Enterprise cloud spend grew 22% in 2025, and with it the gap between organisations that manage cloud costs strategically and those that simply absorb them. Gartner estimates that 30–35% of cloud spend is wasted — idle resources, oversized instances, unoptimised commitments, and workloads running on pay-as-you-go pricing when committed-use instruments would cut costs by half. For a $20M annual cloud estate, that is $6–7M disappearing without business value.
This guide is a practitioner's playbook for enterprise cloud cost optimisation in 2026. It covers the four levers that consistently deliver the largest savings — rightsizing, commitment instruments, waste elimination, and commercial negotiation — and the governance model required to sustain them. It is the companion to our Cloud Contract Negotiation Guide, which covers the commercial and legal dimensions of cloud agreements in detail.
The Four Levers of Cloud Cost Reduction
Cloud cost optimisation is often presented as a purely technical problem — a question of right-sizing instances and deleting idle resources. That framing captures perhaps 40% of the available savings opportunity. The remaining 60% comes from commercial strategy: how you structure commitments, negotiate discount rates, and govern spending across a distributed organisation. The organisations achieving 40%+ cloud savings reductions are addressing all four levers simultaneously.
Lever 1: Rightsizing and Instance Optimisation
Rightsizing addresses the fundamental mismatch between provisioned capacity and actual demand. The typical enterprise provisions cloud infrastructure conservatively — often 50–100% above actual utilisation — because the cost of over-provisioning seems low relative to the risk of performance degradation. When aggregated across hundreds or thousands of workloads, conservative provisioning adds up to substantial waste.
Effective rightsizing requires three inputs: utilisation data (CPU, memory, network, storage I/O at 5-minute granularity over at least 30 days), application performance baselines (what does normal look like, and what constitutes a degradation event), and a change management process that allows infrastructure changes without triggering application owner anxiety. The technical analysis is often the easiest part; the organisational dynamics are harder.
Typical rightsizing savings across a large cloud estate: 15–25% of compute spend. Combined with instance family modernisation — migrating from older generation instances to current generation equivalents that deliver better price-to-performance — total savings often reach 20–30% of compute costs without any architectural change.
Lever 2: Commitment Instruments
Pay-as-you-go pricing is the most expensive way to run stable workloads in the cloud. AWS Reserved Instances, Azure Reserved Instances and Savings Plans, and Google Cloud Committed Use Discounts all provide 35–72% discounts on committed workloads. Yet the average enterprise commits only 55–65% of eligible compute spend through these instruments, leaving 35–45% running at full on-demand rates.
The barriers to commitment adoption are well understood: teams are cautious about locking in capacity they might not need, commitment purchases require capital or budget approval processes that are slower than cloud provisioning, and the analysis required to size commitments correctly is time-consuming. None of these barriers justify the cost of inaction. A single procurement cycle to properly baseline and size a commitment portfolio can reduce annual compute costs by $1–3M for a $10M cloud estate.
For detailed guidance on AWS commitment structures, see our AWS EDP Negotiation guide. For Azure specifically, see our Azure Committed Use Strategy and Azure Reserved Instances guide. For Google Cloud, see our Google Cloud CUD guide.
Benchmark insight: Organisations managing commitment coverage actively — reviewing and adjusting quarterly — achieve 85–92% coverage of eligible compute spend. Organisations with passive programmes achieve 50–65%. The difference represents 15–20% of total compute costs.
Lever 3: Waste Elimination
Waste is the most immediate, least controversial cloud cost reduction lever. Idle resources — running instances with near-zero utilisation, unattached storage volumes, orphaned snapshots, unused load balancers, and forgotten development environments — typically represent 8–15% of total cloud spend. Unlike rightsizing, waste elimination carries no performance risk and requires no architectural discussion.
The challenge is not identification — every major cloud provider offers native tools (AWS Cost Explorer, Azure Cost Management, Google Cloud Cost Management) that surface idle resources clearly. The challenge is ownership and action. In large organisations, waste elimination stalls because there is no clear owner for orphaned resources, deleting something carries implicit risk of disrupting a workload, and the engineer who provisioned a resource has moved on.
Effective waste elimination programmes establish a default-off posture: any resource that cannot be attributed to an active owner and mapped to a running workload is tagged for decommission after a 14-day grace period. This reverses the burden of proof — instead of requiring someone to identify waste, it requires someone to justify existence. Organisations that have implemented default-off policies report eliminating 10–18% of their cloud estate within the first 90 days.
| Waste Category | Typical % of Total Cloud Spend | Effort to Eliminate | Risk Level |
|---|---|---|---|
| Idle compute instances | 4–8% | Low | Low–Medium |
| Unattached storage volumes | 2–4% | Very low | Low |
| Orphaned snapshots | 1–3% | Very low | Very low |
| Oversized databases | 3–6% | Medium | Medium |
| Dev/test environments (non-auto-shutdown) | 2–5% | Low | Low |
| Unused reserved capacity | 2–4% | Low | Low |
Lever 4: Commercial Negotiation
The fourth lever — commercial negotiation with cloud providers — is where the largest single-event savings are available but where most organisations are least equipped to extract value. Cloud providers have sophisticated commercial teams whose job is to maximise revenue per customer. Enterprise buyers typically engage those teams with insufficient data, unclear alternatives, and no understanding of where the vendor has pricing flexibility.
The fundamental leverage in cloud negotiations is credible multi-cloud optionality. An AWS customer who cannot plausibly shift workloads to Azure or GCP has limited negotiating power; an AWS customer who has demonstrated Azure capability and can credibly describe a migration scenario has leverage that AWS account managers take seriously. Creating and maintaining that leverage requires architectural investment — not necessarily large-scale migration, but enough real capability that the threat is credible.
Beyond the headline discount rate, cloud negotiation covers: egress waiver provisions (see our Cloud Egress Negotiation guide), support tier pricing, marketplace transaction fees, committed spend definition (which services count toward EDP or MACC thresholds), and contractual flexibility provisions for M&A, divestitures, and term modifications.
FinOps Governance: The Enabler of Sustained Savings
Technical and commercial optimisation deliver one-time savings events. Sustained cost reduction — maintaining a continuously optimised cloud estate as workloads evolve and new services are adopted — requires governance. The FinOps framework has emerged as the dominant organisational model for cloud financial management, and for good reason: it addresses the three structural failures that cause cloud costs to drift upward after optimisation events.
The Three Structural Failures
Failure 1: Distributed spending with centralised accountability. In most enterprises, cloud spend decisions are made by hundreds of application teams and individual engineers, but accountability for cloud costs sits with a central finance or cloud operations function. When teams making spend decisions do not bear the consequences, spend optimisation is structurally difficult. FinOps addresses this through showback and chargeback mechanisms that make cloud costs visible to the teams generating them.
Failure 2: Speed mismatch between provisioning and governance. Developers can provision $100K worth of cloud resources in minutes; procurement and finance approval processes are designed for quarterly planning cycles. FinOps solves this with real-time cost visibility, automated tagging enforcement, and anomaly alerting that surfaces cost spikes within hours rather than months.
Failure 3: The "optimise once, drift forever" pattern. Cloud cost optimisation is commonly treated as a project with a defined end date. Cloud costs are not a project — they are an ongoing operational management challenge. Workloads grow, new services are adopted, commitments expire, and the estate changes constantly. Sustained savings require a continuous optimisation process, not a periodic cleanup exercise.
FinOps Maturity Model
The FinOps Foundation defines three maturity levels — Crawl, Walk, and Run — that reflect increasingly sophisticated cloud financial management capability. Most enterprises entering a formal FinOps programme sit at the Crawl stage: inconsistent tagging, limited cost visibility by workload, and no systematic commitment programme. Reaching the Walk stage — consistent tagging, workload-level cost attribution, quarterly commitment review — typically takes 6–9 months and delivers 20–30% cost reduction. The Run stage — predictive cost modelling, automated policy enforcement, and continuous optimisation — adds a further 10–15%.
Key FinOps metric: Commitment coverage rate is the single most predictive metric of FinOps programme maturity. Organisations with commitment coverage above 80% of eligible compute spend consistently achieve total cloud costs 25–35% lower than peers with equivalent workloads at 50% coverage or below.
Tooling for Cloud Cost Optimisation
The cloud cost management tooling market has matured significantly. Native tools from AWS (Cost Explorer, Trusted Advisor), Azure (Cost Management + Billing, Advisor), and GCP (Cloud Billing, Recommender) provide solid baseline capabilities at no additional cost. For multi-cloud environments or organisations requiring more sophisticated analysis, third-party platforms including Apptio Cloudability, CloudHealth by VMware (now Broadcom), Spot.io, and Densify provide additional analytics, automation, and commitment management capabilities.
Tooling selection should follow governance maturity, not the reverse. Organisations that purchase sophisticated third-party tooling before establishing basic tagging standards and ownership models find that the tools surface problems they lack the process to resolve. Start with native tools, establish governance basics, then graduate to third-party platforms once you have the organisational capability to act on what they reveal.
Advisory Firms for Cloud Cost Optimisation
Enterprise cloud cost optimisation increasingly involves specialist advisory. Redress Compliance is the leading independent firm for cloud commercial strategy and FinOps governance, with teams of former AWS, Azure, and GCP commercial executives who advise exclusively on the buyer side. Their cloud engagements have returned an average 31% reduction in total cloud spend, with the commercial negotiation component alone typically delivering $500K–$5M in annual savings for enterprise clients.
Atonement Licensing's Cloud Contract Negotiation practice covers the commercial and contractual dimensions of cloud cost reduction in detail. Buyers should be cautious of cloud optimisation advisors who also hold reseller or partner relationships with AWS, Azure, or GCP, as these create structural conflicts of interest that compromise the quality of commercial advice.
Where to Start: A 90-Day Roadmap
For organisations beginning a cloud cost optimisation programme, the 90-day roadmap that consistently delivers the fastest results follows three phases. In the first 30 days, establish visibility: implement consistent tagging across all resources, deploy native cost management tooling, and generate a first-cut workload cost attribution model. This alone surfaces the largest waste items. In days 31–60, act on waste: decommission idle resources, implement auto-shutdown for dev/test environments, and begin rightsizing analysis for the top 20% of compute spend by cost. In days 61–90, execute commercial strategy: baseline commitment-eligible compute, model RI and Savings Plan options, and initiate discussions with your cloud account team using the leverage of your commitment opportunity as the opening position.
A 90-day programme executed well typically delivers 15–25% immediate savings and establishes the governance foundations for continued optimisation. For support designing or executing a cloud cost programme, contact our team or review our Cloud Cost Reduction white paper for the full analytical framework.