? In the 1950s, Eisenhower's Interstate Highway System faced opposition as an extraordinary expense for something nobody strictly needed.

The pattern is always the same: a big new demand arrives. It seems expensive, risky, maybe reckless. Skeptics focus on the cost. And then the infrastructure built to meet that demand becomes the foundation for the next hundred years of growth.

Here's the thing about each of those moments: in hindsight, the infrastructure was inevitable. The only question was whether America would lead it or watch someone else do it first. We led. And the jobs, the expertise, the geopolitical leverage — all of it followed.

What the demand actually looks like

Let's take the concern seriously, because the scale is genuinely striking.

Meta's Hyperion center in Louisiana alone will use roughly two to three times the power that New Orleans uses in a year. That's the scale we're talking about.

But read those numbers again. Notice what else they tell you.

They tell you that the richest, most capital-efficient companies in the history of capitalism have decided this demand is real, durable, and worth betting on. They've signed 20-year nuclear deals. They've committed billions. When Amazon, Google, and Microsoft are all building toward the same infrastructure future, that's not a trend — that's a signal about where the next century is going.

And they've done something else, something I think got buried in the headlines.

What we should actually be talking about

Here is the question that keeps nagging at me: if AI's energy demand is the forcing function, what exactly does it force?

It forces an upgrade of an electrical grid that, in many places, still runs on infrastructure and thinking from the 1940s. It creates the first real market signal at scale for small modular reactors, for advanced nuclear, for fusion research. It makes the economics work for clean energy projects that have been stuck on drawing boards for years, waiting for a buyer large enough to justify the investment.

Renewable energy is projected to cover nearly half of AI's demand growth by 2030. That's 280,000 solar workers today growing toward 400,000 by the early 2030s, in one of the few high-growth career tracks that doesn't require a four-year degree.

And that's just the energy piece. The same dynamic plays out everywhere AI is pushing demand.

Manufacturing. AI-powered robotics is making the economics of reshoring work for the first time in decades. A robotic factory in Ohio or Tennessee can compete with cheap offshore labor in ways a human-staffed factory simply cannot. Those jobs aren't the same jobs that left — they're better: engineers, systems integrators, technicians, maintenance specialists. But they're here.

Quantum computing. AI's computational demands are accelerating the development timeline for quantum hardware. The nations that lead in quantum will define intelligence capabilities for the next fifty years. Right now, that race is winnable.

Space. The compact high-output nuclear reactors being developed to power AI data centers are the same reactors that make serious deep-space exploration possible. The infrastructure we build to power AI on Earth is, in a very literal sense, infrastructure that could power humanity's expansion beyond it.

What leadership actually looks like

There's a version of this story where America gets cautious. We slow-walk nuclear permitting. We pass costs onto ratepayers. We treat data center construction as an environmental liability rather than an infrastructure investment. We wait for the rest of the world to prove the models work and then try to catch up.

That is a version of this story. It's not the one we should want.

When America led in aerospace, we built an industry that employs hundreds of thousands of people and exports billions in value annually. When we led in semiconductors, we created the leverage that shapes global supply chains to this day. When we led in software, we created the platforms that billions of people use and that have made American companies the most valuable on the planet.

Clean energy infrastructure built to power AI is the same kind of foundational bet. Countries that want the benefits of AI will need the energy infrastructure to support it. Who builds that infrastructure — and who they buy it from — matters enormously. Not just economically. Geopolitically.

The nations that sign long-term energy deals with American companies are not neutral parties. They're aligned. They're dependent on our standards, our expertise, our continued investment. That's not an accident — it's how infrastructure leadership works, and it's worked for us before.

Fear sells articles. Opportunity builds countries.

Here's the honest thing: 72% of Americans say they're concerned about AI's environmental impact. And the press covering those concerns isn't wrong about the facts. The demand is real. The grid stress is real. The water consumption is real.

But the framing has become almost entirely about cost, burden, and risk. Almost nothing about opportunity, investment, and what gets unlocked on the other side.

At the exact moment when people are worried about AI taking their jobs, we're also pushing back against the investments that would create the next generation of high-skilled American jobs — in solar, in nuclear, in grid modernization, in advanced manufacturing, in all the physical infrastructure that the digital economy depends on.

The Brooklyn Bridge builders didn't look at the East River and see a problem. They saw a river that needed a bridge. And then they built one that lasted 140 years and made everything around it more valuable.

We are standing at the edge of a similar moment. The demand exists. The technology is ready. The capital is willing to move. The only question is whether we treat this as a burden or an opportunity to finally build what we've been waiting to build.

AI might have just handed us the justification — in the form of demand, capital, and urgency — to upgrade an energy system that has been tapped out for decades, to reshore industries we've spent a generation offshoring, to build infrastructure that will still be running when our grandchildren are working in jobs we can't quite picture yet.

That's not a cost. That's a gift.

The question is whether we're going to treat it like one.

The Jobs Nobody Has Heard Of Yet

The real reason AI feels threatening has nothing to do with intelligence. It's about fear. And if history is any guide, we're afraid of exactly the wrong thing.

In the summer of 1996, we launched LinkShare. Affiliate marketing didn't exist as a business category. There was no job title for it, no industry conference, no trade association. When I told people what we were building, the most common response was a polite nod that meant: I have no idea what you're talking about.

Within a few years, millions of people were earning real income from it. Influencers, as we'd later call them — though that word didn't exist either — were a direct descendant of what we built. An entire economic layer appeared that nobody had predicted, nobody had trained for, and nobody had a five-year plan to enter.

That's how it always works. The jobs that define the next era don't show up in any forecast. They emerge from the infrastructure of the new thing, and then suddenly they're everywhere, and nobody can imagine the world without them.

I've been thinking about this constantly as I watch the conversation about AI and jobs play out. Because what I keep seeing isn't an honest reckoning with the future. It's a very old fear wearing a new mask.

The Graduate Moment

There's a scene in The Graduate where a family friend corners Dustin Hoffman at a party and leans in conspiratorially: "I just want to say one word to you. Just one word. Plastics."

It's played as comedy — the banality of it, the misplaced confidence, the collision between that generation's certainty and the protagonist's existential drift. But here's the thing: he wasn't wrong. Plastics was the future. The person who said it just couldn't explain why in a way that meant anything to someone who hadn't seen it yet.

I keep thinking about all the jobs being created right now that feel like that. The people building out AI infrastructure, new energy systems, autonomous aircraft, satellite intelligence networks, robotics platforms — they're the ones leaning in at the party. The rest of the room is doing the polite nod.

The jobs that didn't exist

Let me make this concrete, because it's easy to say "new jobs will emerge" as an abstract comfort and easy to dismiss it as the same thing optimists always say. So let's actually look at the record.

Notice the pattern. Not just that new jobs appeared, but that each wave paid better than what it replaced, required more interesting skills, and opened to people who had no obvious path into the previous category. The commercial internet alone now supports 28.4 million American jobs and drives 18% of US GDP. None of that was in any forecast from 1994.

The question isn't whether new jobs will come. They always have. The question is whether we're looking for them or looking backwards.

The UBI conversation bothers me

I want to be honest about something, even though it cuts against the grain of a lot of serious people I respect.

Universal basic income has become the implicit frame for a lot of the conversation about AI and work. The argument is essentially: automation will eliminate so many jobs that we'll need to pay people just to exist. It comes up at presidential debates. It comes up at Davos. Serious economists defend it with serious models.

And I find it, at some level, insulting to humanity.

Not because the concern isn't real — the disruption is real, and the transition will genuinely hurt people who are mid-career in fields that are about to change fast. That's not nothing. But the UBI frame assumes the endpoint is a world where human beings don't have enough useful things to do. That people, given time and resources and tools of extraordinary power, will... what, exactly? Run out of problems to solve? Run out of things to build? Run out of curiosity?

The thought experiment worth running

Here's the exercise I actually want to do. Not "what jobs will AI eliminate?" — that's the question everyone is already running, and it produces nothing but anxiety. The more interesting question is:

What happens to a civilization when the cost of executing ideas approaches zero?

What deflation actually means for regular people

There's one more piece of this that almost nobody is talking about, and it might be the most practically important.

AI, robotics, new energy sources, and advanced manufacturing don't just change what jobs exist. They change the price of everything. And not in a small way.

Think about what happens when energy gets dramatically cheaper — which is what building out nuclear and renewable infrastructure at scale actually does. Cheaper energy means cheaper manufacturing, cheaper food production, cheaper transportation, cheaper heating and cooling. Think about what happens when robotics lowers the cost of construction — housing, the biggest financial burden most families carry, becomes less expensive to build. Think about what happens when AI dramatically lowers the cost of legal advice, medical diagnosis, financial planning — services that have historically been available only to people who could afford them.

The people who lose the most from this transition are the incumbents in expensive, protected, inefficient industries. The people who gain are everyone else — especially people in the middle and at the bottom, for whom the high cost of those services has always been the biggest barrier to a better life.

So what do we actually do

I'm not arguing that the transition is painless. It isn't. People mid-career in fields that are about to change fast are going to face real disruption, and pretending otherwise is its own kind of dishonesty.

But the response to that disruption can't be to slow down the technology or to assume the endpoint is people sitting home collecting checks. The response has to be to move toward the new thing, not away from it.

From where I sit — having watched the affiliate marketing economy emerge from nothing, having helped build the intelligence infrastructure at Collective[i], having seen from my work with Spire what satellite data is doing to industries that didn't know they needed it — the pattern is always the same. The people who win are the ones who get curious about the new thing before everyone else does. The people who lose are the ones who spend their energy protecting the old thing.

The jobs coming out of AI, new energy, robotics, space, and autonomous systems are going to be better than the jobs they replace. Better paid, more interesting, more connected to things that actually matter. They're going to be distributed across geographies and backgrounds in ways the old economy wasn't. And they're going to include entire categories that don't exist yet, that we'll look back on in twenty years the way we look back on "social media influencer" — as something that was obviously going to be enormous and that almost nobody saw coming.

The question isn't whether those jobs will exist. They will.

The question is whether you're going to be looking for them.