Anchoring the Limits of Computation in Physical Reality

In the baseline production function, capital represents tangible capital. Historically, this meant factories, machinery, and raw materials. In the AI era, capital represents compute: data centers, GPUs, fiber-optic networks, and the energy required to power them.

The Fallacy of Infinite Scale:

Classical economic applications of technology often suffer from a "digital blindness." Because software historically scaled with near-zero marginal cost, management theory incorrectly assumed that the infrastructure supporting AI, the new capital, could scale infinitely in the "cloud." Consequently, corporate investment models treat compute as an abstract, limitless resource where scaling up invariably drives proportional increases in total economic output.

The Endogenous Reality:

Internal boundaries dictates that continuous growth must be sustainable from within the system. Current AI deployment models are fundamentally exogenous and extractive; they ignore the natural laws of physics. The true theoretical boundary of capital is thermodynamic. Massive compute requires physical land, rare earth mineral extraction, and immense cooling capacity (water and power).

The Economic Impact of Mismanagement:

When organizations ignore the physical realities of capital, they misprice their capital investments. As AI models scale, the law of diminishing returns forcefully applies to the physical world. The marginal cost of cooling a data center and securing reliable grid power eventually outpaces the marginal efficiency gained by a slightly larger language model. When capital breaches its ecological carrying capacity, its contribution to output turns negative due to supply chain shocks, grid failures, and the prohibitive costs of thermodynamic management.