AI Hastens Limits to Exponential Growth

TL;DR

This paper argues that AI-driven growth in energy demand could rapidly deplete Earth's resources, necessitating star system colonization and highlighting fundamental physical limits to expansion.

Contribution

It introduces a simple algebraic and system dynamics model showing how AI accelerates resource depletion and the physical limits of expansion beyond Earth.

Findings

AI could increase electricity growth from 3% to 15% annually.

Earth's resources are finite and depletion timelines are compressed by AI-driven demand.

Expansion beyond Earth requires capturing the entire Sun's output or colonizing other star systems.

Abstract

At any sustained positive growth rate of energy demand, depletion of all terrestrial energy resources, including non-renewable deuterium fusion and renewable solar, occurs within a remarkably compressed period. The time to depletion is inversely proportional to the demand growth rate. Artificial Intelligence (AI) has the potential to increase the growth rate of electricity from three percent per year to fifteen percent per year, effectively collapsing multi-millennial expansion timelines into decades. To grow after terrestrial depletion will require capturing more of the sun's output than the earth's cross-sectional area, eventually capturing the entire sun's output (Kardashev Type~II civilization). Expansion beyond that threshold requires colonizing other star systems. Simple algebraic models yield the main conclusions of the paper, supported by a system dynamics simulation. This analysis reveals that even unthinkably vast resources, such as total oceanic deuterium or the full luminosity of the Sun, are decidedly finite when viewed through a logarithmic lens. Uncertainties in the exact remaining resources of coal, oil, natural gas, and uranium do not affect the conclusions of this paper, as the fundamental physical limit is dictated by the geometry of expansion and the universal speed of light.