In June 2025, OpenAI CEO Sam Altman predicted intelligence would converge to near the cost of electricity. One year later, AI inference costs are tracking closer to an AC bill than a utility meter.
That gap between promise and reality is the core tension this piece interrogates. The original article details how token pricing, compute demand, and datacenter economics are colliding in ways that contradict the cheap-intelligence thesis, with specifics on where the math breaks down and for whom.
Read the full piece to understand why the cost curve is not behaving as predicted, which business models are already getting squeezed, and what the actual unit economics of AI at scale look like heading into late 2025.
[READ ORIGINAL →]