Dedicated RPC node hosting for Web3
Production RPC endpoints on dedicated Kubernetes, with bandwidth bundled and no per-gigabyte egress bill. The standard workload runs $194 a month, roughly a fifth of the AWS cost, with a 99.99% SLA in your contract.
LinkPool has operated Web3 infrastructure since 2017, at 5.7x cheaper than AWS on the standard RPC workload, against a 99.99% uptime SLA target, with 10 Gbps of dedicated network per pod.
The bill is really about bandwidth
RPC is the one Web3 workload where bandwidth drives the monthly cost, and bandwidth is the line hyperscalers price most aggressively.
Egress dominates the bill
A moderately loaded public endpoint serves multiple terabytes of JSON-RPC traffic a month. AWS bills that at $0.09 per GB. We bundle bandwidth as a flat monthly port, so egress is not a line item.
State keeps growing
A 4 TB archive-feature endpoint today is on track for 5 to 6 TB by the end of 2026. NVMe at scale is priced very differently on a hyperscaler than on hardware we own.
Read replicas are cheap here
Hyperscalers price read replicas aggressively and charge for traffic between zones. We run replicas on the same fabric with no per-replica licensing and no inter-zone egress.
One RPC endpoint, priced across providers
The same workload, 16 vCPU, 64 GB RAM, and 4 TB NVMe, across the providers operators actually compare. Monthly, at a 12-month commitment.
| Provider | Monthly cost | Pricing model |
|---|---|---|
| AWS EC2 | $1,112 | Per-resource, egress billed on top |
| Servers.com | $1,219 | Whole server |
| Latitude.sh | $720 | Whole server |
| Hetzner AX162 | $219 | Whole server, egress not bundled |
| LinkPool Standard | $194 | Component-priced, bandwidth bundled |
| LinkPool Performance | $335 | Guaranteed QoS, 10 Gbps per pod |
Built for endpoints that stay up
Dedicated Kubernetes
Managed clusters on dedicated hosts. No shared tenancy, and no hypervisor tax on JSON-RPC latency.
Guaranteed QoS on Performance
Requests equal limits. No throttling, no eviction, and 10 Gbps of dedicated network per pod.
Three availability zones
Endpoints fail over across three zones in Manchester, the location institutional buyers audit.
GitOps and a full audit trail
Every change is version-controlled and PR-reviewed. There is no interactive SSH login on our hosts.
What operators ask us
How much does dedicated RPC hosting cost?
The standard RPC workload (16 vCPU, 64 GB RAM, 4 TB NVMe) is $194 a month on our Standard tier at a 12-month commitment, or $335 on Performance. The same workload runs around $1,112 a month on AWS before egress.
Do you charge for bandwidth or egress?
There is no per-gigabyte egress fee. Bandwidth is a flat monthly port, up to 10 Gbps. On a busy public endpoint that is the difference between a predictable bill and a surprise one, because egress is the line hyperscalers price most aggressively.
Standard or Performance tier for RPC?
Standard is burstable and handles most RPC workloads well, because RPC CPU sits high under sustained load rather than spiking. Choose Performance when JSON-RPC tail latency matters or read-replica scaling pressure is unacceptable: institutional endpoints, large dApp backends, and archive-feature endpoints.
Which chains do you support?
Ethereum and EVM-compatible chains, with more available on request. Tell us the chain and the traffic profile and we will confirm what we can run.
Get pricing for your RPC workload
Tell us the chain, the traffic profile, and the scale. We reply with numbers, not a deck.