isbaremetal

Servers that don’t
flinch at real traffic.

Truly unmetered uplink — no counter, at any port speed. Machines in Amsterdam and on both US coasts. You get root, IPMI and a port — nothing in between you and the wire.

traffic cap · every plan
100Gbps
max port · custom build
4
locations · ams · us-east · us-west
what’s in the box
specbare-metaltrafficunmetered · no counter, no overage
specbare-metalport1 / 10 / 100 Gbps · line rate, sustained
specbare-metaladdressing2 × ipv4 included · native ipv6
specbare-metallocationAmsterdam · Netherlands
specbare-metallocationMeppel · Netherlands
specbare-metallocationManassas, Virginia · US East
specbare-metallocationPhoenix, Arizona · US West
specbare-metalmachinesingle tenant · no hypervisor, no neighbours
specbare-metalaccessroot · administrator · your keys from hour one
specbare-metalout-of-bandipmi / kvm-over-ip · reboot and reinstall yourself
specbare-metalosany iso you bring · any kernel
specbare-metalsupportnetwork and hardware · os is yours
specbare-metalbillingflat monthly · the invoice does not move
terms of service →everything above, in writing
by subtraction
No meter.
No neighbours.
No upsell.

We built this brand by removing the things that make hosting feel like a subscription. What’s left is a server, a port, and your password.

Three things that push a workload off shared hardware.

None of them are failures of the cloud — they are how it is priced and scheduled. They just stop making sense once your traffic is steady and large.

01

Steal time you cannot see

On shared hardware the hypervisor decides when your process runs, and you get no view of the host it runs on. Throughput moves without your code changing, and the only honest answer to “why was p99 double last Tuesday” is that a neighbour was busy.

latency you cannot explain
02

Unlimited, with a footnote

The headline says unlimited; the terms say 95th percentile, fair use, or a transfer allowance measured per month. The limit is real and it applies exactly when the traffic matters — on the busiest day, not the quiet ones.

throttle on the day it counts
03

Egress is the actual price

The machine is cheap and the bytes leaving it are not. Per-gigabyte egress, overage tiers and premium-transit surcharges mean the invoice is a function of how well your product did last month.

a bill you cannot forecast

We’re picky about what’s in the box.
So you don’t pay for things you didn’t ask for.

+ what you get
  • dedicated bare-metal host no neighbours
  • unmetered port · 1 / 10 / 100 Gbps
  • full root · IPMI / KVM-over-IP
  • your own OS · any iso, any kernel
  • native ipv4 + ipv6 from our own space 2 addresses included
  • noc on call · network-level support
  • flat monthly invoice · no metered extras billed in the HostiServer panel
what you don’t get
  • os patching, package installs self-serve
  • cpanel / plesk / managed control panels the panel you do get is for invoices
  • app deploys, db tuning, devops on retainer
  • web-firewall rules, content moderation
  • migrations · backup orchestration
  • “unlimited” with footnotes no caps. period.

Need managed OS-level work? Our parent brand HostiServer does that — bare-metal stays out of the way.

Built for traffic that breaks normal hosts.

The five things that matter to an ops engineer, stated as numbers and limits rather than adjectives.

▌ headline · what people actually buy us for

True unmetered uplink.

No fair-use clause. No 95th percentile billing. No soft throttle. Saturate the port for 30 days straight — your bill is the same.

100Gbps
max sustained · single host
/mo
no monthly cap. ever.
  • ports of 1, 10, 100 Gbps
  • native ipv4 and ipv6 on every machine
  • no traffic counter anywhere in the stack
hardware

Enterprise iron only.

Xeon, EPYC, ECC, enterprise NVMe. No desktop CPUs dressed up as servers.

  • Intel Xeon · AMD EPYC
  • ECC across the fleet
  • Enterprise NVMe (no QLC junk)
access

Root, IPMI, KVM.

Full hardware control. Custom ISO boot. Reboot, reinstall, repartition on your schedule.

  • IPMI · serial-over-LAN
  • KVM-over-IP for hard cases
  • boot any iso · any kernel
footprint

Four sites, one network.

Carrier-grade facilities either side of the Atlantic, all announcing out of the same address space.

  • AMS · Amsterdam
  • MEP · Meppel
  • IAD · Manassas, Virginia
  • PHX · Phoenix, Arizona
support

Network-level NOC.

We answer routing, port and hardware questions. We don’t touch your OS. The line is honest and clear.

  • routing, port and hardware tickets
  • humans who read a traceroute
  • escalation to HostiServer for OS work

Four stock builds. No egress meter.

The same price in all four locations, 2 IPv4 addresses on every machine and $0 setup. Nothing on the invoice moves with your traffic.

Built for the traffic nobody else wants.

There is no clever filter behind this list — it is simply what an uncounted port and a whole machine are good for. If your workload is lawful and it moves bytes, it fits.

Amsterdam and both US coasts.

The same deal in every building: a port with no traffic counter on it, native v4 and v6, and hardware you hold the keys to. Pick the side of the Atlantic your users are on.

amsmepiadphx
AMS · NLAmsterdam1 / 10 / 100 Gbps · unmetered
MEP · NLMeppel1 / 10 / 100 Gbps · unmetered
IAD · US-EASTManassas, Virginia1 / 10 / 100 Gbps · unmetered
PHX · US-WESTPhoenix, Arizona1 / 10 / 100 Gbps · unmetered
same in every location
single-tenant hardware2 × ipv4 included · native ipv6root and ipmi from hour oneyour own os image
port sizes
1 Gbps10 Gbps100 Gbpsup to 100 Gbps per host

Locations and what ships with each machine on /network →

The questions people ask before they trust a port.

What’s the difference between VPS and Bare Metal?

A VPS is a slice of a shared physical host. CPU steal time spikes, IO is contested, and noisy-neighbour effects are invisible to you. Bare-metal removes all of that — you see the whole machine, not a guest view of it.

How truly “unmetered” is your bandwidth?

Literally unmetered: no fair-use clause, no soft 95th-percentile, no rate-limit that appears at month-end. Line rate on a 1, 10 or 100 Gbps port, sustained around the clock. Saturate it every day of the month and the bill does not move.

Do you protect against DDoS attacks?

No — not as a product. We do not sell mitigation, we do not advertise an absorption figure, and we will not promise you an outcome for a given attack. What exists is filtering at the network edge and monitoring that pages a human. If your workload attracts attacks and you need guaranteed mitigation, tell us before you order: that is a fair reason for us not to be the right host, and hearing it on day zero beats finding out on the day it matters.

How does support work?

Network and hardware tickets reach a human who can read a traceroute. OS-level questions are out of scope by design — we will point you at the documentation, or escalate to HostiServer if you want someone to do the work for you.

What uptime do you promise?

We do not publish a percentage, because a number without a measurement method and a credit schedule is decoration. What we do run: redundant power and cooling in every facility, multiple independent uplinks, and monitoring that pages a human. If an outage is ours, you will hear it from us first.

Pick a build, or describe the one you actually need.

The list price is a real price and the custom column is a real configurator — neither hides behind a qualification call. If you would rather just send a traceroute and let us work it out, that works too.