The current consensus, delivered breathlessly by pundits who have discovered that “bubble” fits every headline, is that we are living in a giant, expensive mistake. The predictions are familiar: imminent recession, the inevitable pop of a silicon-flavored fever dream, and a generation of workers contemplating a horizon that has apparently been recalled by the manufacturer. They aren entirely wrong about the money part. The capital flows are genuinely erratic, and the hype has achieved a kind of ambient exhaustion—you can hear it in the sighing of anyone who has been asked to “leverage AI synergies” in a quarterly review.
But there is a profound difference between a financial bubble and a structural shift in reality. One involves irrational exuberance. The other involves physics.
We are, at this precise moment in history, living through the dim lightbulb phase of machine intelligence. The first incandescent bulbs were expensive, inefficient, and—depending on whom you asked—a dangerous distraction from the perfectly serviceable gaslight. They didn power cities. They made it slightly less dark in one specific room, while occasionally threatening to burn that room down. Most observers concluded this was a passing novelty. They were not entirely right about that.
We are not building better chatbots. We are laying cognitive copper wire for a new kind of global infrastructure. We are building the grid.
The evidence for this is not in the volatility of stock tickers. It is in the quiet laboratories where something stranger is happening. AlphaFold has effectively solved the protein folding problem—a puzzle biology had declared too complex for human intuition alone—and turned it into an engineering discipline. GNoME has surfaced millions of stable crystal structures that materials science didn know existed, compressing what would otherwise have been centuries of work into months. These are not features. They are not “capabilities.” They are the first flickers of a light that, given time and care, will eventually illuminate the hardest problems our species has ever tried to solve.
We did not build the electrical grid because the investors were patient. We built it because the problems it would eventually solve were large enough to demand it.
And so we find ourselves at a threshold. The question is no longer whether this shift is real—it is already too late for that question. The question is whether we can move past the noise long enough to do anything useful with it. The petty squabbles. The race to claim territory. The “no kings” problem: the tendency of every era-defining technology to get captured by people more interested in who controls the lever than what the lever might lift.
This particular lever could move clean energy forward. It could provide traction against climate instability and resource scarcity. It could solve problems that human cognition, working alone, has simply not been fast enough to solve. But only if we look up from the quarterly report.
The path forward is not certain, and it is not easy. That is not the pitch. The pitch is simply this: stagnation—the alternative—is a far more certain catastrophe. We engage with this future not because we are blind to the risks, but because the problems we face do not offer a wait-and-see option.
It is time to listen to our better angels. Not the loud, nervous ones. The steadier ones—the ones that look at a dim, flickering, slightly dangerous bulb in a room that smells faintly of smoke and think: yes, but imagine the grid.
The current is coming. We might as well start building machines that can harness it.