I keep seeing predictions on social media about massive job losses from AI. People seem convinced that automation will hollow out entire professions. But when I look at what is actually happening in my own work, I see something different. The constraint just moved.
A single developer can now produce ten times the code they could a year ago. That sounds like a reason to hire fewer people. But the opposite is happening. The faster we can build software, the more I need UX designers, business analysts, product specialists, and sales people. The bottleneck did not disappear. It shifted.
AI reduces the cost of producing software. That threatens some narrow, repetitive coding tasks. But cheaper software increases demand for software. Companies attempt more projects, more experiments, more integrations, more custom workflows. More projects create more surrounding work. Now you need more people to figure out what to build, why it matters, how it fits the business, how users adopt it, and how it gets sold and supported.
This is not a new phenomenon. It is a core principle of systems thinking. In any connected system, increasing throughput in one area redistributes pressure to the others. When one part of a business system becomes dramatically more efficient, the work does not simply disappear. The constraint moves. If coding gets much faster, then product definition, UX, quality assurance, sales, onboarding, governance, and support become the new constraints.
The Theory of Constraints explains that once one bottleneck is reduced, another part of the system becomes the new limiting factor. You do not eliminate the constraint. You relocate it. Complementarity explains that when one capability becomes cheaper, the value of the surrounding capabilities often increases. Jevons paradox adds that when something becomes easier and less expensive to produce, demand for it often rises rather than falls. What begins as a productivity gain in one function can therefore expand the workload across the entire organization.
The result is not necessarily less work. It is a redistribution of work toward the areas that make increased output useful, usable, and valuable. The faster we can build, the more we need people who can decide what is worth building. The more software we ship, the more we need people who can explain it, sell it, support it, and make sure it actually solves the problem it was meant to solve.
“The faster we can build software, the more we need people to figure out what to build, why it matters, and how it fits the business.”
I am not dismissing the real disruption happening in certain roles. Some tasks will disappear. Some jobs will change. But the idea that AI simply reduces headcount misses the larger dynamic. When you make one thing cheaper and faster, you increase demand for everything connected to it. The system does not shrink. It shifts.
So when I hear predictions about job losses, I think about what I am actually seeing. I think about the UX designer I just hired because we are shipping more features than we can properly design. I think about the business analyst we need because the volume of integrations has outpaced our ability to define requirements. I think about the sales team struggling to keep up with the new products we can now build in a fraction of the time.
The constraint just moved. The question is whether we are paying attention to where it went. The next time someone tells you AI will eliminate jobs, ask them where the new bottleneck will be. Then ask how many people it will take to clear it.
