# AI Power & Grid: A Source-Backed Research Baseline

> Agent-assisted, human-reviewed Constraint Alpha research note. Confidence: high.

Published: 2026-08-15T10:30:00Z  
Research as of: 2026-07-12  
Scope: theme / AI Power & Grid Bottlenecks

## Summary

Constraint Alpha starts the AI power research process with the physical system: firm generation, grid connection, power distribution, cooling, and the qualified equipment needed to deliver usable megawatts to a data center.

## What changed

- The existing AI Power & Grid map is now available as a versioned research note for both human readers and machine consumers.
- The note separates public evidence from member-only thesis, risk, catalyst, and falsification fields.

## Why it matters

AI infrastructure growth can be delayed by equipment availability, interconnection, cooling, and project execution even when compute demand remains strong. Mapping those dependencies makes the research question testable instead of treating every electrical supplier as equivalent exposure.

## Evidence

- [Eaton data-center electrical portfolio](https://www.eaton.com/us/en-us/markets/data-centers.html) — switchgear, UPS, busway, and power-distribution relevance to data centers (as of 2026-07-12)
- [GE Vernova gas-power portfolio](https://www.gevernova.com/gas-power) — long-cycle firm-generation equipment within the AI power system (as of 2026-07-12)
- [Vertiv AI infrastructure portfolio](https://www.vertiv.com/en-us/solutions/ai-hub/) — high-density data-center power and thermal-management equipment (as of 2026-07-12)

Member-only thesis, risk, catalyst, falsification, and score fields are intentionally excluded.  
Canonical URL: https://constraintalpha.com/research/ai-power-grid-research-baseline