ETH validator hosting cost: AWS vs dedicated infrastructure (2026)

A single ETH validator workload costs $556 per month on AWS. The same workload fits on a $219 whole server, and costs less again when you pay only for the slice of a machine it actually uses. Same memory footprint, same storage growth, same 99.99% uptime target, same operational outcome at the chain.

Most cost comparisons stop at that headline. The real question is why each price lands where it does. And when each option is actually the right call for an operator.

The workload we are pricing

One ETH validator steady-state. 8 vCPU of sustained consumption. 32 GB of RAM, fixed by the execution and consensus client memory footprint. 2 TB of network-attached NVMe covering roughly a year of chain growth. Continuous network traffic at a low kbps sustained rate. A 99.99% uptime target because missed attestations and proposals cost money directly.

This shape is steady, not bursty. It is memory-bound, not CPU-bound. It is sensitive to latency jitter, not to peak throughput.

That matters because the pricing model of each provider is tuned for a different workload shape.

Side-by-side comparison

Prices are 12-month commitments in USD, for the same 8 vCPU, 32 GB RAM, 2 TB network NVMe workload, as published by each provider at the time of writing.

LinkPool is deliberately absent from the table. We quote managed validator infrastructure per workload rather than from a public rate card, and pricing for LP MKS, our self-service bare-metal Kubernetes platform, will be published when it launches. The rest of this article explains the pricing model it uses, because that is what moves the number.

Why each price lands where it does

AWS is optimised for bursty, general-purpose workloads. A validator runs at roughly 20-40% CPU steady-state and never bursts. The reserved instance price still reflects the compute-optimised SKU pricing that the platform is built around. You pay for flexibility that this workload will not use.

Hetzner and Latitude charge per server. The AX162 at $219 is the floor. You get the whole box regardless of whether the validator consumes 10% of it or 90%. Cheap if you have a single workload that happens to match the hardware. Expensive per watt of validator work if your workload is smaller, and capacity-constrained if it is larger.

Servers.com sells the smallest single-tenant server that fits, which for this spec lands at $453. Same whole-server logic, slightly different floor.

Component pricing charges for the resources the workload actually consumes: the 8 vCPU, the 32 GB, the 2 TB, and nothing else. The validator pays for its footprint, not for a whole server or for a general-purpose cloud SKU. For a single validator that is a fraction of a 224-vCPU host, so the number lands well under the whole-server floor.

A guaranteed-QoS tier is the same workload with CPU requests equal to limits, so the pod never yields to a burstable neighbour. For validators that cannot tolerate throttling during attestation windows, that is the tier to pick, and it costs more than the burstable equivalent because the capacity is reserved whether or not it is used.

The hidden line: operational overhead

The monthly price is not the full cost. Running your own dedicated infrastructure carries operational overhead. DIY validator operators typically measure it at $4,700 to $16,000 per month in engineering time. Server procurement, rack and stack, and network configuration are part of it. So are OS hardening, client upgrades, incident response, and on-call rotation.

Hetzner and Latitude hand you a server. Everything above the hypervisor is your team. AWS bundles a managed control plane but not managed validator operations. Managed Kubernetes with support included removes that overhead without charging for a full server you will not use.

Run the math at your own burdened engineering rate before comparing raw hardware prices. The $219 number on Hetzner often sits underneath $5,000+ in monthly ops cost for a single-validator shop.

Platform risk is part of the price

In November 2022, Hetzner banned over 1,000 Solana validators from its infrastructure. The workloads were shut down inside the provider's terms of service, not on the chain. Operators moved or lost consensus participation. The Solana Foundation responded publicly.

This is a real line item. Rented infrastructure sits under someone else's acceptable use policy. If the policy changes, the workload moves. Owned infrastructure and contracts with Web3-native providers carry a different risk profile on that axis.

SLA ceiling

An AWS customer cannot offer a higher uptime guarantee than AWS offers them. A validator operator running on AWS is bounded by the cloud SLA they sit beneath. Same for GCP, Azure, or any hyperscaler. Owned infrastructure sets its own SLA ceiling. LinkPool's 99.99% target comes from three availability zones in Manchester with failover across them. It does not come from a pass-through cloud contract.

When each option is right

AWS fits short-lived test workloads and testnet validators. Also teams that already run on AWS and want to consolidate billing.

Hetzner AX162 fits operators running exactly one validator per server, with no managed Kubernetes requirement. They need enough in-house ops to absorb platform risk.

Latitude and Servers.com fit teams that want dedicated hardware with a polished provisioning layer and are comfortable paying per server.

Component-priced managed infrastructure fits continuous validator workloads where managed Kubernetes, paying for the footprint rather than the server, and Web3-native support are the priority. Pick the guaranteed-QoS tier where any throttling during attestation or proposal windows is unacceptable.

For the architecture pattern behind the pricing, see bare-metal Kubernetes hosting: cloud vs dedicated infrastructure and we built bare-metal Kubernetes for Web3 before the enterprise vendors caught up.

Common questions

How much does it cost to host an ETH validator on AWS?

One ETH validator workload (8 vCPU, 32 GB RAM, 2 TB NVMe) costs approximately $556 per month on AWS using Reserved EC2 pricing. Data transfer costs are on top of that figure.

What is the cheapest way to host an Ethereum validator?

Per server, Hetzner's AX162 at $219 per month is the cheapest production-grade option, but you pay for the whole machine whether the validator uses 10% of it or 90%. Component-priced managed infrastructure charges for the 8 vCPU, 32 GB and 2 TB the validator actually consumes, which is why it undercuts whole-server pricing for a single validator.

Why is dedicated infrastructure cheaper than cloud for validators?

ETH validators have a stable, predictable workload: steady CPU consumption, fixed memory footprint, low network traffic. Cloud providers price for bursty workloads. Component pricing charges for the exact resources the validator uses, with no idle headroom baked in.

Is Hetzner a good option for ETH validator hosting?

Cost-competitive at $219 per month (AX162), but you pay for the whole server. In November 2022, Hetzner enforced its terms of service against 1,000+ Solana validators in a single policy change. That platform risk remains.

Run the number for your own workload

Pull your current invoice. Multiply your validator count by the number above that matches your provider. Then tell us the workload and we will quote the same spec on managed infrastructure. The delta is either noise or your Q3 budget. Which one is it?

Frequently asked questions

How much does it cost to host an ETH validator on AWS?

One ETH validator workload (8 vCPU, 32 GB RAM, 2 TB NVMe) costs approximately $556 per month on AWS using Reserved EC2 pricing. This includes compute and storage but does not include data transfer costs.

What is the cheapest way to host an Ethereum validator?

Per server, Hetzner's AX162 at $219 per month is the cheapest production-grade option, but you pay for the whole machine whether the validator uses 10% of it or 90%. Component-priced managed infrastructure charges for the 8 vCPU, 32 GB and 2 TB the validator actually consumes, which is why it undercuts whole-server pricing for a single validator.

Why is ETH validator hosting cheaper on dedicated infrastructure than cloud?

ETH validators have a stable, predictable workload: steady CPU load, fixed memory footprint, and low network traffic. Cloud providers price for bursty workloads and charge for headroom you don't use. Dedicated component pricing charges for the exact resources the validator consumes.

Is Hetzner a good option for ETH validator hosting?

Hetzner is cost-competitive at $219 per month (AX162 server), but you pay for the whole server regardless of utilisation. In November 2022, Hetzner enforced its terms of service against 1,000+ Solana validators in a single policy change. That platform risk has not been removed.