Wed, October 28, 2026
18 min read
Designing Google Cloud Spending Caps Around a Four-Service Native Cap — Defense-in-Depth for the AI Agent Era
#google-cloud#finops#cost-management#governance#cloud
Google Cloud's commercial contracts do not provide a hard cap that stops all services at a target dollar amount across an entire billing account. Spend Cap Budgets, GA since late October 2026, cover only Gemini API, Agent Platform (formerly Vertex AI), Cloud Run, and Cloud Run functions, limited to single-project × single-service × monthly scope. Reseller-billed accounts are excluded, in-flight requests keep billing, and persistent resource fees do not stop. Alerts-only Budgets are notification-only, and billing data can lag more than 24 hours. With AI coding agents able to drive Google Cloud APIs directly, preventing runaway bills means layering Cloud Quotas, Organization Policies, IAM, and billing link locks on production as prevention, Budgets and anomaly detection as detection, and Spend Cap and Pub/Sub automation as enforcement and response. This post walks through that design against the official behavior as of November 2026, including the dual cap structure of Gemini API's tier limit and AI Studio project cap.
read more →