LearnGlossary

What is a vCPU? vCPU vs core in AWS and Azure

Updated 2 November 2026 · Published by Zoorik

A vCPU, or virtual CPU, is a processor as a virtual machine (VM) sees it: one thread of a physical core, or a whole core.

vCPU vs core: threads and hyper-threading

With simultaneous multithreading (SMT), also called hyper-threading, one CPU core runs more than one thread at once. Each thread is one vCPU.

Same 2 vCPUs: 1 core with 2 threads, or 2 cores AWS’s specifications for two EC2 sizes, each with 2 vCPUs and 8 GiB of memory, drawn side by side. The left box, m7i.large, on Intel Xeon Sapphire Rapids, holds 1 core with 2 shaded vCPU boxes inside it, and reads 1 core, 2 threads per core. The right box, m7g.large, on AWS Graviton3, holds 2 cores with 1 shaded vCPU box inside each, and reads 2 cores, 1 thread per core. A key says each shaded box is one thread, which the VM sees as one vCPU. Same 2 vCPUs: 1 core with 2 threads, or 2 cores m7i.largeIntel Xeon Sapphire Rapids2 vCPUs, 8 GiBcorevCPUvCPU1 core, 2 threads per corem7g.largeAWS Graviton32 vCPUs, 8 GiBcorevCPUcorevCPU2 cores, 1 thread per coreOne thread, which the VM sees as one vCPU
AWS’s specifications: each size has 2 vCPUs and 8 GiB. On m7i.large a vCPU is a thread, and on m7g.large it’s a whole core.

Some VMs give each vCPU a whole core, with one thread per core. AWS Graviton and AMD-based M7a instances do this, and so do Azure’s Arm-based Dpsv5 VMs.

vCPUs on AWS EC2 and Azure

On AWS, an m5.xlarge has 2 CPU cores with 2 threads each by default, so 4 vCPUs.

On Azure, a VM size name carries its vCPU count: Standard_D4s_v5 has 4. The Dsv5 series runs on hyper-threaded Intel Xeon processors, so each vCPU is a thread.

vCPU vs CPU, and what 1 CPU means in Kubernetes

The CPU is the physical processor in the server that hosts the VM. A VM’s vCPUs are a share of that CPU’s cores or threads.

In Kubernetes, 1 CPU is one physical core on a physical host, or one virtual core on a VM. So on AKS and EKS nodes, 1 CPU is 1 vCPU.

Pods can’t use all of a node’s vCPUs: system daemons use a share, so the node’s allocatable CPU is less than its capacity.

Why vCPUs matter for your cloud cost

  • EC2 On-Demand pricing lists each instance type’s vCPUs and memory beside its hourly rate, so you can compare the price per vCPU. Prices change.
  • Some fees count vCPUs. On EC2, Red Hat Enterprise Linux (RHEL) is priced per vCPU-hour. License-included Windows and SQL Server fees follow the vCPUs you configure.
  • For software licensed per core, you can keep the memory and run fewer vCPUs. AWS CPU options and Azure constrained vCPU sizes, such as Standard_E32-16s_v5, do this at the same VM price.
  • The Kubernetes scheduler places pods by their CPU requests, not their use. Requests above real use hold vCPUs that sit idle, so the cluster needs more nodes.
  • Coral rebalances CPU and memory requests on AKS and EKS to match real use, and drains spare nodes so your autoscaler can remove them.

Read next: Kubernetes allocatable vs capacity: what pods can use

Sources: AWS, “CPU options for Amazon EC2 instances” (opens in a new tab); AWS, “Rules for specifying CPU options for an Amazon EC2 instance” (opens in a new tab); AWS, “Specifications for Amazon EC2 general purpose instances” (opens in a new tab); AWS, “Amazon EC2 On-Demand Pricing” (opens in a new tab); Microsoft Learn, “Dsv5 size series” (opens in a new tab); Microsoft Learn, “Dpsv5 size series” (opens in a new tab); Microsoft Learn, “Azure VM sizes naming conventions” (opens in a new tab); Microsoft Learn, “Constrained vCPU sizes” (opens in a new tab); Kubernetes, “Resource Management for Pods and Containers” (opens in a new tab).

Find your own number

One line in your cloud console. One PDF report with your estimated saving at list price.

Estimate your saving
Talk to the builders

Thirty minutes with the people who build Coral.

Book a demo (opens in a new tab)

Or write to founders@zoorik.com