Skip to content
FinOps workspace
COST LEDGER / DECISION GUIDE

Azure reservations vs savings plan: compare coverage and commitment risk

Reviewed: 20 October 2026Get a quote for my project

Azure reservations vs savings plan is a choice between different coverage constraints and commitment risk. Compare the eligible workload, demand stability and flexibility you actually need before approving a financial obligation.

Illustrative rows of server cabinets in a large room lit with mint light
Illustration : cloud infrastructure.
On this page
  1. Azure reservations vs savings plan: Two commitments, two kinds of lock-in
  2. Reservation scope and the order in which discounts apply
  3. Exchange, refund and trade-in: the rules to check
  4. The sequence Microsoft recommends
  5. Choosing by workload pattern
  6. Data to gather before modeling
  7. Official sources used
  8. Frequently asked questions

Azure reservations vs savings plan: Two commitments, two kinds of lock-in

A reservation commits you to a specific resource type or family in a specific region. A savings plan commits you to an hourly spend on eligible services across regions. Microsoft currently offers a savings plan for compute and a savings plan for databases, each for one or three years.

  • Savings plan benefits are applied each hour to eligible usage in scope, starting with the usage that earns the highest discount. Unused commitment for an hour expires and does not roll over.
  • Savings plan discounts vary by product and by term, not by the size of the commitment.
  • Savings plan purchases cannot be canceled or refunded.
  • Microsoft states that reservations give the greatest savings when fully utilized, and that savings plans suit dynamic workloads or usage that moves between regions and services.

Reservation scope and the order in which discounts apply

A reservation can be scoped to a single resource group, a single subscription, a management group or the shared scope of the billing context. Azure applies discounts in that order, from the narrowest scope to the shared one. Changing the scope afterwards is not a commercial transaction and does not reset the term. A wrongly scoped reservation can leave one team's discount unused while another team pays full rate, so scope is worth a review as much as quantity.

Exchange, refund and trade-in: the rules to check

  • Reservations purchased after 1 February 2027 are not eligible for exchange when the service is supported by savings plans, which includes Virtual Machines, App Service and SQL Database. Reservations bought before that date keep one final exchange.
  • Refunds are capped: the total canceled commitment cannot exceed 50,000 USD in a rolling 12-month window for a billing profile or single enrollment, and the cap is per customer for CSP.
  • Microsoft does not currently charge an early termination fee on refunds and says it may do so in the future.

Existing reservations covering variable workloads may be eligible for a savings plan trade-in. Review the current exchange, refund and trade-in policy on the purchase date and keep the applicable terms with the approval.

The sequence Microsoft recommends

  1. Right-size first. Discounts reduce rates, not waste.
  2. Exchange underutilized reservations into configurations that fit current use.
  3. Trade in underutilized reservations for savings plans when usage is variable.
  4. Buy new reservations only for stable, well understood workloads.
  5. Buy savings plans last, sized to a clean baseline.

The Azure cost optimization guide shows where this sequence sits in a wider review.

Choosing by workload pattern

Workload patternLeans towardReason to check
Always-on VMs in one region with a fixed familyReservationUtilization must stay high for the whole term
VM families or regions likely to changeSavings plan for computeCompute plan does not cover software, networking or storage
SQL or Cosmos DB provisioned capacityReservation or savings plan for databasesCompare coverage lists, because serverless tiers are not covered by reservations
Mixed estate with containers and App ServiceSavings plan for computeConfirm each service appears in the included products price sheet
Illustrative server rack in the foreground with a row of cabinets behind it
Illustration only, not a customer's infrastructure.

Data to gather before modeling

Collect hourly usage by SKU family and region for at least a full business cycle, current reservation utilization, expiry dates and scopes, and the amortized cost view. Azure Advisor, the purchase experience in the portal and the Benefit Recommendation APIs all produce commitment recommendations, so ask the provider which one it used and why. Also note that savings plan prices can change on the first day of a month, and the current price is applied to active plans. A first model can be built with the commitment break-even calculator, and the guide on Savings Plans and Reserved Instances covers the AWS side of the same decision.

Want the commitment modeled on your own usage? Get a quote for this review.

Official sources used

Documents behind the statements above, as read in October 2026.

Frequently asked questions

Is the strongest discount always preferable?

No. Microsoft notes that a reservation gives its largest savings only when it is fully utilized. A lower rate on capacity you stop using costs more than a flexible plan that keeps being consumed.

Does a savings plan cover the entire Azure invoice?

No. The compute plan covers infrastructure cost of eligible compute services and excludes software, networking and storage. The databases plan covers the listed database services. Check each line of the invoice against the included products list.

Related workload decisions

Continue the review with azure vm cost. Each guide addresses a separate decision within the same cloud cost boundary.

Continue your review

Get support on my project

A specialist in the subject reviews your request and sends you a quote.

Get support on my project
Get my FinOps quote