The Waste That Refuses to Go Away
Flexera has surveyed cloud waste annually for over five years. The number has been stable: roughly 30 percent of cloud spend is waste (Flexera, "2024 State of the Cloud Report"). Year over year, in a category that consumes hundreds of billions of dollars globally, 30 percent disappears.
The waste is not because the optimization techniques are unknown. The waste is because the discipline of pulling the optimization levers in sequence requires organizational commitment that most enterprises do not sustain.
If your cloud spend has grown faster than your business and finance is now asking pointed questions, the FinOps playbook is well-established. The work is executing it. Six levers, in the right order, eliminate most of the waste.
Coined Frame: The Six Levers in Sequence
Six levers cut cloud cost. The sequence matters because they unlock each other and because early wins build organizational momentum for the harder work later.
Lever 1 - Visibility. You cannot optimize what you cannot see. Cost dashboards by team, product, environment, and workload. Allocation tags applied consistently. Cost monitoring that produces alerts on anomalies. Most enterprises have partial visibility and discover they have more waste than they expected once visibility is complete.
Lever 2 - Right-sizing. Instance sizes matched to actual utilization. Storage tiers matched to access patterns. Database instance sizes matched to workload. The mechanical lever that produces the fastest wins. Typical savings: 15-25 percent of compute spend.
Lever 3 - Reserved capacity. Long-term commitments for steady workloads. Reserved instances, savings plans, committed use discounts. Typical savings: 25-45 percent over on-demand pricing for the workloads that qualify.
Lever 4 - Idle and orphan elimination. Resources running with no purpose. Stopped instances paying for storage. Unattached volumes. Stale snapshots. Forgotten test environments. Typical savings: 8-15 percent of total spend.
Lever 5 - Architectural optimization. Workload redesign for cloud-native cost patterns. Containerization, serverless adoption, ARM-based instances, autoscaling tuning. Typical savings: 20-35 percent on the workloads that get the treatment.
Lever 6 - Contract negotiation. Enterprise agreements, negotiated rates, multi-product bundles. For enterprises spending more than $5M annually with a single provider, contract negotiation typically captures 10-25 percent additional savings.
Pulled in sequence, these six levers typically eliminate 40-60 percent of the 30 percent waste figure within 12 months. Pulled out of sequence, savings are typically half that.
The Visibility Foundation
Lever 1 is foundational because nothing else works without it. Three components define the visibility bar.
Allocation tags. Every resource tagged with team, product, environment, and cost center. Automated enforcement of tagging at provisioning. Untagged resources flagged and assigned weekly.
Showback or chargeback. Costs allocated back to teams or products with visibility that team leaders can act on. Showback (visibility without billing) is the minimum. Chargeback (actual cost transfer) is more impactful but politically harder.
Anomaly detection. Cost spike detection with alerts to responsible owners. The cost spike that nobody notices for two weeks compounds. The cost spike that triggers an alert within hours typically gets resolved within days.
The teams that get visibility right discover their actual waste is higher than they thought. The teams that skip visibility never know.
The Cultural Discipline
FinOps is part technical, part cultural. The technical part is well-documented. The cultural part determines whether the technical work actually happens.
Three cultural practices distinguish FinOps programs that sustain results from programs that produce one-time savings and then regress.
Cost accountability with teams. Engineering teams own their cloud costs as part of their work. Not a separate FinOps team's responsibility. Teams see their costs, decide on their tradeoffs, and explain their numbers. The accountability produces ongoing discipline.
Regular cost reviews. Weekly cost reviews at engineering leadership level. Quarterly cost reviews with finance. Annual cost reviews with executive leadership. The cadence catches regression before it compounds.
Cost gates in major decisions. New service adoption, architectural changes, major workload launches all include cost estimates and impact analysis. Cost becomes part of the design decision, not a surprise after deployment.
The teams without these practices reduce costs once and watch them grow back. The teams with these practices reduce costs and keep them down.
The AI Workload Twist
AI workloads have specific FinOps considerations covered in dedicated frameworks. The six-lever model applies but the specific moves differ.
Visibility for AI workloads requires per-feature, per-model, and per-tenant cost tracking that traditional cost dashboards do not naturally provide. The tooling has caught up in 2024-2025 with platforms like Helicone, Langfuse, and Portkey providing AI-specific cost visibility.
Right-sizing for AI workloads is model tier selection rather than instance size selection. The principle is the same; the implementation is different.
Reserved capacity for AI workloads is provider contract negotiation rather than reserved instance purchases. For enterprises spending more than $50K monthly on any single AI provider, negotiation is the right move.
The combination of traditional FinOps and AI-specific FinOps is what well-run enterprises operate in 2026. Single-discipline FinOps (traditional cloud only, AI only) misses material savings.
What This Costs
A serious FinOps program for a mid-market enterprise typically requires one to two dedicated FinOps engineers plus tooling investment of $50K-$300K annually. For enterprises above $10M annual cloud spend, the FinOps team typically pays for itself within one quarter and produces ongoing savings of 20-40 percent of cloud spend.
For smaller enterprises, FinOps can run as a portion of an engineering manager or finance partner role. The investment is smaller but the discipline still matters.
What Logiciel Does Here
Logiciel works with engineering and finance leadership on FinOps programs. The work is typically structured around the six-lever sequence with priority on whichever lever is the binding constraint on cost reduction for the specific environment.
The Cost-Performance Cloud Architecture framework covers the architectural side of cost optimization. The AI FinOps Framework framework covers the AI-specific FinOps discipline that sits alongside this broader playbook.
A 30-minute working session is enough to assess your current FinOps maturity and identify the highest-leverage lever to pull next.
Frequently Asked Questions
Should I build an internal FinOps team or use a third-party service?
Internal for sustained operations. Third-party for initial setup, episodic optimization, and specialized expertise. Most mature enterprises have both: an internal team for ongoing discipline and external partners for periodic deep audits.
How do I get engineering teams to care about cloud costs?
Showback or chargeback with team-level visibility. Cost accountability inside engineering performance frameworks. Recognition for cost-conscious architectural decisions. Engineering teams will optimize what they see and own. They will ignore what they do not.
What is the right FinOps tooling stack?
Cloud-native cost tools (AWS Cost Explorer, Azure Cost Management, GCP Cost Management) for baseline visibility. Third-party FinOps platforms (Cloudability, CloudHealth, Vantage, Zesty) for cross-cloud and advanced features. Most enterprises use a combination.
How do I prevent cost optimization from reducing performance?
Performance monitoring through the optimization work with explicit performance thresholds that cannot be crossed. Performance regression treated as a release blocker. The pattern that works is improving cost and performance together rather than treating them as opposites.
What is the right cadence for FinOps work?
Weekly tactical review. Quarterly strategic review. Annual contract and architecture review. The cadence catches regression before it becomes a quarterly problem. Sources: - Flexera, "2024 State of the Cloud Report" - FinOps Foundation, "State of FinOps 2024"
