Finance & cloud cost teams
Google Cloud Committed Use Discount Optimization
Google Cloud bills compute at an on-demand rate, but committing to 1- or 3-year Committed Use Discounts (CUDs) cuts that cost substantially. The hard part is deciding how much to commit — and most tools answer it by sizing to your usage history and hedging the rest. Solvicus instead takes the demand you already know (contracts, migrations, a steady baseline) and optimizes commitments for it directly: over-commit and you waste capital on idle capacity; under-commit and you keep paying full on-demand rates.
Solvicus computes the cost-optimal CUD purchase plan for your Google Cloud usage — across any machine family and region — net of the commitments you already own, and shows you the coverage, utilization, and savings behind every recommendation.
Under the hood, each resource is modeled as its own linear or mixed-integer program, built from your contracts' demand and existing CUDs, and solved for the commitments to buy within your decision window plus the resulting on-demand usage over a longer evaluation horizon that accounts for demand tapering as contracts lapse and renew. You can then edit any purchase, run what-if scenarios, and stress-test the plan in an interactive dashboard — Solvicus recommends and evaluates, but never touches your cloud console.
Required inputs
- Your Google Cloud usage: per-resource monthly demand from your contracts or workloads (for example vCPUs and RAM)
- The committed use discounts you already own, with their coverage and expiry
- On-demand and 1-/3-year CUD pricing for the resources you run
Solver output
- The cost-optimal set of 1- and 3-year CUD purchases to make within your decision window
- Month-by-month on-demand usage, coverage, and utilization over a longer evaluation horizon, with the savings and break-even behind each commitment
Example agent prompt
“Act as a Solvicus FinOps agent and use gcp_optimize_capacity. We run two Google Cloud resources on N2 in us-central1: vCPU and RAM. Contract Atlas needs 120 vCPU/month from month 1 through 24; contract Borealis needs 200 GiB RAM/month from month 1 through 36. We already own a CUD covering 60 vCPU/month through month 12. All prices are per unit per month: vCPU is $23.08 on-demand, $14.54 on a 12-month commitment and $10.38 on a 36-month commitment; RAM is $3.09 on-demand, $1.95 on a 12-month commitment and $1.39 on a 36-month commitment — or look the current rates up with gcp_get_machine_pricing. Assume demand decays 20% per year on any contract that lapses without renewal. Recommend the 12- and 36-month commitments to purchase within the next 6 months, evaluated over a 42-month horizon, that minimize our total spend across both resources. Then use gcp_evaluate_capacity to show what that plan costs month by month.”
Have a recurring optimization problem in mind? We'll build a solver for it.