NP family · Accelerator Optimizedx86_64 · Intel/AMD

NP20s

The Azure NP20s (Standard_NP20s) is a virtual machine from the NP series, with 20 vCPUs, 336 GB of RAM and 1400 GB of temporary storage. On-demand pricing starts at $3.3000/hour in East US.

Updated September 3, 2026
vCPUs
20
Memory
336 GB
Network
Up to 8–16 Gbps
Storage
1400 GB Temp SSD

Save up to $1964/mo on every NP20s

Connect Azure — we'll find the optimal pricing model for your workload.

Pricing Overview

The Azure NP20s costs $3.30/hour ($2409.00/month) for Linux and $4.22/hour ($3080.60/month) for Windows in East US, as of 3 September 2026.

All pricing options

East US · USD · Linux OS · monthly = 730 hrs
Prices exclude local taxes
Term / TierHourlyMonthlySavings
Standard Tier — On-Demand
$3.3000$2409.00 Baseline
1-Year Reserved Instance
N/AN/A Baseline
3-Year Reserved Instance
N/AN/A Baseline
Spot Instance
$0.6098$445.18 -82%
You could save $1963.82/mo on this instance with the right pricing model.
Find my savings

Monthly cost estimator

Estimate your spend based on actual usage.
730 h
1h365h730h (24/7)
1
Estimated monthly
$2409.00
730 hrs × $3.3000/hr × 1

Compute

x86_64 · Intel/AMD
vCPUs i20
Memory i336 GB
Physical processorAmpere® Altra® ARM Processor
Nested VirtualizationSupported
Sole TenantSupported
GPUNone
Within NP family
NP10s
10 vCPU
NP20s
20 vCPU
NP40s
40 vCPU

Lock in this rate across your fleet — typically save 30–40%

Connect your Azure account to apply this pricing logic to every running instance automatically.

Networking

Max egress bandwidth8–16 Gbps
Tier 1 bandwidthN/A
Enhanced networking iYes (gvnic/virtio)
IPv6 supportYes (Dual-stack VPC)

Storage

Max persistent disks64
Max disk size32 TB
Local SSDsSupported (Scratch Disk)

Compare with another instance

vs.
SpecNP20sNP10sΔ
vCPUs2010-50%
Memory336 GB168 GB-50%
Hourly Price$3.3000$1.6500-50%
Monthly Price$2409.00$1204.50-50%
Cost optimization

Azure recommender alerts can cut this bill ~30-40%.

Connect your Azure account. We'll identify every underutilized NP.* instance and show exactly where downsizing, cleaning up orphaned disks, or shifting to reservations saves money — with zero code changes.