Google Cloud Cost Optimisation: CUD, Flex CUD and Sustained Use Discount Strategy

Google Cloud's pricing model is built on a simple premise — if you don't commit, you pay full price. But the mechanics of Google's discount programmes are genuinely complex: Committed Use Discounts (CUDs), Flexible CUDs, Sustained Use Discounts (SUDs), and negotiated spend commitments all interact in ways that most enterprise FinOps teams don't fully understand. Get it right and you can cut your GCP bill by 30–55%. Get it wrong and you're leaving millions on the table.

How Google Cloud Prices Compute by Default

Google Cloud's on-demand pricing for Compute Engine is straightforward: you pay per second of compute usage, per gigabyte of memory, with no discounts applied unless you explicitly commit. For a standard n1-standard-4 machine in us-central1, the on-demand rate is approximately $0.19/hour. Run that workload for a year without any commitment, and you're paying around $1,664 per year per instance — a baseline from which all other discount mechanisms are measured.

The critical difference between Google Cloud and AWS is that Google automatically applies Sustained Use Discounts (SUDs) to your compute usage without any action on your part. If you run an instance continuously for 30% of a month, Google applies a 10% discount automatically. Run it for 75% of a month, and you get a 30% discount. This is fundamentally different from AWS, where Reserved Instances and Savings Plans require explicit commitment and upfront analysis.

This automatic SUD system creates a subtle trap for enterprise FinOps teams. Because Google is already applying discounts automatically, many teams assume they have reached cost optimisation and stop digging. In reality, SUDs represent a baseline discount floor, not a ceiling. The vast majority of enterprise Google Cloud users are dramatically underutilising the deeper discount mechanisms that Google provides — CUDs, Flex CUDs, and enterprise spend commitments.

Committed Use Discounts (CUDs) Explained

Google Cloud's Committed Use Discounts provide discounts of 25–57% off on-demand pricing in exchange for a 1 or 3-year commitment to purchase a specific amount of vCPU and memory resources in a specific region. The two main CUD models are resource-based CUDs and spend-based CUDs.

Resource-Based CUDs: vCPU and Memory Commitments

Resource-based CUDs lock you into purchasing a specific combination of vCPU cores and gigabytes of memory in a specific region. For example, you might commit to 100 vCPUs and 400GB of memory in us-central1 for three years. Google will apply the discount to any instance using those resources, regardless of machine type. This flexibility within a resource type is a significant advantage over AWS Standard Reserved Instances, which lock you to a specific instance type.

The discount tiers for 3-year resource-based CUDs reach 57% off on-demand for compute and 55% for memory. 1-year CUDs provide 25–35% discounts. The tradeoff is clear: longer commitment terms unlock deeper discounts, but expose you to the risk of underutilisation if your workload shrinks or your compute architecture changes.

Spend-Based CUDs: Dollar-Amount Commitments

Spend-based CUDs function differently. Instead of committing to specific vCPU and memory quantities, you commit to spending a specific dollar amount per month on Google Cloud services. These commitments are far more flexible than resource-based CUDs — they apply across compute types, regions, services, and even to different Google Cloud products like BigQuery and Cloud Storage. However, the discount rates are lower: 3-year spend-based CUDs provide approximately 30–35% discounts, compared to 50–57% for resource-based CUDs.

Spend-based CUDs are the better choice for enterprises with highly variable or unpredictable workloads, multi-region deployments, or those planning to expand their GCP footprint during the commitment period. The lower discount is a worthwhile tradeoff for the elimination of overcommitment risk.

Flexible CUDs: The New Approach

Google Cloud introduced Flexible Commitments (commonly called Flex CUDs) in 2023 as a response to customer demand for commitment discounts without the rigidity of 1 or 3-year terms. Flex CUDs allow you to commit to a specific dollar amount of monthly spending on compute services, with commitment terms as short as one month. The tradeoff: 1-month Flex CUDs provide approximately 40% discounts versus 25–57% for traditional CUDs, and there is no discount improvement for longer terms.

Flex CUDs fundamentally change the FinOps calculus for enterprises with highly variable compute loads. Previously, the choice was binary: commit to a CUD and lock in resources for 1–3 years, or run entirely on-demand and receive only the automatic SUD baseline. Flex CUDs create a middle path: capture deep discounts (40%) without long-term commitment risk. For workloads that scale seasonally, support temporary customer initiatives, or are in architectural transition, Flex CUDs represent a substantial improvement over pure on-demand pricing.

The optimal strategy for many enterprises is a layered approach: use traditional resource-based CUDs for the stable core of your workload (base load that runs 24/7 continuously), use Flex CUDs for the variable mid-tier (workloads that fluctuate 20–40% month-to-month), and keep on-demand capacity available for unpredictable peaks. This combination typically achieves 35–45% blended savings versus running entirely on-demand.

Sustained Use Discounts: The Automatic Baseline

Sustained Use Discounts (SUDs) are Google Cloud's automatic discount system — no commitment required. If an instance runs continuously for 25% of a month, Google automatically applies a 10% discount. If it runs 50% of the month, you get 20%. Run continuously for the entire month (100% utilisation) and you receive a 30% discount. This automatic application is the key difference from AWS, where commitment discounts must be explicitly purchased.

The critical nuance is that SUDs and CUDs operate on different logic and do not stack. When you purchase a CUD, Google applies the CUD discount instead of the SUD, not in addition to it. This means that if you have a workload running at 75% utilisation (qualifying for 30% SUD), a 3-year resource-based CUD providing 57% discount replaces the 30% SUD, not adds to it. The effective savings improvement from CUD is therefore 57% minus the 30% SUD you would have received anyway — a net improvement of 27 percentage points, not 57.

Understanding this interaction is crucial for FinOps teams evaluating whether to commit. For workloads that are already receiving significant SUD discounts (20%+ due to high utilisation), the marginal improvement from a CUD is smaller. For workloads that are sporadic or low-utilisation (receiving only 0–10% SUD), a CUD commitment delivers substantially more value.

Google Cloud Spend Commitments: Negotiating Beyond CUD

Google Cloud, like AWS, offers enterprise-level spend commitments that go beyond the standard CUD framework. These commitments require a minimum threshold (typically $1M+/year) and provide additional discounts negotiated case-by-case. Google's Enterprise Agreements can layer additional discounts on top of CUDs, creating scenarios where truly enterprise-scale customers negotiate total discounts of 40–55% across their entire GCP footprint.

The negotiation landscape for Google Cloud enterprise agreements differs significantly from AWS. Google's enterprise sales team is more conservative in leading with discount offers — they generally wait for customers to reach out rather than proactively pitching commitment programmes. Additionally, Google's discount framework is more opaque: unlike AWS, which publishes EDP discount ranges, Google treats each negotiation as bespoke. This opacity is actually an opportunity for well-informed buyers: having external negotiation expertise substantially improves leverage and final discount outcomes.

What Enterprise Spend Commitments Actually Achieve

• $1M–$3M annual GCP commitment: Enterprise discount typically 5–12% on top of CUD; sometimes co-invest credits ($25K–$100K); ability to convert spend commitments to CUDs mid-term
• $3M–$10M annual GCP commitment: Enterprise discount typically 10–18% on top of CUD; co-invest credits typically $100K–$300K; accelerator clauses for 20%+ growth
• $10M+ annual GCP commitment: Enterprise discount typically 15–25%+ on top of CUD; significant co-invest and innovation credits; custom SKU discounts for GCP Marketplace; dedicated sales engineering

BigQuery and Storage Cost Optimisation

Compute is typically 40–50% of total GCP spend for infrastructure-heavy organisations, but for data-driven enterprises, BigQuery and Cloud Storage costs can exceed compute entirely. BigQuery offers three pricing models: on-demand (you pay per TB queried), flat-rate (you commit to capacity via monthly reservations), and Editions (a newer model combining on-demand query pricing with per-hour commitment for compute capacity). Storage pricing varies by class: Standard ($0.020/GB/month in us-central1) down to Archive ($0.0036/GB/month), with automated lifecycle policies enabling substantial savings.

For organisations running significant BigQuery workloads (typically $500K+/year in query costs), the flat-rate or Editions pricing models deliver 30–50% savings versus on-demand pricing. The tradeoff is upfront commitment: you must commit to purchasing capacity by the month or year, and unused capacity expires. For analytics and data warehouse workloads, this is typically a worthwhile tradeoff. For data exploration, development, and ad-hoc query patterns, on-demand pricing often proves cheaper.

Cloud Storage cost optimisation centres on three levers: storage class optimisation (moving aged data to cheaper tiers), lifecycle policies (automatically transitioning or deleting data), and data deduplication/compression (reducing total data volume). Organisations commonly find that 40–60% of their Cloud Storage usage is aged data qualifying for Archive tier storage at 82% lower cost. Automated lifecycle policies capturing this opportunity typically take 4–6 weeks to deploy and reduce storage bills by 20–35%.

Google Workspace Enterprise Pricing: Often Overlooked

Many organisations negotiate their GCP compute and storage costs meticulously, then completely overlook Google Workspace (Gmail, Drive, Docs, Meet, etc.) pricing, which often represents 15–25% of total Google Cloud spending for enterprise customers. Google Workspace pricing is published but not negotiated — or so most FinOps teams assume. In reality, Google routinely negotiates Workspace pricing for large customers (typically 500+ seats), bundling Workspace cost reductions with cloud commitments.

Enterprise Workspace negotiations typically unlock 10–20% discounts on per-seat pricing for large deployments (1,000+ seats), with better rates for longer commitments and bundling with GCP infrastructure commitments. For a 2,000-person organisation running Workspace Enterprise (the standard tier for large enterprises), a negotiated 15% discount saves approximately $240K/year. This is not a minor opportunity, yet it sits outside the standard FinOps toolkit that most cost-optimisation teams operate with.

Common GCP Cost Mistakes

The biggest GCP cost mistake we see from enterprises is overcommitting to resource-based CUDs without sufficient visibility into actual workload shapes. Teams often commit to resources based on peak usage rather than average usage, then find themselves with significant unused CUD capacity as workloads shift or scale down. A common pattern: commit to 500 vCPUs across us-central1 based on year-end peak, then discover that average utilisation is only 60% of committed capacity. The result: you've paid for 500 vCPUs but only used 300, wasting the discount on the excess.

The second major mistake is ignoring the interaction between SUDs and CUDs. Teams will purchase a CUD for a workload that's already running at 75% utilisation (receiving 30% SUD), only to discover that the CUD discount replaces rather than stacks with the SUD. The marginal benefit is smaller than expected, and the commitment risk is higher than it initially appeared.

The third mistake is negotiating enterprise agreements without a clear understanding of what discounts are actually achievable at your spend level. Many teams accept Google's initial offer without counter-negotiation, not realising that additional 5–10% discounts are typically available with leverage and preparation. The difference between accepting Google's initial 8% enterprise discount and negotiating to 15% on a $5M annual commitment is $350K/year.

Building Your GCP Cost Optimisation Strategy

A complete GCP cost optimisation approach follows this sequence: (1) Audit your actual spend by service and resource type, identifying where your highest costs live. (2) Identify which workloads are candidates for commitment. Rule of thumb: workloads with utilisation >50% and expected tenure >12 months are strong CUD candidates. (3) Model commitment levels carefully. For each CUD candidate, calculate the payback period (when the savings exceed the commitment amount) and sensitivity to utilisation change. (4) Layer in Flex CUDs for variable workloads to capture 40% savings without locking in rigid resource quantities. (5) Negotiate your enterprise agreement with a clear understanding of your leverage and alternative scenarios. (6) Implement automation to enforce your commitment strategy — automatically assign CUD-discounted capacity to qualifying workloads.

The optimal result is a blended commitment strategy where your top 60% of compute spend is covered by a mix of traditional CUDs (35%) and Flex CUDs (25%), with the remaining 40% running on-demand or pure SUD basis. This combination typically delivers 35–45% blended savings versus pure on-demand pricing, while maintaining flexibility for workload changes and business scaling.

The Gainshare Approach to Google Cloud Negotiation

Traditional cloud cost optimisation forces teams to choose between DIY analysis (requiring deep technical expertise and time investment) or hiring consulting firms (requiring $100K–$500K upfront fees regardless of results). The gainshare model inverts this: we handle the entire process — audit, strategy, negotiation, implementation — and charge a percentage (typically 25%) of verified savings only. If we save you $1M/year, we charge $250K. If we save you nothing, you pay nothing.

For GCP specifically, the gainshare approach is particularly powerful because the negotiation landscape is opaque. Google's enterprise team doesn't publish discount frameworks, and most enterprise buyers have no reference point for what's actually achievable. Having negotiation specialists who understand Google's internal pricing structure, discount frameworks, and negotiation patterns substantially improves outcomes. Our clients typically realise 25–40% total GCP cost reductions — a figure that would require months of internal analysis to achieve independently, if it's achievable at all without external leverage.

The process is straightforward: we conduct a detailed audit of your GCP environment, model different commitment scenarios, and present you with a clear cost-benefit analysis. If you proceed, we negotiate your enterprise agreement and CUD strategy directly with Google on your behalf. You maintain complete control over the final terms — we simply provide expertise and negotiating leverage.