What AI is costing us
The hub's first dollars-and-cents view: estimated AI token spend for every project that runs on Vertex AI, built from real usage priced against public list rates. Here's what we can show you today, and exactly how it's calculated — read the methodology below before trusting any number on this page.
How these numbers are calculated
Every dollar figure below is an estimate, not a bill. It covers AI token spend only — not Cloud Run, storage, or database costs, which this pipeline doesn't have access to query. Shown verbatim, exactly as the pipeline records it:
Summed across 3 GCP projects
The portfolio runs AI workloads in 3 GCP projects. Together they host all 4 hub products with AI usage — one GCP project hosts two products combined, so this total is a sum of projects, not an even split across products.
Estimated spend by project
Ordered by estimated cost over the last 7 days, highest first. Each project shows its 7-day and 30-day totals, the per-model breakdown behind them, and the pricing provenance for every model — including any model with no defensible price yet, shown honestly rather than dropped.
What's actually billed
A second, complementary data source: real dollars from the Google Cloud Billing Console, with discounts already applied, covering every GCP cost per project — not just AI. This is not the same as the Estimated tab, and it is not a live feed — read the capture details below before comparing the two.