How Azure Reserved Instances Work

Azure Reserved Instances (RIs) are a pre-purchase commitment for specific virtual machine types in specific Azure regions. When you reserve capacity for one or three years, Microsoft applies a discount — up to 72% off pay-as-you-go rates — to the matching VM usage that runs against your subscription. The discount is automatic: Microsoft matches reserved capacity to qualifying usage in the billing period. There are no coupons, no activation steps, and no change to how your workloads run.

The critical word is specific. An RI is tied to a VM series (e.g., D-series, E-series), a region (e.g., UK South, East US 2), and an instance size unless you enable instance size flexibility within a family. This specificity is what delivers the highest discount — but it is also what creates risk when workloads shift. If you reserve 50 × D4s_v5 in UK South and your team migrates half of those VMs to West Europe six months in, your RI sits unused. Microsoft does not give refunds for underutilised reservations beyond the 12% cancellation fee cap on early returns, subject to the $50,000 annual return limit. Our Azure Hybrid Benefit guide explains how layering Windows Server BYOL on top of RIs can push total discounts well above 72% for qualifying workloads.

RI purchases sit outside your Enterprise Agreement's Azure Monetary Commitment (AMC). They can be paid upfront (maximum discount) or monthly (slightly lower discount, same term). For enterprises with tight capital budgets, monthly RI payments are common — but always compare the total cost carefully, as the monthly surcharge is typically 3–5% over the upfront price across a three-year term.

Scope, Instance Size Flexibility and Centralised Purchasing

RIs can be scoped to a single subscription or shared across all subscriptions in a billing account. Shared scope is almost always the right choice for large enterprises: it maximises utilisation by pooling all matching usage across teams. Instance size flexibility — where one RI covers all VM sizes within a family, weighted by normalisation factors — adds further flexibility without sacrificing the full discount rate. Enable both settings unless you have a specific reason to restrict scope.

How Azure Savings Plans Work

Azure Savings Plans (ASPs), introduced in October 2022, take a fundamentally different approach. Instead of committing to a specific VM type, you commit to a minimum hourly spend on compute — expressed in USD per hour, regardless of VM series, region, or operating system. Microsoft then applies the Savings Plan discount to all eligible compute usage up to your committed spend. Usage above the commitment reverts to pay-as-you-go rates.

The flexibility is real and valuable: if your team moves workloads between regions, switches from D-series to E-series VMs, or runs Azure Kubernetes Service (AKS) containers, your Savings Plan covers the shift automatically. The trade-off is a lower maximum discount — up to 66% off pay-as-you-go for a three-year compute Savings Plan — versus up to 72% for a specific RI. For the MCA-E procurement teams we work with, Savings Plans are increasingly the preferred baseline commitment vehicle because workload composition changes significantly year-on-year.

There are two Savings Plan types: Compute (applies to VMs, AKS, Azure Functions Premium, Azure Container Instances) and Machine Learning (covers Azure Machine Learning compute). Compute Savings Plans are the default choice for general enterprise workloads. Download our Azure Cost Containment playbook for a full walkthrough of how to size your initial Savings Plan commitment without over-committing.

Assess Your Azure Commitment Discount Opportunity

Understand your optimal RI vs Savings Plan mix before committing spend. Our assessment benchmarks your workload profile against actual Azure discount data.

Start Free Azure Assessment →

Azure Reserved Instances vs Savings Plans: Side-by-Side

The two models are not mutually exclusive — Microsoft designed them to stack. The practical decision is how to allocate your commitment budget between them. The following comparisons govern most enterprise procurement decisions:

For most large enterprises running over £1M of annual Azure compute, the optimal structure is a Savings Plan covering 50–60% of baseline compute spend, with targeted RIs layered on top for the most stable, highest-spend workload families.

Need Help Sizing Your Azure Commitment Strategy?

Redress has advised over 500 enterprise clients on Microsoft Azure commercial strategy. We model your workload data against both commitment models and recommend the optimal split — before you sign anything.

Talk to an Azure Specialist

Combining RIs, Savings Plans and Azure Hybrid Benefit

The maximum Azure discount available to an enterprise is not either/or — it is all three applied simultaneously. Microsoft allows RIs and Savings Plans to stack with Azure Hybrid Benefit (AHB), which removes the Windows Server or SQL Server licence cost from the VM price before the RI or Savings Plan discount applies. The effective savings stack as follows:

In practice, enterprises with active Software Assurance on Windows Server running on Azure can achieve effective discounts of 80–85% versus fully paid Windows VMs on pay-as-you-go. Across a 500-VM estate running Windows Server, this translates to £2–4M annual savings without changing a single workload. The prerequisite is valid Software Assurance coverage — track expiry dates carefully, as AHB entitlements lapse with SA and the cost impact is immediate.

For SQL Server workloads, the combination of SQL RI and AHB SQL is even more powerful. A SQL Server Enterprise on a large VM series can carry a pay-as-you-go cost of over £40,000 per year per VM. With RI + AHB SQL applied, the same VM costs under £10,000 per year. To explore the MACC (Microsoft Azure Consumption Commitment) structure that governs how large enterprises lock in Azure pricing across their EA, see our guide on Azure MACC negotiation strategy.

Procurement and Negotiation Tactics for Maximum Savings

Microsoft's account teams have one incentive in RI and Savings Plan discussions: maximise your committed spend relative to your actual usage. You have a different incentive: commit only what your workloads will consume, at the lowest possible rate, with maximum flexibility to adjust. The following tactics reflect what we apply across 500+ enterprise Microsoft engagements:

Start with Savings Plans, then layer RIs. Sign a Savings Plan at 60–70% of your modelled compute spend. This gives you a safety net against over-commitment while capturing the majority of available discounts. Once you have six months of post-RI utilisation data, identify the stable workloads where specific RIs would add incremental discount and layer them selectively.

Use Azure Advisor data against Microsoft. Azure Advisor's reservation recommendations are generated from your own usage data. Bring these recommendations to your renewal negotiation — they confirm the discount opportunity and set the benchmark for your RI sizing conversation. If Microsoft's team proposes a commitment level above Advisor's recommendation, push back with your own data.

Negotiate RI flexibility terms into the EA. Standard RI terms do not allow cancellation beyond the $50,000 annual cap. For large enterprises with rapidly evolving cloud strategies, negotiate enhanced exchange rights or broader return allowances during EA negotiation. Microsoft does grant these commercially, particularly for first-time RI buyers or customers consolidating multi-cloud commitments. To book a confidential call with our Microsoft specialist team, use our online scheduler — we regularly recover $500,000+ in misaligned RI commitments for new clients in the first engagement.

Review utilisation quarterly. Unused RIs are silent cost drains. Set a 90-day review cadence via Azure Cost Management to track RI and Savings Plan utilisation reports. Any reservation running below 70% utilisation for two consecutive months should be exchanged or returned within the allowance window.