Cloud computing costs can easily spiral out of control if you only rely on pay-as-you-go pricing. You start with a few servers. Your team builds new features. Traffic grows. Suddenly, your monthly Google Cloud bill is taking a huge bite out of your profit margins. o get a handle on this, you should follow a comprehensive GCP Cost Management Guide to identify where your money is going.
CFOs often pay full retail price for servers that run 24 hours a day, 7 days a week. This is an expensive mistake.
The solution is a pricing model called Committed Use Discounts (CUDs).
CUDs are the main way businesses achieve big Google Cloud cost savings. This blog explains exactly how CUDs work. You will learn the different types of discounts, see real mathematical examples, learn how to buy them in the console, and understand the hidden risks you must avoid.
Key Takeaways:
A Google Cloud Committed Use Discount is a billing agreement. You promise to use a specific amount of cloud resources or spend a certain amount of money for a term of one or three years. In exchange for this promise, Google gives you a heavily discounted rate.
The maximum discount you can achieve with a Google Cloud CUD is up to 55% for standard machine types and up to 70% for memory-optimised instances.
Think of it like leasing an office building. If you rent a space month-to-month, you pay a premium rate. If you sign a three-year lease, the landlord lowers your monthly rent. CUDs apply this exact business logic to your cloud servers and databases.
CFOs often confuse CUDs with Sustained Use Discounts (SUDs). Both save you money, but they operate very differently.
A Sustained Use Discount is an automatic discount. Google applies it to your bill when a specific resource runs for more than 25% of the billing month. You do not need to sign a contract to get a SUD. However, the savings are much lower than a CUD.
Feature | Committed Use Discounts (CUDs) | Sustained Use Discounts (SUDs) |
Action Required | Manual purchase and contract signature. | None. Applied automatically by Google. |
Commitment Length | 1-year or 3-year term. | No term. Based on current month usage. |
Maximum Savings | Up to 55% (Standard) or 70% (Memory). | Up to 30%. |
Best Used For | Predictable, stable, long-term workloads. | Spiky, variable, or short-term workloads. |
Financial Risk | High. You pay even if you do not use the resource. | Zero. You only get the discount if you use the resource. |
You cannot stack a CUD and an SUD on the exact same resource. If a server is covered by your CUD contract, it receives the CUD pricing. It will not receive an additional SUD discount on top of that.
Google offers mainly two types of CUD contracts. You must choose the right type to avoid wasting money.
A resource-based CUD requires you to commit to specific hardware amounts in a specific location. You commit to a set number of vCPUs, a set amount of Memory (RAM), or specific GPUs.
You must also pick a specific Google Cloud Region (for example, us-central1). Your discount only applies to that exact hardware in that exact region.
Best for: Businesses with highly predictable workloads. If you know you will run the exact same database servers in the exact same location for the next three years, resource-based CUDs offer the deepest possible price cuts.
A spend-based CUD is much more forgiving. Instead of committing to specific hardware, you commit to spending a specific dollar amount per hour. For example, you might commit to spending $50 per hour on Google Cloud services.
Google applies your discount to any eligible service you run, regardless of the region or the machine type. If you turn off a server in New York and start a different server in London, your spend-based discount automatically applies to the new server.
Best for: Growing companies, dynamic environments, and teams using modern services like Cloud Run, Cloud SQL, or BigQuery. Spend-based CUDs give you financial flexibility while your engineering team changes the architecture.
Not every service inside Google Cloud is eligible for a CUD. If you buy a discount, you must ensure your team is using the correct services.
The most common eligible services include:
For a deeper dive into reducing cluster expenses, check out our guide on GKE Cost Optimization.
How do CUDs interact with your company's cloud account structure?
Most businesses use one main Cloud Billing Account that pays for multiple different Projects.
By default, Google applies a CUD only to the specific Project where you purchased it. However, you can turn on Discount Sharing.
When you enable Discount Sharing, Google looks at your entire Billing Account. If Project A does not use all of its committed resources, Google will automatically apply the leftover discount to eligible resources running in Project B. This ensures you do not waste your discount.
For finance teams, Google provides Proportional Attribution. This means the billing reports will clearly show which specific Project received the financial benefit of the shared discount, making internal accounting simple.
Let us look at exactly how the math works. We will compare standard pricing to a CUD contract.
Imagine your business runs a steady fleet of virtual machines (VMs) to host your main software application.
CFOs always ask two questions about these contracts.
Question 1: What happens if my business grows and I use more resources than I committed to?
Answer: Your cloud environment will not break. Your servers will keep running. However, any usage above your committed amount is simply billed at the standard, expensive on-demand rate.
Question 2: What happens if I use fewer resources than I committed to?
Answer: You lose money. If you commit to paying for 100 vCPUs, but your team deletes servers, and you only use 60 vCPUs, Google will still charge you for 100 vCPUs. This is called underutilization. You must pay your committed fee every single hour of the contract term, no matter what.
Buying a CUD is a permanent financial decision. Only users with the Billing Account Administrator role can make this purchase.
Here are the exact steps to buy a CUD:
After you buy a CUD, you must monitor it. Go to the CUD Analysis Tool inside the Billing dashboard. This tool shows a visual chart of your commitment coverage. While these native features are helpful, using specialized GCP Cost management tools can provide deeper insights into your spending patterns over time.
You want your usage line to sit slightly above your commitment line. If your usage drops below your commitment line, you are wasting money. Google also provides a Recommendations page. This page uses your past billing data to suggest new CUDs you should buy to save more money.
CUDs offer massive savings, but they carry strict business risks. You must understand these rules before you sign a contract.
Vendor and Hardware Lock-In: Resource-based CUDs lock you into specific hardware. If you buy a 3-year commitment for an N2 machine family, you cannot simply transfer that discount to a newer C3 machine family next year. Your engineering team is locked into using older hardware to maintain the discount.
Non-Eligible Services: Hardware commitments do not apply to every compute service. For example, they do not cover resources created by the App Engine flexible environment or Dataflow. If your team uses these services, a standard CUD will not reduce those costs.
No Cancellations or Edits:
Can you cancel a CUD if your business downsizes? No. Once you click purchase, the contract is permanent. You cannot cancel it, pause it, or reduce the size of the commitment. You are legally obligated to pay the monthly fee for the entire 1-year or 3-year term.
"Spend-based CUDs are easier to manage but give slightly lower discounts. Resource-based requires exact, flawless capacity planning, which is nearly impossible for a growing startup." - Sourabh Kapoor, CEO at Costimizer"
Managing cloud costs is a dedicated business practice called FinOps. Here is how experienced FinOps professionals handle Google Cloud committed use discounts.
Never Guess Your Baseline
Do not buy CUDs based on what you think you will use. Look at your historical billing data over the last 90 days. Find your absolute lowest point of usage. This is your "baseline."
Start Small and Layer Commitments
Do not try to cover 100% of your cloud spend with one massive 3-year contract. Buy a CUD to cover just 60% of your baseline usage. Let the remaining 40% run on pay-as-you-go pricing or automatic Sustained Use Discounts. As your business grows and your baseline rises, you can buy a second, smaller CUD and layer it on top of the first one.
Centralize Purchasing
Do not let individual engineering teams buy their own CUDs. All purchases should be made centrally by the finance or FinOps team. This ensures the company can use Discount Sharing to move savings around to whichever project needs it most.
Google Cloud Committed Use Discounts are the most powerful tool you have to lower your cloud bills. If your business runs stable, predictable software applications, relying on pay-as-you-go pricing is a complete waste of capital. By understanding the difference between spend-based and resource-based contracts, and by avoiding the trap of over-committing, you can safely cut your cloud expenses by up to 70%.
However, identifying your exact baseline and managing these rigid contracts manually is difficult and risky. One wrong decision locks you into a three-year financial mistake.
"This is why modern businesses use Costimizer."
Costimizer is an agentic AI platform built specifically for FinOps. Instead of staring at confusing billing reports, Costimizer automatically analyses your AWS, Azure, and GCP environments. It identifies your exact baseline, enforces strict budgets, and safely executes the best cloud cost optimisation strategies for you.
Connect your cloud account to Costimizer today and let our AI automatically secure your savings.
No, unlike AWS Reserved Instances, Google Cloud does not require an upfront capital payment for CUDs. You are simply billed the discounted rate monthly for the duration of your 1-year or 3-year contract.
Costimizer is built natively for multi-cloud environments, unifying cost management for AWS, Azure, and GCP into a single dashboard. Your finance team can manage AWS Savings Plans, Azure Reservations, and GCP CUDs simultaneously without jumping between different vendor consoles.
No, Google Cloud CUDs are strictly tied to the Cloud Billing Account where they were purchased. They cannot be sold on a secondary market, transferred to an external organisation, or cancelled under any circumstances.
We operate on a Zero-Risk Guarantee: if Costimizer does not identify more savings than the cost of your subscription, your first month is free. Our Growth and Enterprise plans scale transparently based on your total annual cloud spend, ensuring the tool always pays for itself multiple times over.
CUDs are applied to your eligible resource usage first to lower the base rate. Your custom negotiated enterprise discounts are then typically applied to the remaining balance, ensuring you benefit from both, though you should verify exact stacking rules with your Google account manager.
Setup begins with secure, read-only access so our platform can analyse your billing data and provide immediate optimisation insights. You only grant execution permissions when you are fully comfortable letting the AI autonomously execute discount purchases and resource right-sizing.
If you hold a Spend-Based (Flexible) CUD, your discount will automatically float to cover the new Cloud Run usage. If you purchased a Resource-Based (Hardware) CUD, it will remain locked to the abandoned VMs, resulting in underutilization charges.
Costimizer’s Agentic AI analyses your precise historical workload data to calculate your absolute lowest continuous usage baseline. We only automate commitments up to a safe, mathematical threshold, leaving variable traffic on pay-as-you-go to completely eliminate underutilization risk.
