Power is not merely a footnote in the AI race, some secondary constraint to be solved with better chips and cleverer code. It is the constraint. The quiet, physical governor that decides what scales, what stalls, and who ultimately gets to play at the highest level.
And it does not stop at AI.
It bleeds directly into the broader reindustrialization now underway—a process that, if you listen closely beneath the noise, marks a reversal of the last three to four decades. Those decades were defined by a single, elegant idea: chase cost downward. Outsource relentlessly. Move production to wherever labor is cheapest, regulation is lightest, and friction is lowest. Let supply chains stretch across oceans and continents, because efficiency—on paper—demands it.
The result was predictable. Manufacturing migrated. First gradually, then decisively. China became the gravitational center, with others orbiting in similar roles. Even services—once thought immovable—began to follow, digitized and distributed across time zones.
It worked. Until it didn’t.
Then came the stress test.
COVID did not introduce fragility into the system; it revealed it. A hyper-optimized, globally interdependent network suddenly found itself unable to perform the most basic of functions. Goods stalled. Components vanished. Entire industries discovered, often too late, that their “just-in-time” brilliance had quietly transformed into “just-too-late” paralysis.
At one point, you could not reliably get something as trivial as a cotton swab without discovering that it, too, had been folded into a global supply chain stretching far beyond any reasonable margin of safety.
It was, for a brief moment, a lesson.
And like most lessons that are both obvious and inconvenient, it is already fading.
Memory, particularly collective memory, is short. The urgency dissolves. The shelves refill. The narrative resets. But something did shift, even if people prefer not to dwell on it.
Globalism, as a coherent, unchallenged doctrine, was already fraying before COVID. The pandemic did not kill it, but it stripped away the illusion that it was universally beneficial, universally stable, universally rational.
Because a truly globalized system assumes something that reality stubbornly refuses to provide: uniformity. It assumes that nations operate under comparable incentives, comparable constraints, comparable levels of trust. That they will continue to cooperate even when cooperation becomes inconvenient.
They don’t.
They never have.
Different political systems, different priorities, different thresholds for risk and disruption—these are not bugs in the system. They are the system. And when stress arrives, those differences do not smooth out. They amplify.
So the grand theory—that global integration would lift all boats, harmonize interests, and produce a kind of economic equilibrium beneficial to all—has run into the less accommodating shoreline of reality.
And now, slowly, unevenly, we see the response.
Regionalization.
Not as an ideological statement, but as a practical adjustment. Shorter supply chains. Redundancy where once there was singular dependence. A renewed interest in domestic or near-domestic production—not because it is always cheaper in the narrow sense, but because it is more controllable.
More resilient.
Reindustrialization follows naturally from this. The developed world, having spent decades exporting its industrial base, now finds itself attempting to rebuild it. Not in its old form—smokestacks and assembly lines as far as the eye can see—but in a hybrid configuration.
Supported by automation. By robotics. By AI. By technologies that compress labor requirements while increasing precision and output. Even demographics—aging populations—become part of the equation, not as a constraint to be lamented, but as a driver for systems that can operate with fewer hands.
But all of this rests on a foundation that is far less glamorous and far more decisive.
Energy.
Cheap. Reliable. Scalable energy.
Without it, the rest is theory.
With it, the rest becomes possible.
Nuclear, in an ideal world, would be the obvious anchor. Dense, stable, capable of delivering large volumes of power with minimal variability. It ticks the boxes that planners care about when they think in decades rather than quarters.
But we do not operate in an ideal world.
Nuclear projects are slow. Painfully slow. Years stretch into decades. Costs escalate. Regulatory and political hurdles accumulate. By the time a plant comes online, the context in which it was conceived may have shifted entirely.
So in the here and now, another fuel carries the burden.
Gas.
Flexible, abundant—at least in certain regions—and capable of being deployed at the pace that modern industry demands. It is not perfect, but perfection is not the requirement. Availability is.
And this is where the asymmetry becomes difficult to ignore.
The United States, by virtue of shale, occupies a position that is not easily replicated. It has access to volumes of gas that keep domestic prices low, that support industrial activity, that provide a buffer against external shocks. It can power data centers, factories, and the sprawling infrastructure of an AI-driven economy without negotiating every unit of energy on the global market.
Others are not so fortunate.
They must import. Compete. Navigate chokepoints. Pay premiums. Their reindustrialization efforts, their AI ambitions, their attempts to rebuild what was once exported—all of it runs through an energy constraint that is far tighter.
In a world that is fragmenting, that is rediscovering the importance of proximity and control, this matters more than it has in decades.
Power—literal power—is becoming the differentiator.
Not slogans. Not strategies written in policy papers. Not aspirations.
Megawatts.
Reliable ones.
And those who have them, at scale and at cost, will find that many other pieces of the puzzle begin to fall into place with surprising ease.
The rest will spend their time compensating.
https://wattsupwiththat.com/2026/03/21/the-ai-race-isnt-just-about-chips-its-about-power/