Committed use discounts cut Google Cloud compute cost by up to 70 percent, but the resource lock and the contract terms decide whether the commitment pays off.
Google Cloud committed use discounts reach deep rate cuts, but the region and resource lock on the resource based version strands value when workloads move, and the contract is negotiable.
A committed use discount commits you to a level of usage for one or three years in exchange for a deep rate cut. Google bills the discounted rate on usage that matches the commitment and On Demand on anything above it.
The commitment runs whether you use it or not. That is the core risk, and it is why the choice between the two CUD types matters more than the headline rate.
Google documents both models on its committed use discounts overview and the resource based detail on the Compute Engine CUD page. Read both before committing.
A resource based CUD commits you to a quantity of vCPU and memory in a single region. It reaches the deepest rate, up to 70 percent, but it cannot move. If the workload leaves that region, the commitment stays behind.
A spend based CUD commits you to a dollar per hour amount rather than specific resources. It covers a wider set of services and is not pinned to a region in the same way, which makes it the safer instrument for estates that still change shape.
Google Cloud CUD types compared
| CUD type | Commitment unit | Max rate cut | Portability |
|---|---|---|---|
| Resource based | vCPU and memory in one region | Up to 70% | Region and family locked |
| Spend based (Compute) | Dollar per hour of compute | Up to 46% | Flexible across families |
| Spend based (eligible services) | Dollar per hour of spend | Varies by service | Broad service coverage |
| Sustained use discount | Automatic, no commitment | Up to 30% | Applies automatically |
Commit resource based only where the workload is genuinely fixed in a region for the full term. Everywhere else, spend based CUDs protect you from the region lock for a modest rate trade.
The decision is the same as on any cloud commitment. The deepest rate is worthless if the commitment strands when your architecture moves.
Sustained use discounts apply automatically to Compute Engine usage that runs for a large share of the month, and they stack with CUDs without a separate commitment. Google explains the interaction in its sustained use discount documentation. Model them as a floor, not as additive savings on the same committed hours.
The standard Google account team pitch is to maximize savings with three year resource based CUDs across the estate. We disagree. In roughly half the Google Cloud estates we benchmarked in 2024 and 2025, the region locked resource based CUD stranded 15 to 30 percent of its value when a workload migrated region before the term ended. The buyer side move is to cover only the workloads you know are fixed in region with resource based CUDs, and use spend based CUDs for everything that might move. The deepest rate is a trap when the commitment cannot follow the workload.
Source: Redress Compliance advisory engagement file, 2024 to 2025.
On Google Cloud the deepest committed rate is only the best deal if the workload never leaves the region you committed it to.
The common mistakes are over committing on the three year term, double counting sustained use discounts, and ignoring the enterprise agreement layer entirely. Each adds avoidable cost.
Large customers negotiate a Google Cloud agreement that can add a discount layer on top of public CUD rates and adjust contract terms. Google publishes its general pricing on the Google Cloud pricing page, but the negotiated rate sits in the agreement. Treat the public rate as the floor of the conversation, not the ceiling.
Size to the stable usage floor and layer additional commitment as usage proves out. Use the billing reports in the Cloud Billing documentation to find the floor, then commit below it, not at the recommended line.
The tactics are the same discipline applied to a negotiable contract. Size conservatively, prefer the flexible instrument, and negotiate the agreement, not just the published rate.
Committed use discounts cut eligible Compute Engine cost by up to 70 percent on a three year resource based commitment. Spend based commitments save less, up to about 46 percent on compute, in exchange for far more flexibility.
A resource based CUD commits you to a fixed amount of vCPU and memory in one region for the deepest rate but cannot move. A spend based CUD commits you to a dollar per hour amount, covers more services, and is not pinned to a region in the same way.
Sustained use discounts apply automatically to long running Compute Engine usage and do combine with committed use discounts, but they are not additive on the same committed hours. Model them as a baseline floor rather than as separate savings.
No. Committed use discounts run for the full one or three year term and cannot be cancelled. This is why sizing the commitment to stable usage and preferring the flexible spend based instrument matters.
The public CUD rate is fixed, but a large customer's overall Google Cloud agreement is negotiable and can add a discount layer above the public rate. Treat the published rate as the floor of the negotiation.
The commitment stays in the original region. If the workload migrates, the resource based CUD continues to bill in the region it was bought in, which is how estates strand 15 to 30 percent of committed value.
Commit only the usage floor you have held stable for two quarters or more, prefer spend based CUDs for anything that might move, and layer additional commitment as usage proves out rather than in one large bet.
Choose three years only for workloads genuinely fixed in region and family for that period. For everything else the one year term limits the shelfware risk while still delivering a meaningful rate cut.
CUD types and ratios, the region lock, sustained use discount stacking, and the contract levers that cut an over committed Google Cloud estate.
Used across more than five hundred enterprise engagements. Independent. Buyer side. Built for procurement leaders running the next renewal cycle.
