A2 family · Accelerator Optimizedx86_64 · Intel/AMD

a2-highgpu-2g

The a2-highgpu-2g machine type has 24 vCPUs and 170 GB of memory. Pricing for this instance starts at $7.35 per hour and $5363.14 monthly in the us-central1 region.

Updated June 7, 2026
vCPUs
24
Memory
170 GB
Network
Up to 32 Gbps
Storage
Local SSD

Save up to $2730/mo on every a2-highgpu-2g

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

All pricing options

us-central1 · USD · monthly = 730 hrs
Prices exclude local taxes
TermHourlyMonthlySavings
On-Demand
$7.3468$5363.14 Baseline
1-Year Committed Use Discount (CUD)
$4.6284$3378.74 -37%
3-Year Committed Use Discount (CUD)
$2.5714$1877.13 -65%
Preemptible / Spot
$3.6077$2633.62 -51%
You could save $2729.52/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
$5363.14
730 hrs × $7.3468/hr × 1

Compute

x86_64 · Intel/AMD
vCPUs i24
Memory i170 GB
Physical processorIntel Cascade Lake
Nested VirtualizationNot supported
Sole TenantNot supported
GPU2 GPU
Within A2 family
a2-highgpu-1g
12 vCPU
a2-ultragpu-1g
12 vCPU
a2-highgpu-2g
24 vCPU
a2-ultragpu-2g
24 vCPU
a2-highgpu-4g
48 vCPU
a2-ultragpu-4g
48 vCPU
a2-highgpu-8g
96 vCPU
a2-ultragpu-8g
96 vCPU
a2-megagpu-16g
96 vCPU

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

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

Networking

Max egress bandwidth32 Gbps
Tier 1 bandwidthN/A
Enhanced networking iYes (gvnic/virtio)
IPv6 supportYes (Dual-stack VPC)

Storage

Max persistent disks128
Max disk size257 TB
Local SSDsSupported (Scratch Disk)

Compare with another instance

vs.
Speca2-highgpu-2ga2-highgpu-1gΔ
vCPUs2412-50%
Memory170 GB85 GB-50%
Hourly Price$7.3468$3.6734-50%
Monthly Price$5363.14$2681.57-50%
Cost optimization

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

Connect your GCP account. We'll identify every underutilized a2.* instance and show exactly where downsizing, cleaning up orphaned disks, or shifting to CUDs saves money — with zero code changes.