The AI Infrastructure Stack
Overview  /  Tier VI Energy
Layer 17

Transmission, Distribution & Grid Build-out

Wire, transformers, switchgear, EPC. The years-long backlog hidden behind every AI build.

What this layer does

A power plant 200 miles from a data center is useless without transmission. The US grid spent two decades barely growing — now it has to add hundreds of gigawatts of transmission and substation capacity to connect AI load to generation. The interconnection queue at most ISOs stretches 5+ years; large transformers have 2–3 year lead times and prices have roughly doubled since 2020.

This is one of the most under-appreciated cyclical-into-secular pivots in the public market. Eaton, Hubbell, Quanta, GE Vernova grid solutions, Hitachi Energy — all of them have backlog visibility further out than the cycle they’re reporting on.

Sub-categories

Analysis coming soon — will cover: transformer lead time as a leading capex indicator, Quanta backlog vs. revenue conversion, why Powell & Hubbell were the silent winners of 2024, copper exposure as derivative AI play (links to Layer 18), and reasons US grid build still under-runs the AI demand thesis.