Editorial photograph of a network operations room with engineers reviewing cloud capacity dashboards
Guide · Google Cloud · Commitments

Google Cloud commitments. CUD versus the savings model.

Google Cloud committed use discounts versus the flexible savings model in 2026. Resource and spend based commitment shapes, the realized discount bands, the three year lock in trap, and the buyer side coverage ladder.

Read the Guide Google Cloud Practice
500+Enterprise clients
$2B+Under advisory
Industry Recognized
500+ Enterprise Clients
$2B+ Under Advisory
11 Vendor Practices
100% Buyer Side Independent

Google Cloud committed use discounts buy the deepest rate but the most lock in, while the flexible savings model trades a few points of discount for the freedom to move.

Key takeaways

  • Two shapes. Resource based CUDs lock a machine shape for the deepest rate. Spend based and flexible CUDs commit a dollar figure and float across services.
  • Realized below headline. A three year resource based CUD lists up to 55 percent off but lands 38 to 50 percent once idle slices are subtracted.
  • The savings model. The flexible spend commitment behaves like a savings plan, trading a few points of discount for freedom to change machine family or region.
  • Lock in trap. A stranded three year commitment plus full rate new usage is the classic double pay. Google does not refund an unused resource based CUD.
  • Coverage ladder. Cover the base load deep, the swing layer flexibly, and leave the peak on demand.
  • Run it on data. Size commitments from the billing export, not from the Google account team coverage target.

How do Google Cloud committed use discounts work?

A Google Cloud committed use discount trades a one or three year commitment for a lower rate. You commit, Google discounts. The discount only pays off when the commitment matches real, durable usage.

Resource based and spend based commitments

Google sells two commitment shapes. Each fits a different workload pattern, and mixing them wrong is where the money leaks.

  • Resource based CUDs. A commitment to a fixed vCPU and memory amount in a region. Best for steady Compute Engine fleets.
  • Spend based CUDs. A commitment to an hourly dollar amount across eligible services. More flexible, slightly lower discount.
  • Flexible CUDs. A spend commitment that floats across machine families and regions, the closest thing to a savings plan.

Google documents both shapes on its committed use discounts page and the Compute Engine CUD overview.

What is the savings model and how does it differ from a CUD?

The savings model is the spend based, flexible commitment. It behaves like a savings plan. You commit to a dollar per hour figure and Google applies the discount automatically across qualifying usage. The trade is a smaller headline discount for far less lock in to one machine shape.

Where each model wins

  • Predictable fleets. Resource based CUDs win when the workload runs the same shape for three years.
  • Changing fleets. Flexible commitments win when machine families or regions shift inside the term.
  • Mixed estates. Most real estates need a layered blend, not a single model.

Compare the published rates on the Google Cloud pricing page and the spend based CUD documentation before you size anything.

What discounts do CUDs and savings commitments really deliver?

The headline discount and the realized discount are rarely the same number. The table below is the band we apply on Google Cloud commitment reviews, drawn from the engagements we ran in 2024 and 2025.

Google Cloud commitment discount bands 2026

ModelTermHeadline rate cutRealized band
Resource based CUD1 yearup to 37 percent25 to 33 percent
Resource based CUD3 yearup to 55 percent38 to 50 percent
Flexible CUD1 yearup to 28 percent20 to 28 percent
Flexible CUD3 yearup to 46 percent32 to 44 percent

Why the realized band sits below the headline

The headline assumes perfect utilization for the full term. Real estates idle a slice of every commitment. The realized band is the rate after that idle slice is subtracted.

What is the lock in risk on a three year commitment?

A three year resource based CUD locks the rate and the machine shape. When the workload moves to a newer family or another region, the commitment keeps billing while the new usage bills at full rate. That is the classic double pay.

How the double pay happens

  • Stranded commitment. The committed resource sits idle after a migration.
  • Full rate new usage. The replacement workload runs uncovered at on demand rates.
  • No clawback. Google does not refund an unused resource based CUD.

How should a buyer mix on demand, CUD, and savings commitments?

The right mix covers the steady base with the deepest discount and leaves the variable top uncommitted. Commit to the floor, never to the peak.

The coverage ladder

  • Base load. Cover the always on floor with resource based CUDs for the deepest rate.
  • Swing load. Cover the predictable but shifting middle with flexible commitments.
  • Peak load. Leave the spiky top on demand, where flexibility is worth the premium.

Track utilization through the Cloud Billing export to BigQuery so coverage decisions run on data, not on a forecast.

Where the common advice on Google Cloud committed use discounts is wrong

The common advice is to maximize coverage and commit as much of the estate as possible to the three year resource based CUD for the deepest rate. We disagree. In the Google Cloud commitment reviews Fredrik Filipsson ran in 2024 and 2025, the deepest rate routinely became the most expensive choice once a migration stranded the commitment. The buyer side move is to commit only the durable base load, blend flexible commitments into the swing layer, and price the option value of staying uncommitted on the peak. Coverage is not the goal. Net effective rate after stranded commitments is the goal.

Editorial photograph of server racks in a cloud data center with status lights along the aisle
A resource based commitment locks the machine shape as well as the rate, which is why a hardware refresh inside the term can strand the discount the buyer paid for.
30
Google Cloud commitment reviews 2024 to 2025
22%
Median spend sitting on idle commitments
3x
Stranding rate of resource versus flexible CUDs

Source: Redress Compliance advisory engagement file, 2024 to 2025.

The deepest published rate is not the cheapest outcome. The cheapest outcome is the rate that survives your next migration.

How does Redress engage on Google Cloud commitments?

The Redress engagement framework

Redress engages on Google Cloud commitments from the buyer side. Every engagement starts from your own usage and contract data, not from the vendor account team forecast.

  • Commitment review. A buyer side audit of current CUDs against real utilization and migration plans.
  • Negotiation support. Sizing the next commitment against durable base load, not against headline coverage.
  • Vendor Shield. An always on subscription that tracks commitment expiry and utilization across the cloud estate.

What to do next

  1. Pull the Cloud Billing export and measure the always on floor, the swing layer, and the spiky peak separately.
  2. Cover only the durable base load with resource based CUDs, sized to the floor not the average.
  3. Layer flexible commitments over the swing load so a family or region change does not strand the discount.
  4. Leave the peak on demand and price the option value of staying uncommitted there.
  5. Map every existing commitment to its expiry date and any planned migration before you renew.
  6. Re size or let lapse any commitment that a migration will strand inside its remaining term.
  7. Run the proposed coverage past Vendor Shield or a buyer side advisor before signing.
Cover of the Google Cloud FinOps CUD Guide white paper from Redress Compliance

White Paper · Google Cloud

Google Cloud FinOps CUD Guide

FinOps tools miss the committed use side of Google Cloud cost. Read it free.

Read the white paper

Frequently asked questions

What is the difference between a Google Cloud CUD and a savings plan?

A committed use discount commits either a fixed resource amount or a dollar spend for a one or three year term. The flexible spend based CUD is the savings plan equivalent on Google Cloud, applying automatically across qualifying usage with less lock in than a resource based commitment.

How much do Google Cloud committed use discounts save?

Resource based CUDs land roughly 25 to 33 percent off on a one year term and 38 to 50 percent off on a three year term after idle slices. Flexible commitments land a few points lower in exchange for the freedom to move across machine families and regions.

Can you cancel a Google Cloud committed use discount?

No. A committed use discount cannot be cancelled and an unused resource based commitment is not refunded. The commitment bills for the full term whether or not the resource runs, which is why sizing to the durable base load matters more than maximizing coverage.

What is the lock in risk on a three year CUD?

The lock in risk is the double pay. A migration to a newer machine family or another region strands the original commitment while the new usage bills at full on demand rates. Three year resource based commitments strand far more often than flexible ones.

Should I use resource based or flexible commitments?

Use resource based commitments for the always on base load that will run the same shape for the full term. Use flexible commitments for the swing layer that may shift family or region. Most real estates need a blend rather than a single model.

How do I size a Google Cloud commitment correctly?

Size the commitment to the always on floor measured from the billing export, not to the average and not to the account team coverage target. Cover the floor deeply, the swing layer flexibly, and leave the spiky peak on demand.

Do committed use discounts apply across projects?

Spend based and flexible commitments apply across eligible usage in the billing account, which lets them float across projects. Resource based commitments are tied to a specific region and resource type, so their reach is narrower by design.

How does Redress help with Google Cloud commitments?

Redress runs a buyer side commitment review from your billing export and utilization data, sizes the next commitment against durable base load, and tracks expiry and idle spend through Vendor Shield. Every recommendation runs on your data, not the Google forecast.

Redress is independent. Buyer side. Industry Recognized. Five hundred plus enterprise software engagements. $2B plus in client spend under advisory. Read the related Google Cloud practice, the software spend health check, and the Vendor Shield program.

Run the software spend health check against your actual Google Cloud commitment estate in under five minutes.
Open the Software Spend Health Check →
Advisory · Google Cloud

Talk to the Google Cloud practice.

A buyer side review of your committed use discount strategy, your flexible savings coverage, and your renewal mechanics. Used across more than five hundred enterprise software engagements.

Independent. Buyer side. Built for Google Cloud customers sizing the next commitment.

Google Cloud Practice

Open the Google Cloud advisory practice. Corporate email only.

Open the Practice →
More Reading

More from this practice.

Google Cloud Practice →
Google Cloud Advisory
Google Cloud · Practice
Google Cloud Advisory
The Google Cloud licensing and commitment practice.
14 min read
Software Spend Health Check
Tool · Assessment
Software Spend Health Check
Benchmark your cloud commitment waste in minutes.
5 min read
Vendor Shield
Program · Subscription
Vendor Shield
Always on advisory across the cloud estate.
10 min read
Renewal Program
Program · Renewal
Renewal Program
A managed twelve month commitment renewal sequence.
10 min read
Benchmark Program
Program · Benchmark
Benchmark Program
Subscription benchmarking across cloud vendors.
10 min read
Editorial photograph of an enterprise office at dusk

The advisor your vendors do not want.

Twenty years on the buy side. 500+ enterprises. $2B in client savings.

Google Cloud cost intelligence, monthly.

Committed use discount signals, flexible savings coverage signals, three year lock in signals, and the broader Google Cloud commitment leverage signals.