← Back to Overview
Cloud Compute Pricing 2026: AWS EC2 vs Azure VMs vs GCP Compute Engine
Compute is 60-70% of most cloud bills. On-demand pricing for equivalent specs differs by less than 5% across providers. The real difference is in discount strategy, not list price.
General Purpose Instances
The most common instance type. US-East region on-demand pricing.
| Spec | AWS m7i | Azure D v5 | GCP n2-standard |
|---|---|---|---|
| 4 vCPU / 16 GB | $0.1664/hr ($121/mo) | $0.1660/hr ($121/mo) | $0.1671/hr ($122/mo) |
| 8 vCPU / 32 GB | $0.3328/hr ($243/mo) | $0.3320/hr ($242/mo) | $0.3342/hr ($244/mo) |
| 16 vCPU / 64 GB | $0.6656/hr ($486/mo) | $0.6640/hr ($485/mo) | $0.6684/hr ($488/mo) |
| 32 vCPU / 128 GB | $1.3312/hr ($972/mo) | $1.3280/hr ($970/mo) | $1.3368/hr ($976/mo) |
| 64 vCPU / 256 GB | $2.6624/hr ($1,943/mo) | $2.6560/hr ($1,939/mo) | $2.6736/hr ($1,952/mo) |
Monthly estimates based on 730 hours. On-demand pricing only.
Compute-Optimised (8 vCPU / 16 GB)
| Provider | Instance | On-Demand |
|---|---|---|
| AWS | c7i.2xlarge | $0.2856/hr |
| Azure | F8s v2 | $0.2720/hr |
| GCP | c3-standard-8 | $0.2944/hr |
Memory-Optimised (8 vCPU / 64 GB)
| Provider | Instance | On-Demand |
|---|---|---|
| AWS | r7i.2xlarge | $0.4368/hr |
| Azure | E8s v5 | $0.4480/hr |
| GCP | n2-highmem-8 | $0.4512/hr |
GPU Instance Pricing
The fastest-growing cost category. On-demand hourly rates for single-GPU instances.
| GPU | AWS | Azure | GCP |
|---|---|---|---|
| NVIDIA T4 | $0.526/hr (g4dn.xlarge) | $0.526/hr (NC4as T4 v3) | $0.35/hr (+ VM cost) |
| NVIDIA L4 | $0.8048/hr (g6.xlarge) | N/A | $0.7000/hr (g2-standard-4) |
| NVIDIA A100 (40 GB) | $3.06/hr (p4d) | $3.40/hr (NC A100 v4) | $2.93/hr (a2-highgpu-1g) |
| NVIDIA H100 | $32.77/hr (p5.48xlarge, 8 GPU) | $33.15/hr (NC H100 v5) | $31.22/hr (a3-highgpu-8g) |
ARM / Custom Silicon Comparison
ARM instances offer 20-40% better price-performance for compatible workloads (Linux, containers, most languages except .NET Framework).
| Provider | Instance Family | 8 vCPU / 16 GB | vs x86 Equivalent |
|---|---|---|---|
| AWS | Graviton4 (m7g) | $0.2608/hr | 22% cheaper than m7i |
| Azure | Cobalt 100 (Dpds v6) | $0.2720/hr | 18% cheaper than Dds v5 |
| GCP | Tau T2A | $0.2528/hr | 24% cheaper than n2-standard |
Spot / Preemptible Pricing Comparison
AWS Spot
Up to 90% off
- 2-minute interruption notice
- Dynamic pricing based on demand
- Spot Fleet diversifies across pools
- Best availability: us-east-1, us-west-2
Azure Spot
Up to 90% off
- 30-second eviction notice
- Set max price or use market rate
- Eviction policy: Stop/Deallocate or Delete
- Available on most VM sizes
GCP Spot / Preemptible
Up to 80% off
- Preemptible: 24-hour max, 30-second notice
- Spot: No time limit, dynamic pricing
- Simpler model than AWS/Azure
- Predictable 60-91% discounts
Region Pricing Variation
The same instance type costs 10-30% more in certain regions. Price-sensitive workloads should target low-cost regions.
| Region | AWS m7i.xlarge | Azure D4s v5 | GCP n2-standard-4 |
|---|---|---|---|
| US East | $0.1664/hr | $0.1660/hr | $0.1671/hr |
| Europe (London/Frankfurt) | $0.1864/hr (+12%) | $0.1860/hr (+12%) | $0.1838/hr (+10%) |
| Asia Pacific (Singapore) | $0.2029/hr (+22%) | $0.2070/hr (+25%) | $0.1954/hr (+17%) |
| South America (Sao Paulo) | $0.2330/hr (+40%) | $0.2240/hr (+35%) | $0.2173/hr (+30%) |