Google Cloud CUD Negotiation Tactics

Google Cloud Committed Use Discounts trade flexibility for price. The customer commits to a level of resource use or spend for one or three years. In exchange, GCP discounts the rate by twenty to fifty seven percent depending on the resource family and the term length.

The CUD choice carries shortfall risk. Resource based CUDs lock the customer to specific machine families. Spend based CUDs offer flexibility at a smaller discount. BigQuery slots run their own commitment model. Customers who buy CUDs without modeling shortfall pay back the discount and more.

Read this alongside the Google Cloud knowledge hub and services, the GCP negotiation framework, and the Vendor Shield subscription.

Key Takeaways

What a CIO and procurement leader need to know in 90 seconds

Two CUD types apply. Resource based and spend based, with different discount levels and flexibility.
Resource CUDs discount thirty seven to fifty seven percent. Locked to a machine family and region.
Spend CUDs discount twenty to twenty eight percent. Flexible across machine families.
BigQuery slot commitments run separately. Annual and three year slot commits at thirty to fifty percent below on demand.
Shortfall risk is the most expensive mistake. Unused CUDs cannot be transferred or refunded.
Most estates run a mix of CUD types. Plus per second on demand for the bursty layer.
Enterprise negotiation adds an EDP layer. Enterprise Discount Program stacks on top of CUDs.

CUD types and machine families

Google offers two CUD shapes. Each shape covers a different resource scope and carries a different discount level and flexibility profile.

Two CUD shapes

CUD type	Scope	1 year discount	3 year discount	Flexibility
Resource CUD	Specific machine family, region	37 percent	55 to 57 percent	Low
Spend CUD	Family group across regions	20 to 25 percent	28 percent	High
BigQuery slot commit	BigQuery slot pool	~30 percent	~50 percent	Pool flexibility
Memorystore CUD	Memorystore Redis	17 to 27 percent	Up to 70 percent	Service specific

Resource CUD machine family coverage

General purpose. N2, N2D, E2, T2D, T2A.
Compute optimized. C2, C2D, C3.
Memory optimized. M1, M2, M3.
Accelerator optimized. A2, A3 with attached GPUs.
Storage optimized. Z3.

Discount math on a real workload

A typical mixed estate runs a baseline layer of steady consumption plus a bursty layer that scales with traffic. The CUD strategy covers the baseline. The on demand layer flexes the bursty traffic.

Cost model on a 1M monthly account

Layer	Share of compute	Strategy	Effective discount
Steady baseline	50 to 60 percent	3 year resource CUD	55 percent
Steady cross family	20 to 25 percent	1 year spend CUD	20 to 25 percent
Bursty top	15 to 25 percent	On demand per second	0 percent
Sustained use discount	Automatic on remaining	Auto applied	20 to 30 percent

Effective blended savings

Heavy CUD strategy. Effective discount of forty to forty five percent versus list, with shortfall risk.
Balanced CUD strategy. Effective discount of thirty to thirty five percent versus list, with cleaner flex.
Light CUD strategy. Effective discount of twenty to twenty five percent, lower commit risk.
No CUD. Sustained use discount applies automatically, six to thirty percent effective.

Shortfall is the silent cost

CUDs cannot be transferred between machine families or regions once purchased. A resource CUD on N2 in europe-west1 stays on N2 in europe-west1 for the life of the commit. Workloads that migrate to a newer family or a different region during the term leave the CUD stranded. Shortfall pays back the discount plus interest in two cycles.

BigQuery slot commitments

BigQuery runs its own commitment model separate from compute CUDs. Three options apply across analytic workloads.

Three slot models

On demand pricing. Per query bytes scanned, no commit, highest variability.
Annual slot commitment. Slots reserved for one year, discount around thirty percent.
Three year slot commitment. Slots reserved for three years, discount around fifty percent.

Which model fits

On demand. Variable workload below 500 slots peak.
Annual commit. Steady analytics workload above 500 slots.
Three year commit. Steady high volume workload with a clear three year horizon.
Mixed. Three year base plus on demand top for peaks.

Four buyer side levers

The four levers compress the GCP cost stack by twenty to thirty five percent against list. All four are negotiated before the EDP order form is signed.

Four levers that work

EDP discount tier. Enterprise Discount Program stacks on top of CUDs, ten to twenty five percent.
CUD type mix. Resource CUD on the locked baseline, spend CUD on the flexible layer.
Shortfall protection clause. Negotiate transfer rights between families on planned migration.
BigQuery flex slots. Mix annual commit plus flex slots for peak periods.

Traps to avoid

Over committing the baseline. Shortfall risk compounds across the term.
Single region commit. Multi region architectures need cross region flex.
Old machine family commit. Newer families launch annually with better price performance.
No shortfall language in EDP. Standard EDP terms do not include shortfall transfer rights.

Google Cloud CUDs are a leverage tool, not a savings tool. The discount only converts to savings when the customer matches the commit to a forecastable workload, picks the right CUD shape per layer, and negotiates the EDP terms above the CUD layer.

What to do next

The seven step checklist is the buyer side starting position before any Google Cloud commitment lands on the desk.

Baseline current GCP consumption. By service, by machine family, by region.
Forecast the workload across three years. Baseline, growth, and migration plan.
Choose the CUD mix. Resource versus spend versus BigQuery slot.
Negotiate the EDP discount tier. Stacking on top of CUDs.
Negotiate shortfall language. Transfer rights between families on planned migration.
Model the breakeven. CUD cost versus on demand cost at expected utilization.
Monitor CUD coverage monthly. CUD utilization, shortfall risk, sustained use discount delta.

Frequently asked questions

What is the difference between resource CUD and spend CUD?

Resource CUD locks to a specific machine family and region for a higher discount of thirty seven to fifty seven percent. Spend CUD flexes across a family group for a lower discount of twenty to twenty eight percent. Resource CUD suits steady workload with predictable family. Spend CUD suits flexible architectures.

Can CUDs be transferred between projects?

CUDs apply at the billing account level by default. They can be shared across projects under the same billing account. They cannot be transferred to a different billing account, a different region, or a different machine family without contractual rights negotiated in advance.

How does sustained use discount interact with CUDs?

Sustained use discount applies automatically to on demand usage that runs for a sustained portion of a month. CUDs apply first. SUD applies to the residual on demand layer above the CUD coverage. Both stack with EDP at the billing account level.

What happens to a CUD if the workload migrates?

The CUD stays locked to the family and region of the original commit. If the workload moves to a new family or region, the CUD remains billing the customer for unused commit while the new workload pays on demand. This is the most expensive CUD mistake.

How does Redress engage on GCP commits?

Redress runs Google Cloud advisory inside the Vendor Shield subscription, the Renewal Program, the Benchmark Program, and the Software Spend Assessment. Every engagement is led by a former Google Cloud commercial executive on the buyer side.

Is EDP separate from CUDs?

Yes. EDP is the Enterprise Discount Program. It is a multi year spend commitment at the billing account level. EDP discount stacks on top of CUD discount. CUDs lock specific resources. EDP discounts overall spend including services that do not support CUDs.

How Redress engages on Google Cloud

Read the related benchmarking, about us, locations, and contact pages.

Vendor Advisory

Cloud & Emerging

Programs

Assessments

Research

Knowledge Hubs

Assessment Tools

Google Cloud CUD. The math, the flex, the levers.