Blog Post

Azure Compute Blog
5 MIN READ

Announcing Public Preview of VM adaptive placement

LavanyaK's avatar
LavanyaK
Icon for Microsoft rankMicrosoft
Nov 17, 2026

Adding more flexibility into your deployments

Today at Microsoft Ignite, we're announcing the public preview of VM adaptive placement, a new way to create Azure virtual machines that helps increase the success rate of your deployment. 

With a standard VM create, you name one exact VM size, and the deployment fails if that size isn't available. With adaptive placement, you name a preferred size instead. Azure deploys your preferred size when it can. If the size isn't available in the region, Azure deploys a compatible alternative size with the same number of vCPUs and the same or newer generation. 

A new way to create a VM, not a new type of VM 

Adaptive placement is an optional operation to create VMs instead of standard VM create operation. It isn't a new VM type, and it isn't a new pricing offer. The VM you get is a regular Azure VM: you pay standard rates for the size that's deployed, and it supports the same lifecycle operations as any other VM. 

At a glance 

Standard VM create 

VM adaptive placement 

What you specify 

One exact size (hardwareProfile.vmSize) 

A preferred size (flexProfile.preferredVMSize) 

When that size has no capacity 

The deployment fails with an allocation error 

Azure deploys a compatible size with the same vCPU count 

The size you get 

The size you named 

Your preferred size or a compatible alternative 

What you pay 

Standard rate for the size you named 

Standard rate for the size Azure deploys 

How it works 

  1. You specify a preferred size. In the VM create request, set flexProfile.preferredVMSize (for example, Standard_D4s_v5) instead of hardwareProfile.vmSize. 
  1. Azure tries your preferred size first. If the region has capacity for it, that's the size you get. 
  1. If your preferred size can't be used, Azure selects a compatible alternative. The alternative has the same vCPU count, is the same or a newer generation (for example, v5 or later for a v5 preferred size), comes from the supported D-, E- and F-series families, and delivers performance equivalent to your preferred size. 

Adaptive placement works within your existing quota. Azure only considers sizes your subscription has quota for, so quota across related D-, E- and F-series families gives Azure more room to find capacity. Adaptive placement improves your chances of getting a VM when capacity is tight, but it isn't a capacity guarantee. 

When to use adaptive placement 

Adaptive placement is a good fit when you create individual VMs, your workload runs well on more than one size with the same vCPU count (for example, any 4-vCPU general purpose size), and getting the VM running matters more to you than landing on one specific hardware generation. It replaces the retry loops and hand-maintained fallback lists many teams build today with a decision Azure makes at the moment of your request, based on where capacity is available. 

Get started 

During the preview, you can create a VM with adaptive placement by using the REST API (API version 2026-04-01 or later), Azure PowerShell (Az 16.5.0 or later), the Azure SDK for .NET (Azure.ResourceManager.Compute 1.19.0 or later) or the Azure SDK for Python (azure-mgmt-compute 39.0.0 or later). Before you start, make sure you have quota in the target region for the family of your preferred size. 

To create a VM, include flexProfile in the request and leave out hardwareProfile.vmSize, because the two properties can't be used together. The rest of the request, including the image, OS profile and network settings, is the same as any VM create: 

PUT https://management.azure.com/subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.Compute/virtualMachines/{vmName}?api-version=2026-04-01 

{ 

  "location": "<region>", 

  "properties": { 

    "flexProfile": { 

      "preferredVMSize": "Standard_D4s_v5" 

    }, 

    "storageProfile": { ... }, 

    "osProfile": { ... }, 

    "networkProfile": { ... } 

  } 

} 

After the VM is created, a GET on the VM shows both your preference and the result. flexProfile.preferredVMSize is the size you asked for, and hardwareProfile.vmSize is the size Azure deployed: 

{ 

  "properties": { 

    "provisioningState": "Succeeded", 

    "flexProfile": { "preferredVMSize": "Standard_D4s_v5" }, 

    "hardwareProfile": { "vmSize": "<deployed-size>" } 

  } 

} 

You can also turn on adaptive placement for an existing VM by updating it to specify flexProfile.preferredVMSize instead of hardwareProfile.vmSize. That update can change the VM's deployed size. To switch back to a standard VM, send a PUT request that omits flexProfile and specifies hardwareProfile.vmSize. 

Pricing 

There's no additional charge for adaptive placement. You pay standard rates for the VM size Azure deploys, which can be your preferred size or a compatible alternative, so check the deployed size after creation to know which rate applies. Azure Reservations and other pricing benefits apply based on the deployed size, using the same rules as any other VM. For current rates, see Virtual Machines pricing (https://azure.microsoft.com/pricing/details/virtual-machines/). 

What's supported in the preview 

During the preview, your preferred size can come from the D-series (general purpose), E-series (memory optimized) or F-series (compute optimized) families. Other families, such as B, L, M, N and H, aren't supported during the preview. 

Adaptive placement is available in the following regions: centraluseuap, eastasia, eastus2euap, uksouth, australiacentral2, australiaeast, brazilsouth, canadacentral, francesouth, germanywestcentral, japaneast, koreasouth, northcentralus, northeurope, southafricawest, swedencentral, switzerlandwest, ukwest, and westcentralus. 

Keep these limitations in mind: 

  • Adaptive placement is for single VMs. For Scale sets , use instance mix instead 
  • Azure chooses the alternative sizes. You can't provide your own fallback list. 
  • Adaptive placement doesn't support VMs associated with Capacity Reservations, Azure Dedicated Hosts, Availability Sets or Proximity Placement Groups. 
  • You can't create a VM with adaptive placement in the Azure portal during the preview. 

Frequently asked questions 

Could an alternative size give me fewer vCPUs, an older generation or lower performance? 

No. Any alternative has the same vCPU count as your preferred size, is the same or a newer generation, and delivers equivalent performance. You can always see which size was deployed. 

What happens when I stop and start the VM? 

Stopping, deallocating and starting a VM doesn't reselect its size. The VM starts on the size it was deployed with, so the billed size doesn't change because of a stop and start. If you update flexProfile.preferredVMSize while the VM is deallocated, that update can select a different compatible size, and billing follows the newly deployed size. 

What if no alternative size is available? 

Azure deploys your preferred size, the same as a standard VM create. If that size also can't be deployed, the request fails with the usual allocation error. 

Are you attending Microsoft Ignite? 

Microsoft Ignite runs November 17–20, 2026, in San Francisco and online. To see adaptive placement in action, join breakout session BRK430, "Architecting your workloads for maximum compute flexibility." 

Learn more  

  • Create a VM with adaptive placement (preview) [add the Microsoft Learn link] 

VM adaptive placement is in preview and subject to the Supplemental Terms of Use for Microsoft Azure Previews (https://azure.microsoft.com/support/legal/preview-supplemental-terms/). Try it with your workloads and tell us what you think in the comments below. 

 

Updated Nov 10, 2026
Version 1.0