The AI Industrial Explosion — Part 2: Transition Dynamics

This is Part 2 of a series on post-AGI economic growth. Part 1 established that a fully automated economy could double roughly every year using current technology. But the US economy does not currently look like a self-reproducing capital machine. It overproduces consumer goods and services relative to maximum growth, and underproduces machinery and raw metals. It cannot instantaneously switch to rapid growth, because it simply does not produce enough of the stuff that makes stuff. Using the input-output framework from Part 1, I model the transition sector by sector, solving for the optimal path from today's economy to the maximum-growth composition. Each sector's growth is constrained by its installed capacity and by the demands other growing sectors place on it. Consumption is held fixed so living standards don't fall. I find that energy production — probably the best single proxy for overall economic size, and the slowest sector to scale among those critical for growth — cannot double within the first two years after full automation, even with full reinvestment. The capital-goods sectors are only a small fraction of current US production, and recomposing the economy around them is the binding constraint. But the recomposition is fast in absolute terms. I find that energy production could double within roughly four years. Once that first doubling occurs, the economy has largely converged to the maximum-growth composition from Part 1, and the second doubling comes in roughly half the time. Within a decade of full automation, the economy could be many times its current size even with no further technological change. As in Part 1, we assume that every sector continues using exactly the techniques it uses today — the only change is that robots and AIs replace human workers. There are many obvious optimizations available once human labor is removed, even absent any technological change. We will discuss these in Part 3. Part 4 will consider how fast growth could be in t