Skip to content

Sustainability and GreenOps

Last reviewed: August 2026

Cloud sustainability is a shared responsibility between vendor and customer. Vendors are responsible for data center efficiency (PUE) and the transition to renewable energy, while customers reduce unnecessary resource use through workload efficiency.

Vendor Tool Features
AWS Customer Carbon Footprint Tool Per-account carbon emissions dashboard. Breaks down Scope 1/2/3
Azure Emissions Impact Dashboard Integrates with Microsoft Sustainability Manager. Per-region carbon intensity
Google Cloud Carbon Footprint Per-project emissions. Publishes per-region carbon index (CFE%)
OCI Sustainability dashboard Per-region energy efficiency reports

Each region has a different power mix (share of renewable energy). Workloads with flexible latency requirements can choose regions with lower carbon intensity.

  • AWS — reported achieving its 100% renewable energy matching goal for operational power in 2023 (reported as sustained in subsequent years). Check per-account emissions with the Customer Carbon Footprint Tool
  • Azure — check Microsoft’s public sustainability disclosures for progress on renewable energy and carbon goals. Track emissions with the Emissions Impact Dashboard
  • Google Cloud — publishes per-region CFE (Carbon-Free Energy) percentages, with a goal of 24/7 carbon-free energy by 2030
  • OCI — check Oracle’s public disclosures for per-region renewable energy goals and status (e.g., EU regions). Cloud Advisor provides sustainability recommendations
Principle Method Effect
Right-sizing Eliminate overprovisioning. Scale down based on CPU/memory utilization Resource savings = energy savings
Scheduling Stop dev environments during off-hours Eliminates idle resources
Serverless/managed services Shared infrastructure increases density Higher energy efficiency than dedicated servers
Efficient instances Choose Arm-based (Graviton, Ampere, Tau) Lower power consumption for equivalent performance
Data retention policy Delete unnecessary data, move to cold storage Storage energy savings
  • Asynchronous processing — spreads out peak load to reduce maximum infrastructure capacity
  • Caching — reduces compute by eliminating repeated operations/queries
  • Data locality — placing data and compute in the same region to minimize network transfer
Metric Description Source
Carbon emissions trend per workload Monthly change in tCO₂e Vendor carbon dashboard
Carbon efficiency per cost $/tCO₂e Combination of cost and carbon reports
Idle resource ratio Share of instances with CPU utilization < 5% Monitoring tools
Compute time reduced by scheduling Savings from automatic shutdown outside business hours Automation logs
Data retention policy savings Storage reduced via lifecycle policies Storage reports

Scope concepts commonly seen in carbon emissions reports:

Scope Description Meaning in the Cloud
Scope 1 Direct emissions (own facilities) Not applicable to cloud users
Scope 2 Indirect emissions (purchased electricity) Vendor is responsible via data center power
Scope 3 Value chain emissions The user’s cloud usage falls here

Areas users can directly reduce: eliminating idle resources, right-sizing, choosing low-carbon regions, efficient architecture.

They have much in common, but not always:

Situation FinOps View GreenOps View
A low-carbon region is more expensive Choose the lowest-cost region Choose the lowest-carbon region
Spot instances Cost savings ✅ Repeated restarts can actually reduce efficiency
Removing idle resources Cost savings ✅ Carbon savings ✅ (same)
Right-sizing Cost savings ✅ Carbon savings ✅ (same)
  • Is this a workload that cannot change regions due to latency/SLA constraints?
  • Is this a workload, like a batch job, whose execution time can be adjusted?
  • Is the data retention period longer than regulatory requirements demand?
  • Does carbon optimization avoid compromising security, availability, or compliance?
  • Choosing a low-carbon region while ignoring latency requirements — selecting a region far from users, causing response times to exceed the SLA
  • Tracking GreenOps metrics without taking real action — building dashboards without following through on actions like removing idle resources or scheduling
  • Running FinOps and GreenOps as separate initiatives — most cost-saving activities are also carbon-saving, so this creates duplicate organizations/processes
  • Have you enabled the vendor carbon emissions dashboard (e.g., Customer Carbon Footprint Tool) and tracked monthly trends?
  • Have you applied scheduling to automatically stop dev/test environments during off-hours?
  • Do you prioritize Arm-based instances (Graviton, Ampere) for compatible workloads?