Skip to content

Independent marketplace for verified cloud accounts

AWS vs Hetzner

For steady always-on compute the price gap between Hetzner and AWS is wide. This comparison works out what that gap buys, what it costs in operations, and when it closes.

Buy AWS Account
Broadest service catalog

Amazon Web Services

Regions noted in source
Dozens of global regions
Configurations listed
14
From$25 View account
Buy Hetzner Account
Germany, EU hardware value

Hetzner

Regions noted in source
Germany, Finland, United States, Singapore
Configurations listed
3
From$50 View account

There is a specific workload shape that makes Hetzner look unreasonable on paper: a fleet of servers that runs at a fairly constant load, twenty four hours a day, for years, with predictable traffic and no need for exotic managed services. For that shape the cost difference against Amazon Web Services is not a few percent. It is the kind of difference that changes what your company can afford to build.

The interesting question is not whether that gap exists. It does, and anyone who has run both knows it. The question is what the gap pays for, which parts of it you can actually capture, and at what point the saving is eaten by the engineering time required to keep the cheaper estate healthy.

The workload shape that favours Hetzner

Elastic billing is a hedge. You pay a premium per unit so that you can stop paying at any moment, and that hedge is worth a lot if your traffic triples on a Monday and halves in August. If your load is flat, you are buying insurance against a risk you do not carry.

Hetzner sells the opposite product: capacity by the month, dedicated hardware if you want it, with pricing that assumes the machine stays on. Batch processing with a steady queue, a CI fleet that runs all day, game servers, media encoding, an internal analytics cluster, a database replica that will exist for three years: these are the cases where the elastic premium buys you nothing. Our guide on predictable pricing versus pay as you go works through how to tell which category your workload is in before you commit to either.

What you are giving up, stated plainly

The saving is not free, and pretending otherwise leads to teams migrating in anger and migrating back six months later. AWS sells managed services, and the managed part is most of the value. When you move to Hetzner, several things that were somebody else’s problem become yours.

  • Database backups, failover and version upgrades, which you now schedule, test and own.
  • Load balancing and certificate renewal, configured and monitored by you.
  • Identity and fine grained permissions, which are far simpler and far less granular.
  • Object storage with a deep feature set, lifecycle policies and cross region replication.
  • Managed queues, managed caches, managed search and the rest of the catalog middle.
  • Automatic multi zone redundancy as a property rather than something you design.

None of this is impossible to run yourself. All of it is well understood open source. But it is a job, and the person doing that job costs more per year than a lot of infrastructure bills. If your team does not already have someone who enjoys this work, the saving is theoretical.

Bandwidth is the quiet half of the difference

Compute pricing gets the attention, but for many workloads the bandwidth posture is the bigger line. Hetzner includes substantial outbound transfer with its servers as a matter of product design. AWS meters outbound transfer carefully and it is routinely one of the top entries on a bill that came in over forecast.

This has architectural consequences, not just financial ones. On AWS you learn to keep chatty services close together, to cache hard, and to push bulk delivery through a content delivery network. Those are sound habits. They are also effort you can skip when transfer is effectively included. If you serve large files, stream video, or move data between your own services constantly, run that calculation before anything else, and use cloud egress fees explained as the model.

Comparing the operating burden, not the price list

Responsibility Amazon Web Services Hetzner
Database operations Managed service handles backups, patching and failover You install, back up, patch and test restores
Capacity changes API call, minutes, billed by the hour Provision a new machine, often monthly commitment
Redundancy across failure domains Availability zones built into most managed services You design it across locations yourself
Outbound transfer Metered and a common source of bill surprise Generous inclusive allowance by product design
Access control granularity Fine grained policies down to individual actions Simpler project and user level control
Tooling and integrations Assumed by most third party tools Supported, but you check compatibility first

Read that table as a staffing question rather than a feature list. Every row where Hetzner says “you” is a row that consumes engineering attention forever, not once. That is the real exchange rate between the two platforms.

Hardware access changes what you can build

Hetzner’s dedicated machine offering gives you whole physical servers at prices that make certain designs viable again. A database with a very large working set, an in memory analytics job, a build farm that is bound by local disk: these can all be cheaper and faster on one well specified physical box than on a cluster of virtual instances sized to the same workload.

The trade is replacement time. A virtual instance that fails can be replaced by an API call. A physical machine that fails involves a support ticket and a human. That is survivable if your architecture assumes it, and painful if it does not. Design for the failure first, then enjoy the hardware, and read designing for failure in the cloud before you put a single point of failure on cheaper hardware.

The split estate, which is what most teams end up doing

The useful pattern is rarely all of one. It is putting the steady, boring, bandwidth hungry tonnage on Hetzner and keeping the parts that genuinely benefit from managed services on AWS. Batch workers, build agents, media processing and non customer facing analytics move well. The control plane, the primary customer database, anything with a compliance story attached and anything that needs to scale sharply tend to stay put.

Done deliberately this is a strong setup. Done accidentally it becomes two estates with duplicate monitoring, two sets of credentials and an unclear boundary. Decide which side is authoritative for identity and for data, write it down, and keep the integration surface narrow. Our notes on when multi cloud helps are blunt about the cases where this is a mistake.

Make the data boundary the first thing you decide. If your primary database sits on one side and your application tier on the other, you have built a latency problem and a bandwidth bill at the same time. Keep each workload and its hot data in the same place, and let the link between estates carry only what genuinely needs to cross it.

When Hetzner stops being cheaper

The saving disappears in a few recognisable situations. It disappears when your traffic is genuinely spiky, because you end up provisioning for the peak and paying for it all month. It disappears when you rebuild three managed services badly and lose a weekend to a failed restore. It disappears when a compliance requirement forces you onto a platform with the paperwork already in place. And it disappears when your team is small enough that one engineer’s time is the scarcest resource you have.

It holds up when load is flat, when bandwidth matters, when you already run your own databases competently, and when the money saved is going into engineering headcount rather than just into margin. Our write up on right sizing cloud instances is worth reading first, because a correctly sized AWS estate is often much closer to the alternative than an oversized one.

Where to start

If you want to test the thesis, do it with one workload, not with a migration plan. Take the most boring steady service you run, put it on a single machine, run it in parallel for a month, and measure both the bill and the hours. You will learn more from that than from any spreadsheet.

When you are sourcing accounts, read a Buy Hetzner Account listing and a Buy AWS Account listing with the same questions: who holds the login, what payment instrument is attached, what limits apply to new resources, and what happens to support access after handover. If you are also weighing the other value oriented European options, Hetzner against OVHcloud covers that pairing directly.

Other comparisons to consider

AWS vs Vultr
vs

AWS vs Vultr

Some teams need many small locations and fast access to bare metal or GPU capacity more than they need a…

AWS vs Digital Ocean
vs

AWS vs Digital Ocean

How far does a small team get on DigitalOcean before platform complexity starts costing more than it saves? This comparison…

AWS vs IBM Cloud
vs

AWS vs IBM Cloud

IBM Cloud is built around OpenShift, regulated industries and hardware you cannot rent elsewhere. This comparison sets that platform thesis…

Google Cloud vs Hetzner
vs

Google Cloud vs Hetzner

Hetzner sells compute by the unit at a low price. Google Cloud sells a platform. The comparison is really about…

Background reading

How to Read a Cloud Account Listing
Cloud Accounts 7 min read

How to Read a Cloud Account Listing

Every line in an account listing is a claim that can be checked. This guide turns listing language into a set of…

How to Verify Cloud Credits Before and After a Transfer
Cloud Accounts 7 min read

How to Verify Cloud Credits Before and After a Transfer

Credit balances are provider-controlled, time-limited and service-restricted. This guide shows where each billing console displays them and how to verify a balance…

Reading Cloud Storage Performance Specifications
Infrastructure 7 min read

Reading Cloud Storage Performance Specifications

IOPS, throughput and latency are three promises, not one. How to read a storage spec sheet so the performance you get matches…

Decided which platform fits?

Both accounts are documented in the catalog, with configurations, credit balances and service limits listed before you buy.

Telegram