GCP Committed Use Discounts:
Negotiation Guide 2026

What Are Google Cloud Committed Use Discounts?

Google Cloud Committed Use Discounts (CUDs) are contractual agreements between an enterprise and Google Cloud that provide a discounted rate on specific Google Cloud resources in exchange for a 1-year or 3-year usage commitment. Unlike AWS's Enterprise Discount Program — which is a portfolio-level percentage discount on all qualifying spend — GCP CUDs are structured around specific resource types and regions, giving them a different risk/reward profile.

This guide is part of our Cloud FinOps negotiation guide and covers the commercial and negotiation dimensions of GCP CUDs — beyond the basic documentation that Google Cloud publishes. The focus throughout is on what's negotiable, what's not, and how to structure commitments to maximise savings while managing risk.

CUDs come in two primary flavors: resource-based CUDs and spend-based CUDs. Understanding the difference between these models — and when to use each — is foundational to any GCP cost optimisation strategy. Most enterprises use a combination of both types, with the mix depending on workload predictability, product breadth, and the enterprise's risk tolerance.

Resource-Based CUDs

Resource-based CUDs commit to a specific amount of vCPU and memory capacity in a specific region for 1 or 3 years. In exchange, Google Cloud applies a discount to that committed capacity: typically 37% for 1-year and 55% for 3-year commitments on general-purpose machine types, with up to 70% discounts on some memory-optimised configurations. Resource-based CUDs are the most cost-effective option for predictable, stable workloads — but they carry the highest commitment risk if capacity needs change significantly.

A critical characteristic of resource-based CUDs: they apply at the project or folder level and cover any VM instance that uses the committed resource type in the committed region, regardless of which specific instance types are running. This flexibility means you don't need to predict exactly which instance sizes will be running — just the total vCPU and memory footprint in each region. This is materially more flexible than AWS Reserved Instances, which historically required instance-type-level commitments (though Convertible RIs improved this).

Spend-Based CUDs

Spend-based CUDs commit to a minimum monthly spend on specific Google Cloud products — typically Cloud Run, Cloud SQL, VMware Engine, Google Kubernetes Engine, and several others — in exchange for a discount on that spend. The spend-based model trades some discount depth for reduced commitment risk: instead of committing to specific resource capacity, you're committing to a dollar amount of spend on a specific product category.

Spend-based CUDs are particularly valuable for products where resource-based CUDs aren't available — Cloud SQL and Cloud Run being the most commercially significant examples. For enterprises with significant database-as-a-service usage on GCP, Cloud SQL spend-based CUDs can represent one of the highest-impact cost reduction actions available.

CUDs and Sustained Use Discounts: The Interaction

One of the most commonly misunderstood aspects of GCP pricing is how Committed Use Discounts interact with Sustained Use Discounts (SUDs). SUDs are automatic discounts that Google Cloud applies when a VM runs for more than 25% of a month — no commitment required. The discount scales up to 30% for VMs running the full month. CUDs and SUDs don't stack in the way that AWS RIs and EDPs can. When a CUD applies to a resource, the CUD discount replaces the SUD rather than stacking with it.

The practical implication: for VMs running most of the month (getting close to the maximum 30% SUD), a 1-year CUD providing 37% discount is only 7 percentage points better than the automatic SUD. The incremental benefit of committing is lower than it appears when comparing against on-demand pricing. The 3-year CUD (55%) provides a meaningful 25 percentage point improvement over the maximum SUD, making 3-year CUDs far more commercially compelling than 1-year for highly stable workloads. This SUD/CUD dynamic should be factored into every CUD sizing decision — and is one of the areas where inexperienced buyers systematically overpay.

GCP CUD Discount Benchmarks

Google Cloud publishes standard CUD discount rates publicly — unlike AWS's EDP, which is entirely opaque. However, standard published rates are not the only discount lever available. Enterprises with significant GCP spend have access to negotiated pricing that goes beyond CUD rates, through Google Cloud's enterprise pricing programs and custom contract negotiations.

Machine Type	SUD (Automatic)	1-Year CUD	3-Year CUD
General Purpose (N2, E2)	Up to 30%	37%	55%
Compute Optimised (C2)	Up to 30%	37%	55%
Memory Optimised (M1, M2)	Up to 30%	Up to 40%	Up to 60–70%
Cloud SQL (Spend-Based)	None	~20%	~25–30%
Cloud Run (Spend-Based)	None	~17%	~23%
VMware Engine	None	~25%	~46%

Beyond published CUD rates, enterprises spending more than $1M/month on GCP can negotiate additional portfolio-level discounts through Google Cloud's custom pricing framework. These negotiations typically focus on: overall committed spend thresholds (similar to AWS EDP), pricing floor guarantees for specific services, credits for data migration and professional services, and co-investment arrangements for strategic workloads. Our Google Cloud negotiation advisory regularly identifies 15–30% additional savings beyond standard CUD rates for enterprise clients.

CUD Flexibility Provisions

One of the most significant changes Google Cloud made to its CUD program in recent years is the introduction of flexibility provisions for resource-based CUDs. These provisions allow enterprises to adjust their commitments without penalty in specific circumstances — and negotiating the right flexibility terms is often as important as the discount rate itself.

Machine Type Flexibility

Standard resource-based CUDs apply to specific machine families, not specific machine types. A commitment for N2 vCPUs can be used by any N2 machine type — N2-standard-2, N2-standard-4, N2-highmem-8, and so on. This flexibility is built into the standard CUD structure and doesn't need to be negotiated separately. However, it applies only within a machine family: an N2 commitment cannot be used against N2D or C2 instances. Enterprises running a mix of machine families need separate commitments for each family, or should use spend-based mechanisms for the portions of their fleet where flexibility is most valued.

Region Coverage

Resource-based CUDs are regional — a commitment in us-central1 cannot apply to workloads in us-east1. For enterprises with multi-region deployments, this means managing a portfolio of regional commitments. The commercial implication: under-utilised commitments in one region cannot offset over-utilised commitments in another. Proper sizing at the regional level — including headroom for workload migration between regions — is critical to CUD optimisation for multi-region deployments.

Cancellation and Modification

Standard GCP CUDs are non-cancellable and non-modifiable once purchased. This is materially different from AWS Savings Plans, which have a structured resale market on the AWS Marketplace. For enterprises buying CUDs through Google Cloud's self-service console, this inflexibility is a significant risk. For enterprises negotiating enterprise agreements directly with Google Cloud, contract-level flexibility provisions — including the ability to renegotiate commitments in the event of significant infrastructure changes — are achievable and should be pursued.

8 GCP CUD Negotiation Tactics

1. Benchmark Against AWS and Azure Before Negotiating

Google Cloud is more price-sensitive to competitive threat than most enterprise buyers realise. Entering a CUD negotiation with documented AWS equivalent pricing — including RI discounts and EDP rates — consistently produces better outcomes than negotiating based on GCP's own pricing alone. Demonstrating credible multi-cloud optionality changes the commercial dynamic fundamentally. This doesn't require an active migration plan — it requires credible evidence of cost benchmarking and the organisational capability to act on it.

2. Negotiate the Shortfall Provision

Enterprise CUD agreements through direct Google Cloud negotiation can include modified shortfall provisions. Standard CUDs require full payment of committed capacity regardless of utilisation — but custom agreements can include provisions for commitment re-profiling in the event of organisational changes (M&A, divestiture, significant workload migration). Negotiating these provisions before commitment is far more effective than seeking relief after the fact.

3. Bundle Infrastructure with Google Workspace

If your organisation uses Google Workspace (formerly G Suite), bundling your GCP infrastructure commitment with your Workspace renewal creates a larger commercial event that Google's enterprise team is more motivated to discount aggressively. The combined TCV of a 3-year GCP commit plus a 3-year Workspace agreement regularly unlocks 10–20% better pricing on both. This is particularly effective for organisations renewing Workspace in the same fiscal window as a GCP commitment decision.

4. Time to Google's Fiscal Year End

Google's fiscal year ends December 31. The final 6–8 weeks of Q4 (mid-November through December) consistently produce the most aggressive GCP commercial terms. Google's enterprise sales teams have quarterly and annual targets, and large commitments signed in this window regularly include enhanced credits, additional services, and better baseline pricing. If your GCP commitment can be timed to this window without creating operational risk, the commercial benefit is typically 5–15% above what's achievable at other times of year.

5. Use Anthos and AI/ML Ambitions as Leverage

Google Cloud is competing aggressively for enterprise AI and ML workloads — Vertex AI, BigQuery ML, and TPU access are strategic priorities. Enterprises with credible AI/ML workloads can use these as negotiation leverage: committing to run AI workloads on GCP's differentiated infrastructure (including TPUs and specialised ML instances) often unlocks additional discounts that go beyond standard CUD rates. Framing your commitment discussions around AI roadmap investment — not just infrastructure cost — shifts the conversation from price to strategic value, and typically produces better commercial outcomes.

6. Negotiate Credits for Migration and Testing

Enterprises migrating workloads to GCP or expanding their GCP footprint are in a strong position to negotiate migration credits, proof-of-concept credits, and professional services credits as part of a commitment agreement. These credits — which reduce the net cost of the first year of a commitment — are regularly available for enterprises making commitments above $500K/year and are frequently not mentioned by Google's commercial team unless the buyer requests them explicitly.

7. Right-Size Before Committing

The most common GCP CUD mistake is committing to current consumption rather than optimised consumption. Enterprises that right-size their GCP fleet before purchasing CUDs — using Google Cloud Recommender, third-party FinOps tools, or an independent assessment — consistently save 20–40% on their baseline before CUDs are applied. CUDs then apply to a smaller, more efficient baseline. The combination of right-sizing and CUDs produces far better outcomes than CUDs applied to an unreduced baseline.

8. Request a Custom Pricing Framework for >$1M/Year Spend

Enterprises spending more than $1M/year on GCP should be negotiating a Custom Pricing Framework (CPF) rather than relying solely on standard CUD rates. CPFs can include portfolio-level discounts that apply across all services, not just committed resource types. Google's enterprise team typically does not proactively offer CPFs to buyers who appear satisfied with standard CUD discounts — requesting this framework explicitly, backed by spend data and competitive alternatives, is the trigger for the discussion.

Common GCP CUD Mistakes

The most costly GCP CUD mistake enterprises make is purchasing CUDs without a structured FinOps practice in place. CUDs purchased without proper utilisation tracking frequently result in significant wasted commitment spend. Before purchasing any significant GCP commitment, enterprises should implement tooling that tracks CUD utilisation in real time and provides early warning of under-utilisation — Google Cloud's native tooling provides some of this, but third-party platforms typically offer better visibility and forecasting.

The second most common mistake is treating CUD strategy as a one-time event rather than an ongoing discipline. GCP workloads evolve, new services are added, old instances are retired. A CUD portfolio that was well-optimised 18 months ago may now include significant over-commitment in some regions and under-commitment in others. Quarterly CUD portfolio reviews — comparing commitment utilisation, pricing benchmarks, and upcoming commitment expirations — are a core practice for any mature cloud cost optimisation programme.

Third: failing to negotiate at the enterprise level. Many enterprises make CUD decisions at the workload team level, with individual product teams purchasing CUDs for their specific infrastructure without coordinating at the enterprise account level. This fragmented approach consistently produces worse commercial outcomes than a coordinated enterprise CUD strategy negotiated with Google's enterprise commercial team.

Real Result: 35% GCP Savings Through CUD Restructuring

A global technology firm engaged IT Negotiations to review their GCP commercial position ahead of a major commitment renewal. Initial analysis identified: over-committed resource CUDs in two US regions (running at ~60% utilisation), under-committed spend in Cloud SQL (no CUDs in place), and no custom pricing framework despite $2.8M/year in GCP spend. By right-sizing existing commitments, adding Cloud SQL spend CUDs, and negotiating a Custom Pricing Framework, the client achieved 35% total GCP cost reduction. See the full story in our Google Cloud CUD case study.

GCP CUD vs AWS EDP vs Azure MACC

Each hyperscaler's commitment program has a distinct structure, with different risk/reward profiles that suit different enterprise circumstances. AWS's EDP is the most flexible in terms of service coverage — a single percentage discount applies across all qualifying AWS services. Azure's MACC provides deep integration with Microsoft's broader commercial ecosystem, making it particularly valuable for enterprises with significant Microsoft on-premises footprint. GCP's CUD program offers the highest headline discount rates for compute, but requires more granular commitment management by region and machine family.

For enterprises with significant spend across multiple clouds, the interaction between these programs — and the negotiating leverage that multi-cloud optionality provides in each negotiation — is a significant strategic asset. Our multi-cloud cost optimisation guide covers this interaction in depth. Enterprises that negotiate each hyperscaler commitment independently, without coordinating the timing and terms, consistently leave value on the table that a coordinated multi-cloud strategy would capture.

The choice between increasing GCP commitment depth vs. diversifying to multi-cloud is rarely a purely technical decision — it's increasingly a commercial one. The threat of workload migration to a competing cloud, even if not immediately executable, is one of the most powerful cost reduction tools available to enterprise IT procurement teams.