Is a Hosting Business Profitable? What It Costs to Keep Running

Is a hosting business profitable? Yes, and the number is decided by things that never appear on your price list. The launch bill is a spreadsheet you can plan around. What comes after launch is a support load that varies wildly per customer, an abuse queue that never empties, and churn that deletes the customers you paid to acquire.

We run our own fleet, so this is the post-launch reality as we see it. The startup side is covered in what it costs before your first customer; this is the month after that.

Two customers on the same plan are not equally profitable#

Take two customers on the identical plan at the identical price. One of them you hear from twice a year. The other opens a ticket a month, and each ticket is twenty minutes of an engineer who could otherwise be racking hardware. On thin hosting margins, that second customer can consume the profit on the first one and then some.

Price lists do not model this. Revenue is a flat line per plan; cost per customer is not, and the spread between the quiet ones and the loud ones is wider than the spread between your plans.

The lever is not charging talkative customers more. It is drawing the scope line clearly enough that the tickets you get are the ones you are paid to answer. Ours is written down: everything on this site is unmanaged, meaning we run the hardware, network, and virtualization, and you run what is inside the server. Engineers still answer tickets around the clock and will tell you which log to read. What they will not do is administer your application for you, and that boundary is why unmanaged prices can stay where they are.

Abuse handling is a job, not an incident#

Every host with an IP range learns that abuse is a standing job with a person attached. Compromised servers send spam. Someone signs up with a stolen card and starts scanning. A customer gets your address space listed, and mail from an unrelated customer stops being delivered.

Three costs come out of that, only one of them obvious:

• Time. Reading reports, matching them to an account, contacting the customer, deciding whether to suspend. None of that automates down to free.
• Reputation. IP space with a bad history is worth less and delivers mail worse. You are maintaining an asset, not just holding one.
• Refunds and chargebacks. Fraud signups are revenue you were never keeping, plus fees, plus risk to your merchant account if the rate climbs.

What keeps this manageable is writing the rules down before you need them. Our acceptable use policy names what is not allowed and what happens when a report lands, so an abuse decision enforces a published rule rather than an argument invented under pressure. Hours saved per case is margin.

Churn eats acquisition before you see it#

Hosting customers are not bought once. If it costs a month of revenue to win a customer and they leave in month four, you did not make money on them, you rented them. That maths is brutal at the budget end, where the same customer is shopping again the moment a cheaper coupon appears.

Two things move churn, and neither is a discount. Sizing the customer correctly at the start, because a server that was too small on day one becomes a cancellation in month three. And not degrading under them: nobody leaves a server that has quietly done its job for two years.

We also publish a 30-day money-back guarantee on new VPS orders, which sounds like a churn cost and behaves like the opposite. It moves the mismatch to week two, when it is a refund, rather than month six, when it is a bad review.

Someone is always on call#

A hosting business is a promise that a machine keeps working while you are asleep, which means a person is awake, or at least reachable, permanently. Our network uptime commitment is 99.9%, backed by credits, and a commitment like that is a staffing decision before it is a marketing line. So is round-the-clock ticket coverage.

For a solo founder that decides whether the business is still enjoyable in year two, and it is why the honest advice is to start smaller than feels satisfying. Capacity you sold and cannot support costs you the customer, the refund, and the review.

The levers that actually move the number#

The margin decisions worth making reduce future work rather than raise present revenue:

• A hard density ceiling. We cap every node at 60% average CPU load and sell from a fixed stock of slots per host, so a node hits its sales limit before its load limit. That reads as an engineering policy. It is a margin policy, because the tickets you never receive are the cheapest ones you will handle. The reasoning is in why we use stock control.
• Boring, uniform hardware. One platform means one spares inventory, one firmware baseline, and one set of failure modes the team already recognizes. Ours is AMD EPYC 4564P with NVMe storage and uplinks of 10 Gbps and up.
• Not competing on price. Overselling to hit a headline number converts a margin problem into a support problem, at a worse exchange rate. We wrote that up from the buyer's side in the cheap hosting trap. The operator's side is the same argument with your payroll attached.
• Forecastable fixed costs. Committed power and a transit commit are unpleasant to sign and pleasant to own: they turn your largest recurring costs into known numbers. Metered surprises are what make a profitable month look unprofitable in arrears.

So, is it worth doing?#

It is profitable at a scale and a discipline, not at a price point. Capital comes first, and IP space is the gate that stops most people before they reach any of this. What decides the outcome afterwards is how much work each customer costs you, and how much of that work you designed away in advance.

If you are at the stage where your own hardware finally makes sense, our Seattle colocation starts at a single 1U with redundant A and B feeds, and the quote process prices space, power, and transit line by line.

FAQ#

What profit margin does a hosting company make?
It varies more by support load than by plan, which is why a single figure misleads. Two customers paying the same amount can differ by a wide multiple in what they cost to serve.

Is reselling hosting profitable?
It can be, at low margin and low risk, because you pay a monthly fee instead of committing capital. You still inherit the support load without the ability to fix the platform underneath it.

What is the biggest ongoing cost of running a hosting company?
People. Power, transit, and licenses are large but predictable. Support, abuse handling, and on-call coverage scale with customers rather than servers.

Does automation fix the support cost?
It removes the repetitive half: provisioning, reinstalls, backups, and self-service in the portal. It does not remove judgement calls, abuse decisions, or the customer whose problem is inside their own application.