Azure capacity reservations are an on-demand way to reserve virtual machine capacity in a specific region and availability zone, so the instances you need are guaranteed to be available when you deploy them. The crucial point for cost control is that a capacity reservation bills from the moment you create it, whether or not any VMs are running against it, because you are paying Azure to hold that capacity for you. They guarantee availability, not a discount, which makes them a useful but easily misused tool.
This article is part of our Azure cluster. For the wider account picture, start with the complete guide to Azure cost optimization, the pillar this piece links up to. Capacity reservations sit in the Lock step of our See, Cut, Lock, Run method: they secure capacity for workloads that cannot tolerate a deployment failure, and they need governance so you never pay for idle guarantees.
A capacity reservation guarantees that VMs of a given size will be available in a zone. It does not lower the rate on its own. The discount comes separately, from a reservation or savings plan that can apply on top. Keep the two concepts apart and you avoid the most expensive mistake.
What a capacity reservation actually does
When you create an on-demand capacity reservation, you specify a VM size, a region, an availability zone, and a quantity. Azure then sets aside that capacity for you, and it is yours to deploy into whenever you choose. If you deploy fewer VMs than reserved, the unused portion still bills at the equivalent on-demand rate, because the whole purpose is that the capacity is held and unavailable to anyone else. This is what distinguishes a capacity reservation from simply launching a VM: you are buying certainty that the hardware will be there, particularly valuable in constrained regions or for zonal deployments where capacity can be tight during demand spikes.
How it differs from reservations and savings plans
This is where teams get confused, so it is worth being precise. A reserved VM instance or a savings plan is a billing discount: you commit to a level of spend or usage for one or three years and pay a lower rate, but neither guarantees that capacity will be available when you deploy. An on-demand capacity reservation is the opposite: it guarantees capacity but carries no inherent discount. The two are complementary. You can place a capacity reservation to guarantee availability and apply a reservation or savings plan to the same capacity to bring the rate down, getting both certainty and discount together. The discount mechanics themselves are covered in Azure reservations versus Azure savings plan for compute.
| Instrument | Guarantees capacity | Lowers the rate | Commitment |
|---|---|---|---|
| Capacity reservation | Yes | No | None, on-demand |
| Reserved VM instance | No | Yes | 1 or 3 years |
| Savings plan | No | Yes | 1 or 3 years |
When a capacity reservation is worth it
Because you pay whether or not you use the capacity, a capacity reservation only earns its keep where guaranteed availability has real value. The clear cases are disaster recovery, where you need to be certain you can spin up replacement capacity in a target zone the moment a failover triggers; critical production workloads in capacity-constrained regions where a deployment failure would be unacceptable; and planned scale-up events, such as a launch or seasonal peak, where you want the capacity locked in ahead of time. For ordinary, flexible workloads that can deploy in any zone or tolerate a brief retry, a capacity reservation is usually unnecessary cost. The honest default is to use them sparingly, only where the availability guarantee is genuinely required.
Paying to hold Azure capacity you are not deploying into?
Our Azure cost audit finds idle capacity reservations, releases the ones no workload needs, and layers reservations or savings plans onto the ones that stay so you get the discount as well as the guarantee. On the performance model, you pay only from realized savings. No savings, no fee.
Book an Azure cost audit →The cost discipline they require
The risk with capacity reservations is exactly the same as with any always-on resource: they get created for a project, the project ends, and the reservation keeps billing for held capacity nobody is using. Treat them as governed inventory. Tag every reservation with an owner and a purpose, review them on the same cadence you review idle resources, and release any that no longer back an active workload, the discipline laid out in how to find idle and orphaned Azure resources. Right-size the quantity to what you actually deploy rather than a generous round number, and revisit the VM size if your workload has shifted families. Budgets and alerts should flag a reservation whose utilization drops so it surfaces before month end.
Putting it together
Used well, capacity reservations remove deployment risk for the workloads that cannot tolerate it, and combined with a reservation or savings plan they deliver both guaranteed availability and a lower rate. Used carelessly, they are a quiet drain that bills for capacity nobody deploys into. The line between the two is governance. Azure capacity reservation behavior, billing, and the way discounts apply on top can change, so verify the current rules against Microsoft's live Azure documentation before you build a strategy around them, and connect the discount side to your broader plan in Azure Cosmos DB cost control with RU/s and autoscale and the rest of the cluster.
The Azure Cost Optimization Field Guide includes our capacity-reservation review checklist and the commitment-layering pattern we deploy on engagements. It is the downloadable companion to this article.
The short version
Azure capacity reservations guarantee that VM capacity will be available in a specific zone, and they bill whether or not you deploy into them, which makes them distinct from reserved instances and savings plans that discount the rate but do not guarantee capacity. Use them for disaster recovery, capacity-constrained regions, and planned peaks; layer a reservation or savings plan on top for the discount; and govern them so you never pay for idle guarantees. When you want your Azure commitments and capacity arranged for both certainty and lowest cost, that is exactly what our Azure cost optimization service delivers.