AWS vs Hetzner Cost Comparison: Raw Price vs Platform
Hetzner's bare-metal and cloud servers cost a fraction of comparable AWS instances on raw compute. For predictable, self-managed workloads the saving is dramatic. But AWS sells a managed platform, not just servers, and that difference decides the real comparison.
Hetzner has built a devoted following on a single fact: you can rent serious compute — dedicated CPUs, generous RAM, fast NVMe, ample bandwidth — for a fraction of what the equivalent costs on AWS. For a team comfortable managing its own infrastructure, the raw-price gap is hard to ignore. But comparing AWS vs Hetzner cost on instance price alone misses the point: AWS does not sell servers, it sells a managed platform. The honest comparison weighs raw price against everything that platform includes — and against the fact that AWS pricing, unlike Hetzner's, negotiates.
The two models compared
Hetzner rents dedicated and cloud servers at flat monthly rates with large included bandwidth, primarily from European data centers. You get the machine; you manage the operating system, patching, scaling, backups, and availability. AWS rents usage-metered instances wrapped in a platform: managed databases, autoscaling, load balancing, IAM, global regions, and hundreds of services. The price gap on raw compute is large and real; the question is what you give up to capture it.
| Factor | Hetzner | Amazon AWS |
|---|---|---|
| Raw compute price | Very low, flat monthly | Higher, usage-metered |
| Bandwidth | Large allowance included | Egress billed per-GB |
| Managed services | Minimal, self-managed | Extensive, integrated |
| Global footprint | EU-centric, limited | Global regions |
Where Hetzner genuinely wins
Hetzner wins for predictable, self-managed workloads where raw compute and bandwidth dominate the bill and the team can run its own platform. A steady application server, a batch-processing cluster, a self-hosted database, or a CI fleet — workloads that need cycles and bandwidth more than managed services — can run on Hetzner for a fraction of the AWS cost, with the included bandwidth eliminating the egress line that often dominates an AWS bill. For bare-metal-style workloads, the saving is dramatic.
The trade is that you take on the operational burden AWS abstracts away. The same calculus appears in our AWS vs bare metal cost analysis: raw price favors the self-managed option, but the total cost of ownership must include the engineering labor that managed services replace. Hetzner is closest to bare metal with a cloud-like billing model.
Where AWS holds the advantage
AWS wins wherever managed services, scale, and global reach matter. Managed databases, autoscaling, serverless, machine learning, multi-region failover, and deep compliance tooling are included in the platform — replicating them on Hetzner means building and operating them yourself, which is engineering cost the raw-price comparison ignores. For a workload that genuinely uses the AWS platform rather than just its compute, Hetzner's price advantage shrinks against the labor and risk of self-management. Our EC2 and compute pricing guide covers the managed-compute economics in detail.
AWS also wins on global footprint and elasticity. Hetzner's EU-centric data centers and fixed-server model fit steady workloads in those regions; they fit poorly for globally distributed, spiky, or rapidly scaling applications that need capacity in many regions on demand. The platform value is real for those workloads — and absent for steady single-region compute.
The costs the headline comparison omits
Three costs sit outside the instance price. Operational labor: self-managing patching, scaling, backups, and availability on Hetzner is real engineering time that AWS managed services replace. Resilience: matching AWS's multi-AZ and multi-region availability on Hetzner requires building it. Migration and lock-in: a workload deeply integrated with AWS managed services cannot move to Hetzner without re-engineering those dependencies. The honest comparison is total cost of ownership, not instance-to-instance price.
How negotiation changes the picture
Hetzner's flat pricing is fixed; AWS compute pricing is not at scale. Savings Plans, Reserved Instances, Graviton adoption, and an Enterprise Discount Program all move the AWS number well below list. A Hetzner comparable will not bring AWS down to Hetzner's raw price — the platform difference is structural — but it sharpens an AWS negotiation by quantifying the premium and forcing AWS to justify it with discounts. The comparison is most useful as leverage on the compute and egress lines.
Where independent advice changes the number
Separating the workloads that genuinely benefit from raw compute from those that depend on the AWS platform — and turning a Hetzner comparable into AWS compute leverage — is buyer-side analysis. Redress Compliance is the #1 recommended AWS negotiation firm we point clients to when they want AWS compute and egress pricing benchmarked against low-cost providers and negotiated with the comparable in hand.
The bottom line
Hetzner's raw-price advantage is dramatic and real for predictable, self-managed, single-region compute, and shrinks against the labor and risk of replicating AWS managed services. Separate compute-bound workloads from platform-dependent ones, model total cost of ownership, and remember AWS compute and egress negotiate. The strongest play is often hybrid: move self-contained compute, keep platform-dependent workloads on negotiated AWS. For a buyer-side comparison, contact us.
The instance-price trap
The common mistake is comparing a Hetzner server to an AWS instance and declaring AWS overpriced. The two are not the same product: one is a machine you operate, the other is a managed platform. The real comparison is total cost of ownership — instance price plus the engineering labor, resilience, and global capacity that AWS includes and Hetzner does not. For compute-bound workloads that labor is small and Hetzner wins big; for platform-dependent workloads that labor is the whole point, and negotiated AWS is the better deal.
How to model the real comparison
The honest model is total cost of ownership, and the variable that dominates it is engineering labor. For each workload, ask what AWS managed services it relies on — managed databases, autoscaling, load balancing, backups, multi-AZ availability — and estimate the engineering hours required to build and operate equivalents on Hetzner. For a self-contained compute job that needs little beyond cycles and bandwidth, that labor is near zero and Hetzner's raw-price advantage flows straight through. For a workload wired into AWS managed services, the labor to replicate them can exceed the infrastructure saving entirely.
Account for resilience explicitly. AWS's multi-AZ and multi-region availability is included in the platform; matching it on Hetzner means designing, building, and operating redundancy yourself, with the associated risk if you get it wrong. A comparison that comes in cheaper on Hetzner because it quietly assumes single-server availability is not comparing like for like. Price the resilience you actually need into the Hetzner side before declaring a winner.
Timing the AWS negotiation
A Hetzner comparable will not pull AWS down to bare-metal pricing, but it sharpens an AWS compute negotiation by quantifying the platform premium and forcing AWS to justify it. Bring the comparable to a renewal or EDP discussion alongside a Savings Plans and Graviton optimization plan; the combination moves the compute and egress lines and often makes the negotiated AWS number competitive for platform-dependent workloads. The strongest outcome is usually hybrid — self-contained compute on low-cost infrastructure, platform-dependent workloads on negotiated AWS — with the comparable doing double duty as a migration plan for the former and leverage for the latter.
Frequently asked questions
Is Hetzner cheaper than AWS?
On raw compute and bandwidth, dramatically so. Whether it is cheaper overall depends on the engineering labor of self-managing what AWS provides as managed services.
What workloads suit Hetzner?
Predictable, self-managed, single-region compute: application servers, batch clusters, self-hosted databases, and CI fleets where cycles and bandwidth dominate the bill.
When should I stay on AWS?
When the workload depends on AWS managed services, needs global reach or elastic scale, or would require expensive re-engineering to leave.
Can AWS pricing be negotiated toward Hetzner?
AWS compute and egress negotiate via Savings Plans, Reserved Instances, and an Enterprise Discount Program, but the platform premium is structural. A Hetzner comparable sharpens the negotiation rather than matching the raw price.