Bare-metal email servers, your MTA, our operational layer
For teams that already have email-infrastructure expertise and don't want to be told which MTA to run. Real Intel Xeon CPUs (no vCPU, no hypervisor noisy-neighbour problems). Ten to thirty dedicated IPs per plan. Bring your own Postfix, Exim, KumoMTA, or your own PowerMTA license — we handle everything around that. Port 25 open by default. Reverse DNS configured to your sending domain. Clean IP allocations rather than recycled blocks. And an operator that treats a Spamhaus listing as a Tuesday afternoon problem to investigate, not as grounds for instant cancellation.
Hardware for teams who already know what they're doing
This is a different product than the managed services on the rest of the catalogue. It's for teams that already have email expertise: a sysops engineer who knows Postfix down to the routing layer, an operator who has run Exim for a decade, a developer ready to deploy KumoMTA themselves, an agency bringing its own PowerMTA license to a host that will leave them alone to run it. The customer brings the MTA expertise. We bring the dedicated hardware, the clean IPs, the open port 25, the reverse DNS configured to the sending domain, and the operator who treats Spamhaus troubleshooting as part of running email rather than as a reason to suspend the account.
The positioning matters because the dedicated-server market falls into two unhelpful extremes. On one side, the generic bare-metal providers — Hetzner, OVH, Leaseweb, smaller shops — sell hardware at competitive prices but treat email as a liability. Port 25 sometimes blocked. Reverse DNS hard to configure properly. Acceptable Use Policy that suspends accounts on the first abuse complaint, no triage. On the other side, the full email-platform providers (SendGrid, Mailgun, Postmark) handle the operations end-to-end but force you onto their software stack, their templates, their event model, their per-thousand pricing. People searching for a dedicated email server or a dedicated SMTP server are usually looking for the middle ground that almost nobody actually offers: hardware that's truly yours, IPs that no one else sends from, and a host that understands email as a workload instead of a problem.
What's in every plan
Real CPU, not vCPU
The Standard ships with a Xeon E-2388G — 8 physical cores, 16 threads — allocated entirely to your server. No hypervisor in the path, no shared cores, none of the noisy-neighbour variance that creeps into virtualised offerings.
Dedicated IP block
10, 20, or 30 IPs depending on plan, allocated from clean address space. One sending domain per IP — the discipline that's the whole point of dedicated IPs in the first place.
Port 25 open from day one
Outbound SMTP works the moment the server is provisioned. No "request port 25 unblock" ticket queue, no manager approvals, no waiting period before you can actually use what you bought.
Reverse DNS to your sending domain
PTR records configured to whichever hostname you specify. Forward-confirmed reverse DNS (FCrDNS) properly aligned — which is what the serious receivers actually check before accepting your mail.
Your choice of MTA
Postfix, Exim, KumoMTA, your own PowerMTA license if you bring it, MailerQ, Sendmail, Haraka — install whatever fits the operation. No platform lock-in to a stack we'd prefer.
Operator tolerance for the messy parts
The team understands that bounce reports, occasional FBL spikes, and Spamhaus listings happen even to legitimate senders. We help diagnose and remediate rather than suspending on the first complaint.
EU data centre, default
Sweden and Germany by default. EU data residency simplifies the GDPR posture for European data subjects. US and APAC are available on Custom plans for teams with specific latency or regulatory needs.
Hardened Linux baseline
AlmaLinux 9 by default; Rocky, Debian, Ubuntu LTS available. Hardened SSH config, fail2ban, automatic security updates, no preinstalled bloat. The system is yours; we make sure it doesn't ship insecure.
Three plans plus a Custom quote
The price reflects the actual hardware you get plus the email-specific operational layer wrapped around it — clean IP allocation, reverse DNS configured properly, FBL registration assistance, an operator who treats Spamhaus listings as triage work rather than as cancellation grounds. The specs below describe physical CPUs allocated entirely to your server. Not vCPU equivalents, not shared cores, not "comparable performance" — actual silicon.
Note that no daily email volume cap appears in the plan specs. That is deliberate. The volume a dedicated server can sustainably send is determined by your sending domain reputation, your IP warmup discipline, your list hygiene, your complaint-rate management and your authentication posture — not by a number we'd write into the plan. A well-run sending program on the Standard plan can comfortably move hundreds of thousands of messages per day; the same hardware run carelessly can struggle to deliver a fraction of that. Annual prepayment qualifies for a 10% discount across all three published plans.
Why €995, when Hetzner sells a Xeon dedicated for €99?
Generic dedicated server providers sell raw hardware at competitive prices because they are explicitly not in the email-operations business. Their AUP treats outbound email as a liability — port 25 may be open or may require a request, reverse DNS configuration is sometimes self-service and sometimes restricted, and the moment a Spamhaus listing or a Microsoft block lands, the typical response is account suspension and a ticket back to the customer telling them to "fix it." For a team that already runs email infrastructure, the cheap hardware is fine right up until the first real problem, at which point the absence of an operations partner who understands email becomes expensive. The Authorize Hosting price reflects what's actually included beyond the bare hardware: clean dedicated IPs from address space we manage for sending reputation, reverse DNS configured properly to your domain, FBL registration assistance, Spamhaus delisting support when legitimate sending hits bumps, and an operator that has been running this kind of work since 2003. If the team's email expertise is deep enough that they only need raw hardware, Hetzner is genuinely the right answer. If the team wants a hosting partner that understands the difference between sender having a bad day and a sender abusing the platform, this is what that costs.
How this compares to the generic dedicated providers
The fair comparison is on operational philosophy, not on raw hardware specs (which are roughly similar across the segment). Each of the alternatives below makes sense for a different kind of buyer:
Dedicated email server hosting compared, April 2026
Provider
Hardware pricing
Port 25 status
Reverse DNS
Email operational support
Operator tolerance
Authorize Hosting Standard
€995/mo (8C/16T physical)
Open by default
Configured to your domain
FBL, Spamhaus assistance, list cleanup add-on
High — help, not cancellation
Hetzner dedicated
From €39/mo (variable specs)
Open by default
Self-service in panel
Out of scope
Strict AUP enforcement
OVH bare metal
From €60/mo (variable specs)
Open by default
Configurable
Premium support tier upsell
Strict; queue-based hardware replacement
Leaseweb
From $80/mo (entry tier)
Open with KYC
Configurable
Out of scope
Standard AUP enforcement
DigitalOcean droplet (large)
From $96/mo (8C, vCPU)
Blocked, request required
Hostname-based
Out of scope
Strict; vCPU not equivalent
AWS EC2 (m6i.4xlarge)
~$560/mo on-demand
Blocked permanently
Configurable in Route 53
Out of scope without paid support
Strict; SES is the AWS path
Google Cloud / Azure VM
Variable
Blocked permanently
Configurable
Out of scope
Outbound email actively discouraged
The shape is consistent across the segment: generic infrastructure providers sell hardware at competitive prices on the assumption that you handle the email operations entirely yourself. The cloud hyperscalers — AWS, GCP, Azure — actively discourage outbound email by blocking port 25 permanently and treating any unblock request as something close to a hostile act. Our bundle sits in the gap: bare-metal control like a Hetzner-class dedicated server, with the email-aware operational layer that the generic providers explicitly don't sell.
You bring the MTA. We do everything around it.
Dedicated Email Server architecture — your software stack on our infrastructureThe product is the hardware plus the email-aware operational layer underneath. You choose the MTA. You configure the DKIM keys. You set the throttling rules. You build the bounce handling. You bring whatever sending discipline your team prefers. We provide the physical server, the dedicated IPs with clean reputation, port 25 open from day one, reverse DNS configured properly, FBL registration assistance, and Spamhaus remediation help when the occasional bump happens. Each side does what it does best.
The shapes that fit this
Teams running their own MTAPostfix shops, Exim operators, KumoMTA early adopters, agencies bringing their own PowerMTA license. The hardware comes clean; you install what you know.
ESPs building a sending platformEmail Service Providers building their own product on bare-metal infrastructure. Multiple servers in cluster on Custom plans for redundancy.
SaaS platforms with sending-as-a-productMulti-tenant SaaS sending on behalf of customers, where the platform team wants direct control over the sending stack rather than wrapping a third-party API.
Compliance-driven sendingHealthcare, finance, government workloads where data residency, sending domain control and direct operational ownership matter for regulatory posture.
Newsletter operators with serious volumeIndependent newsletters, media companies, Substack-class operators who want to own the sending stack rather than rent it.
Migrating from generic VPS providersTeams who started on Hetzner or DigitalOcean and ran into the limits of generic-hosting AUP when their sending program scaled. The migration target is the same hardware quality with email-aware operations.
Recovering from blocklist incidentsTeams whose previous IPs got Spamhaus-listed or Microsoft-blocked and want to start fresh on clean dedicated IPs with operational support during the recovery and warmup.
Specialized sending workflowsCustom MTAs (Haraka, Sendmail, in-house code), unusual queueing requirements, integration with proprietary CRM stacks. Dedicated hardware leaves room for whatever the team builds.
Operational add-ons, when you want them
A handful of services that experienced sending teams sometimes want are available as paid add-ons rather than rolled into the base price. The reasoning is the same across all of them: customers who buy dedicated email servers usually have strong opinions about which parts of the operation they want help with and which parts they want to own. Bundling everything would force everyone to subsidise services some teams don't need.
Database list cleanup
Syntax validation, MX validity checks, role-account flagging, known spam-trap pattern detection, engagement-history pruning. Priced by list size and depth of analysis. Pays for itself when an old database is the cause of complaint-rate problems.
Initial MTA setup assistance
If you want help installing and tuning Postfix, Exim, KumoMTA or another MTA on the new server, our team can do the setup. Pays off when the time saved exceeds the engagement cost.
Spamhaus delisting assistance
Included on a best-effort basis; structured remediation engagements are available as paid work when the listing is severe and the customer wants direct guidance through the Spamhaus communication process.
FBL registration support
Registering your dedicated IPs with AOL, Yahoo and Hotmail feedback loops, configuring complaint processing endpoints. Available as one-time setup or ongoing.
Custom warmup planning
For new dedicated IPs going into production, a structured 14-to-21-day warmup schedule tailored to your sending profile and target receiver mix.
Migration assistance
Help moving from a previous provider — DNS planning, parallel sending coordination, IP transition, DKIM key migration. Available on time-and-materials basis.
What dedicated servers won't do for you
Worth being explicit about. Hardware doesn't fix sender reputation problems caused by bad list hygiene — that's still a list problem, not something a faster CPU helps with. Hardware doesn't make cold email outreach legitimate; cold sending runs on a different operational profile that needs Cold Email Infrastructure with continuous warming and active monitoring. Operator tolerance applies to legitimate programs that occasionally hit bumps — bounce spikes, an unexpected Spamhaus listing, an FBL surge after a content change — not to intentional spam, harvested lists, or sending that should never have started. And dedicated hardware won't replace the product-side decisions about content quality, list permission, and engagement strategy. The hardware can support whatever sending profile you've built. It can't substitute for building it correctly in the first place.
Migrating in from generic hosting or another email-specialist host
Most teams arriving here are migrating from one of three places: a generic dedicated provider (Hetzner, OVH, Leaseweb) where the AUP started causing friction once sending volume grew; a cloud VPS (DigitalOcean, Linode) where the lack of real CPU and the port 25 restrictions hit a wall; or another email-specialist host whose service quality or pricing took a turn. The pattern is consistent in all three cases: we provision the new server, install your preferred MTA, allocate clean dedicated IPs, configure reverse DNS, and run a 14-to-21-day parallel-sending warmup before the full cutover.
The specifics differ by source. From a generic provider, the main concern is reverse DNS and IP reputation — if your old IPs were already in good standing and the move is purely about operational fit, the new IPs still need their own warmup, but you're not fighting damaged reputation. If the old IPs already had problems, starting fresh on clean allocations is almost always better than trying to bring damaged IPs forward. From a cloud VPS, the visible wins from day one are usually the real CPU and the absence of port 25 friction. From another email-specialist host, your pmta.conf, exim.conf, main.cf or kumomta config can usually be brought forward with minor adjustments for the new IP block.
Common questions about Dedicated Email Servers
FAQ
The questions that come up most often before someone signs
What's the difference between dedicated email servers and a regular dedicated server?
Operationally, almost everything. A regular dedicated server typically runs with port 25 blocked, no PTR record control, no allocated IP block specifically for sending, and an AUP that treats outbound email as a problem to handle by suspending accounts. A dedicated email server is provisioned for sending from day one: port 25 open, reverse DNS configured to your sending domain, dedicated IPs from clean address space, and an operator who understands that bounce reports and occasional spam complaints are part of running outbound email — not a reason to cancel the contract.
Why don't you preinstall an MTA?
Because the customers buying dedicated email servers usually have strong opinions about their MTA. Some teams want Postfix because they know it cold and have ten years of running it. Some want Exim because they came up through cPanel and the syntax fits their head. Some want KumoMTA because they like Lua scripting and not paying for a commercial license. Some are bringing their own PowerMTA license. Picking one for everyone forces a choice that should belong to you. We provision the hardware clean — Linux, hardened, port 25 open, IPs allocated, rDNS configured — and let you install whatever fits your operation. Setup assistance is available as an add-on if you want it.
Why "physical CPU" instead of vCPU?
Because they are physical CPUs. The Standard plan ships a Xeon E-2388G with 8 physical cores and 16 threads, allocated entirely to your server. There is no hypervisor sharing those cores with anyone. Cloud VMs typically describe capacity in vCPU, which is one thread of one physical core, often shared with neighbouring tenants. An "8 vCPU" cloud instance and a Xeon E-2388G dedicated server are not the same product — the dedicated server has roughly twice the actual compute, with no noisy-neighbour variance to fight.
Why don't you publish a daily email volume limit?
Because the limit is set by your sender reputation, not by us. A well-warmed dedicated server with healthy reputation can sustain hundreds of thousands of messages per day. The same hardware with a damaged sending domain or unwarmed IPs will struggle to deliver a few thousand. The volume you can send depends on your IP warmup discipline, list hygiene, complaint-rate management, and authentication posture — not a number we'd write into the plan summary. For teams that want explicit guidance on capacity sizing for a specific sending profile, that's a Custom-plan conversation and we're happy to have it.
What does "operator tolerance for spam complaint troubleshooting" actually mean?
It means we don't suspend your account the first time a Spamhaus listing appears or a Microsoft block lands. Legitimate sending programs hit problems sometimes — a database that wasn't pruned in time, an old segment that got reactivated by accident, an FBL spike from a campaign that didn't land well. Our response is to help you diagnose the cause, walk through the Spamhaus or Microsoft remediation processes alongside you, and get you back to healthy sending. That tolerance is part of why this product costs more than a generic dedicated server. To be clear: it does not extend to intentional spam, abuse of harvested lists, or unsolicited cold outreach outside our Cold Email Infrastructure product. Different problem, different product.
Can I get help with database list cleanup?
Yes — as a paid add-on. We can review a sending list against syntax-validity checks, MX-validity checks, role-account flags (postmaster@, abuse@, etc.), known spam-trap patterns, and engagement history if available. The cleanup is priced based on list size and the depth of analysis you want. This is the kind of work that pays for itself quickly when an old database is the actual cause of a complaint-rate problem — but you'd be surprised how often it isn't.
Are these servers in EU data centres?
By default, yes — Sweden and Germany. EU data residency is the default option, which simplifies the GDPR posture for European data subjects. US and Asia-Pacific data centres are available on Custom plans for teams that need them for latency or specific regulatory reasons.
Why "1 domain per IP" as a limit?
Because it produces clean domain-IP isolation, which is the whole point of dedicated IPs in the first place. Each domain's reputation accumulates against its own IP. The IP's reputation reflects only that domain. Sharing one IP across multiple sending domains commingles reputation in ways that defeat the dedicated-IP model — at which point you're paying premium pricing for the operational equivalent of a shared pool. The 1-domain-per-IP cap is a quality discipline, not an arbitrary restriction.
What MTAs do you have experience supporting?
Postfix, Exim, KumoMTA, PowerMTA (if you bring your own license), MailerQ, Sendmail, Haraka — the common installations across our customer base. Setup assistance covers initial configuration, DKIM key publication, FBL registration, and warmup guidance regardless of which MTA you choose. If you want fully managed PowerMTA with the license bundled into the price, that's our PowerMTA Servers product instead — different shape.
How does this compare to Hetzner or OVH dedicated servers?
Hetzner and OVH sell dedicated hardware at competitive prices — and they're good at that. What they don't sell is the email-specific operational layer. Their teams won't help with Spamhaus delisting, won't handle FBL registration, won't get involved in list-cleanup decisions. Their AUP gets enforced strictly when complaints accumulate. Our bundle includes that layer in the price: dedicated IP allocation with clean reputation, FBL registration on request, Spamhaus remediation assistance, and an operator who can tell the difference between a sender having a bad week and a sender abusing the platform.
Are these servers suitable for cold email outreach?
No — and that's a deliberate boundary, not a side effect. Cold email is a different operational profile. It needs active warmup of fresh dedicated IPs, daily monitoring against Gmail's RETVec classifier and Spamhaus's blocklists, infrastructure built around the peculiar compliance posture cold email occupies post-2025. That's our Cold Email Infrastructure product, built specifically for that work. Dedicated Email Servers are for senders with their own existing list, established sending discipline, and a sending domain with warm reputation already.
Customer engagements
Three deployments where bare metal was the answer
Real engagements, names withheld. Each one shows dedicated email servers solving a problem that a managed service couldn't have — control over the MTA, regulatory residency, multi-tenant reseller architecture. The numbers and the technical details are intact.
01
Swiss hosting reseller — building an email tier on top of dedicated infrastructure
Mid-market Swiss hoster offering web + email to SMB customers, based in Zurich
Context
Previously offered shared cPanel email, which generated steady abuse complaints and time-consuming reputation rescues. Wanted to offer a 'business email' upgrade tier with dedicated IPs and proper deliverability — without operating their own MTA expertise.
Implementation
Two dedicated email servers with PowerMTA, isolated network from the shared hosting. Multi-tenant management layer built by the reseller on top of operator-provisioned vMTAs. Operator handles MTA itself, reseller handles billing and customer UI.
Outcome
The 'business email' upgrade now generates 23% of email-attached revenue with 8% of customer count. Abuse complaints on the dedicated tier: under 1 per 100k messages, versus 30+ on the shared tier.
02
Luxembourg private bank — internal communications and client correspondence
Mid-sized private bank with ~150 RM staff, regulated by CSSF
Context
Sending infrastructure had to be physically located in Luxembourg or an approved EU jurisdiction for client correspondence. Cloud SaaS providers did not satisfy the regulator's data residency interpretation. Internal IT had no MTA expertise.
Implementation
Two dedicated bare-metal servers in Frankfurt (the bank's approved secondary jurisdiction). Postfix selected over PowerMTA for the lighter operational footprint and the regulator's familiarity with it. Operator handles patching, hardening and DNS infrastructure; bank's IT handles application integration.
Outcome
Regulator review approved on first submission. Internal sending now compliant without the bank operating MTA expertise. Operator engagement covered as opex with predictable monthly billing — easier to budget than the headcount alternative.
03
Belgian telco — system notifications and customer service emails
Regional telecom operator in Brussels, ~200k subscribers
Context
System notifications (outage alerts, scheduled maintenance, billing) plus customer service responses had to send from telco-owned IPs and infrastructure for brand and security reasons. Could not outsource to SaaS providers due to procurement policy.
Implementation
Three dedicated email servers running Exim (chosen for the in-house team's existing familiarity). IPs in operator-owned ranges with rDNS configured to telco-branded subdomains. Operator handles MTA + rDNS infrastructure; telco handles application logic.
Outcome
Outage alert delivery latency dropped from a P95 of 14 minutes to 2 minutes. Customer service email reply latency dropped 40% because of cleaner deliverability and faster sender authentication checks at recipient providers.
04
Austrian ISP — bulk transactional plus end-user mail relay
Regional ISP serving Vienna and surrounding states, ~80k consumer subscribers
Context
Mixed workload: ISP system notifications, plus relay for end-user mail clients (residential customers sending through the ISP's SMTP). End-user sending was occasionally generating reputation issues that affected ISP system notifications when both shared infrastructure.
Implementation
Two dedicated email servers with stream separation: PowerMTA on one for high-volume system notifications, Postfix on the other for end-user relay. IPs and rDNS distinct between the two streams.
Outcome
ISP system notifications now consistently above 98% inbox placement. End-user reputation issues — when they occur — stay isolated to their own IPs and do not affect the customer-facing infrastructure.
05
Swedish SaaS infrastructure provider — email as part of a larger product
Internal-tools SaaS based in Stockholm, ~6k paying business customers
Context
Email was one component of a larger platform (think 'observability + alerting' style). Previously used Amazon SES, but the product roadmap required tighter integration: per-customer isolation, branded IPs, and the option for customers to send under their own domains via API. SES could not provide this without significant engineering on top.
Implementation
Two dedicated email servers running PowerMTA. Operator-managed vMTA provisioning exposed to the SaaS provider via API for use during customer onboarding. Customer domain DKIM key management handled by a small layer the provider built on top.
Outcome
Per-customer email isolation became a marketed feature. Two enterprise deals closed in the following quarter cited 'isolated infrastructure' as a primary reason for selecting the SaaS over its competitor. SES bill went to zero.
All engagements anonymised at the customer's request. Industry descriptors, volumes, and technical details reflect actual deployments. Specific company identifiers have been withheld.
Where to look next
Three places that usually answer the next question
Dedicated infrastructure is often the middle ground between a simpler sending service and something more customised on top. The next decision is usually what — if anything — needs to sit on top of it.
Useful when sender posture, authentication, and ongoing performance management matter alongside the infrastructure decision — and you don't have someone in-house with the bandwidth to do it.
