Source: David Pogue
David Pogue documents the concrete environmental cost of current AI systems—not speculative future risks, but present-day energy consumption doubling every six months. This shifts the AI adoption debate from capability or ethics to resource scarcity: if training and inference cycles consume electricity at exponential rates, the infrastructure bottleneck arrives before market saturation does, forcing hard choices about which AI applications justify their environmental footprint. The "make it optional" framing matters because it reveals we've already normalized AI as default infrastructure rather than treating deployment as a deliberate trade-off.