There is a company that will generate $1 billion in revenue this year with five employees. Not five hundred. Not five thousand. Five. The employees are not doing the work in any traditional sense. They are directing AI systems that are doing the work. The AI systems are doing the work of what would have been, ten years ago, a company of several hundred people.
This is not a hypothetical. This is happening now, in multiple industries, at multiple scales. The decoupling of economic value from headcount is not a future trend. It is a present reality. And it changes everything about how we think about work, employment, and economic participation.
The traditional employment relationship bundled several things together: access to capital, access to tools, access to distribution, and access to coordination. You needed an employer because the employer had the factory, the software, the customer relationships, and the management structure that allowed individual effort to produce economic value at scale.
AI is unbundling this. The tools are now accessible to individuals. The capital requirements for starting certain kinds of businesses have dropped to near zero. Distribution, through digital platforms, is available to anyone. Coordination, through AI systems, no longer requires a large management hierarchy.
What remains valuable is judgment. Domain expertise. The ability to ask the right question, to recognize the pattern that matters, to know which of the ten things the AI generated is actually correct and which three are subtly wrong in ways that will cause problems later. That is experience. That is what thirty years of working in a specific domain produces.
The question is not whether AI changes the economics of work. It already has. The question is who designs the systems that determine who captures the value that AI creates.
Right now, the answer is: whoever owns the AI. The platforms, the model providers, the companies with the capital to deploy AI at scale. The experienced professionals who provide the judgment that makes AI useful are not capturing the value their judgment creates. They are providing it, often for free, to systems that aggregate it and sell it back to them.
Building infrastructure that allows experienced professionals to capture the value of their judgment is the work. A billion dollars with five employees is a feature, not a bug, if the five employees are the ones who built the system. It is a problem if the five employees are the ones whose expertise was used to train it.
