Intelligence as Infrastructure

Intelligence as Infrastructure names the transition underway as AI moves from application to substrate—from a thing people choose to use to a hidden layer that every other system rides on, as invisible and as load-bearing as the electrical grid. Leonard Kleinrock spent his career on this transition in an earlier technology: communication networks that began as specialized tools and became the invisible precondition of commerce, communication, and governance. His mathematics of queueing and flow—the iron law that delay explodes as load approaches capacity, the principle of decomposition that makes large-scale computation tractable, the vision of nomadic invisible computing—transfers directly to AI serving infrastructure, because the structural demands of infrastructure are invariant: it must be reliable, scalable, and provisioned for peak load, or it fails the systems that have come to depend on it. When intelligence becomes infrastructure, new concerns become paramount: equity of access (who is connected to the intelligence grid and who is not), the lock-in of early design choices (the internet’s early