Ronald Coase was a student when he started wondering why firms exist. If markets coordinate so well, why do people work inside companies and take orders, when they could contract for every task? He published his answer in 1937 in a short paper, The Nature of the Firm. More than fifty years later it helped win him a Nobel prize.
His answer is cost. Using the market means finding someone and negotiating a contract, again and again for each piece of work. A firm saves that by hiring people once and directing them. It grows until running one more activity inside costs as much as buying it outside.
Agents move the boundary
AI agents make coordination inside a company cheaper. Once agents start bargaining with other agents, search and negotiation in the open market get cheaper too, and whichever effect proves larger in a given industry decides whether its firms grow or shrink.
My strategy game, Agentic Enterprise Sim, assumes the boundary moves a long way: a consulting firm with agents doing all the delivery. Coase would ask for the numbers first. They will differ by industry, since a contract for steel and a contract for strategy advice were never equally hard to write.