AWS alternatives fall into two tiers: hyperscalers (Google Cloud, Azure, Oracle Cloud) with comparable service breadth, and predictable-pricing providers (DigitalOcean, Hetzner, OVH, Linode) with simpler pricing and fewer services. The team that picks a hyperscaler alternative stays in the cloud-services world; the team that picks a smaller provider trades service breadth for cost predictability.
Table of contents
- Hyperscaler tier
- Predictable-pricing tier
- Comparing compute costs
- When to stay on AWS
- When to pick GCP
- When to pick Azure
- When to pick a smaller provider
- Multi-cloud
- FAQ
Hyperscaler tier
Google Cloud (GCP): Best data (BigQuery), ML (Vertex AI, TPUs), Kubernetes (GKE Autopilot). Sustained-use discounts automatic.
Microsoft Azure: Best Microsoft stack (Active Directory, Office 365), hybrid (Azure Arc). Free K8s control plane.
Oracle Cloud (OCI): Best Oracle database workloads, aggressive pricing, strong EU presence.
Alibaba Cloud: Best APAC, especially China market.
Predictable-pricing tier
DigitalOcean: $4-8/mo for small VMs, simple UI, predictable bills. Great for SMBs and developers.
Hetzner: German provider, best EU price/performance. Limited Americas/APAC presence.
OVH: French, global bare-metal and dedicated servers. Good for predictable VM costs.
Linode (Akamai): Simple, predictable. $5-10/mo small VMs.
Vultr: 32+ locations, hourly billing. Good global coverage.
Comparing compute costs
8 vCPU, 32GB RAM, 500GB SSD, 5TB egress:
- AWS EC2 on-demand: ~$280/mo
- AWS EC2 with 1yr RI: ~$180/mo
- GCP sustained-use: ~$200/mo
- Azure D8s v5: ~$280/mo
- DigitalOcean: ~$160/mo
- Hetzner CCX63: ~$60/mo
- OVH B2-60: ~$80/mo
The team that runs steady-state at scale finds 50-70% savings on the smaller tier.
When to stay on AWS
- Need specific AWS services (Lambda, SageMaker, ECS, etc.).
- Team has AWS expertise.
- Compliance (FedRAMP, HIPAA-eligible services).
- Need broadest service catalog.
- Existing AWS investment.
- Deepest enterprise relationships.
When to pick GCP
- BigQuery or Vertex AI is primary workload.
- ML at scale (TPUs).
- Multi-region low-latency (premium network).
- Kubernetes at scale (GKE Autopilot).
When to pick Azure
- Microsoft stack is your IT backbone (Active Directory, Office 365).
- Hybrid with on-prem (Azure Arc).
- Windows workloads dominate.
- Free K8s control plane matters.
When to pick a smaller provider
- Predictable pricing matters more than service breadth.
- Workloads fit on simple VMs (web apps, dev environments).
- Lower ops overhead.
- Cost-sensitive startups.
The team that has simple workloads on a smaller provider saves significantly. The team that needs managed services stays on a hyperscaler.
Multi-cloud
Some teams use multiple clouds:
- GCP for BigQuery + data pipeline
- AWS for the rest of the stack
This adds complexity (IAM, networking, ops expertise in both) but can use each cloud’s strengths. The team that has a specific reason for multi-cloud does it intentionally; the team that defaults to multi-cloud has unnecessary overhead.
FAQ
Which is the cheapest cloud provider?
Hetzner, OVH for compute. For managed services with similar pricing, the comparison varies. The team that runs simple workloads finds 50-70% savings on smaller providers.
Who is AWS’s biggest competitor?
Microsoft Azure by market share (~25% vs AWS ~31%). Google Cloud close behind on technical merit.
Can I use multiple clouds at once?
Yes (multi-cloud). Adds complexity. The team that uses GCP for BigQuery + AWS for the rest is a common pattern.
Should I migrate away from AWS?
The team that has specific reasons (cost, performance, missing service, compliance) migrates. The team that is happy with AWS doesn’t. Migration is expensive and risky.
What about open-source alternatives?
OpenStack (self-hosted IaaS), Kubernetes (orchestration), MinIO (S3-compatible storage). The team that runs on-prem with these has full control; the team that wants managed services stays on a hyperscaler.
If you are sizing the infrastructure for the kind of project this post covers, the RunxBuild hosting calculator is the right place to model the line items. The compute, the memory, the storage, the bandwidth, the database - each one is a separate number, and the team’s mental model for the platform is the sum of those numbers. The RunxBuild dashboard is where the team sees the actual usage in one place.
Useful related references: