Cloud pricing comparison
Amazon Web ServicesvsGoogle Cloud Platform
AWS vs Google Cloud
Compute Pricing
Reference comparison for EC2 and GCP Compute Engine across general purpose, memory-optimized, compute-optimized, and GPU families. The table below is a non-authoritative snapshot; use the catalog for provenance-backed current rows.
Amazon Web Services
- Regions
- 30+
- Instance families
- 750+ types
- Pricing models
- On-Demand, Spot, RI 1yr, RI 3yr, Savings Plans
- Auto discounts
- Savings Plans only
Google Cloud Platform
- Regions
- 40+
- Machine families
- 500+ types
- Pricing models
- On-Demand, Preemptible, CUD 1yr, CUD 3yr
- Auto discounts
- Sustained-use (up to 30%)
Instance pricing comparison
Reference on-demand snapshot for us-east-1 / us-central1. Prices per hour before sustained-use discounts.
| Workload | vCPU | RAM | AWS | GCP | Winner |
|---|---|---|---|---|---|
General purpose 2 vCPU / 8 GB | 2 | 8 GB | $0.0832/hr t3.large | $0.0971/hr n2-standard-2 | AWS −14% |
General purpose 4 vCPU / 16 GB | 4 | 16 GB | $0.2016/hr m7i.xlarge | $0.1942/hr n2-standard-4 | GCP −4% |
Memory optimized 8 vCPU / 64 GB | 8 | 64 GB | $0.5040/hr r7i.2xlarge | $0.4740/hr n2-highmem-8 | GCP −6% |
Compute optimized 16 vCPU / 64 GB | 16 | 64 GB | $0.7140/hr c7i.4xlarge | $0.7672/hr c2-standard-16 | AWS −7% |
GPU — A100 (1×) | 12 | 85 GB | $32.7700/hr p4d.xlarge | $3.6738/hr a2-highgpu-1g | GCP −89% |
Reference-only rows for orientation, not live provider pricing. GCP sustained-use discounts may lower effective costs for persistent workloads.
Coverage & pricing models
| Capability | AWS | GCP |
|---|---|---|
| Regions | 30+ | 40+ |
| Instance types | 750+ | 500+ |
| Pricing models | 5 | 4 |
| Spot / preemptible | Yes | Yes |
| Committed use (1 yr) | Yes | Yes |
| Committed use (3 yr) | Yes | Yes |
| Sustained use discounts | No | Yes |
| Savings plans | Yes | No |
Feature availability
| Feature | AWS | GCP |
|---|---|---|
| Global load balancing | ||
| Managed Kubernetes (free CP) | ||
| Arm-based instances | ||
| Spot / preemptible instances | ||
| Committed-use discounts | ||
| Sustained-use discounts | ||
| Custom machine types | ||
| Savings plans (flexible) | ||
| Free tier (12 months) | ||
| Per-second billing |
When to choose each cloud
Choose AWS when
- You need the deepest catalog of managed services
- Your workloads use Savings Plans for flexible commitment pricing
- You run Windows or SQL Server workloads at scale
- Graviton Arm instances are a cost-optimization priority
- You need the broadest global region footprint for regulated workloads
Choose GCP when
- You want automatic discounts for long-running workloads (sustained use)
- Custom machine types let you right-size without paying for standard shapes
- Your ML/AI workloads benefit from Google's TPU and A100/H100 pricing
- You are a Google Workspace shop and want tight identity integration
- BigQuery and Dataflow are central to your data architecture
Open provenance-backed AWS and GCP pricing
Search, filter, and compare catalog rows with source badges and freshness metadata when normalized provider pricing has been imported.