The Power Curve

Technology has always been a multiplier of human capability, but the nature of that multiplication has changed.

A medieval king could command an army of thousands, but each soldier had to be individually fed, equipped, trained, and led. The king's power was bounded by logistics, by the physical constraints of moving bodies through space. A nineteenth-century industrialist could command factories, but each factory required workers, and workers could organize, strike, or leave. A twentieth-century media mogul could reach millions of minds, but the infrastructure of broadcasting imposed its own limits: the number of channels, the cost of production, the physics of transmission.

In each era, the amplification of individual power was real but friction-bound. There were always bottlenecks, whether physical, economic, or social, that prevented any single actor from scaling their influence without limit.

Artificial intelligence changes the shape of this curve.

Software can be copied at near-zero marginal cost. Networks can scale without adding proportional infrastructure. Algorithms operate continuously, without fatigue, without forgetting, without hesitation. A machine does not need to be convinced; it does not need to be motivated; it does not need to sleep. And critically, it can make decisions, or something functionally indistinguishable from decisions, without waiting for a human operator to intervene.

The result is a qualitative shift. Technology no longer merely makes individuals more productive. It makes organizations more powerful in a way that is increasingly decoupled from the number of human beings those organizations employ.

This distinction matters enormously.

If AI multiplies the capability of an organization, then whoever controls the organization gains a corresponding multiplication of power. A single person directing a sufficiently advanced AI-enabled enterprise could, in principle, exert influence that once required the coordinated effort of thousands. The leverage is not just economic. It is strategic, informational, and potentially coercive.

And the race to acquire this leverage is already well underway. Governments understand what is at stake. Corporations understand it. Military planners understand it. Investors understand it. And the engineers building the systems understand it perhaps best of all.

This creates the oldest strategic dilemma in human history, dressed in new clothing: if you believe a technology could reshape the world, and you believe your adversaries will build it regardless, then choosing not to build it yourself begins to feel not just impractical but irresponsible. Safety becomes a luxury you cannot afford if your competitor does not share your caution.

This is how arms races begin. And arms races have a characteristic pathology: the finish line becomes more important than the road. The question of whether we should build something gets quietly replaced by the question of whether we can afford not to.