Self-Hosting vs. SaaS in 2026: The Real Cost Breakdown

"Just self-host it, it's free" is one of the most repeated pieces of advice on the internet. It is also one of the most misleading. Open-source software has no license fee, but running it still costs something. The question is not whether self-hosting is cheaper. The question is when it is cheaper, and for whom.

This post gives you a simple model to answer that for your own situation.

The three costs nobody puts on the invoice

A SaaS subscription bundles a lot of work into one monthly number. When you self-host, that work does not disappear. It moves to you. There are three buckets.

Infrastructure. A small server, backups, and bandwidth. For most self-hosted tools this is genuinely cheap: a $6 to $20 per month virtual machine covers a surprising amount. Object storage for backups adds a dollar or two.

Setup time. The hours to install the software, configure a domain and TLS, wire up email sending, restore-test your backups, and document what you did so future-you can follow it. This is a one-time cost, but it is real, and it is usually underestimated by a factor of two or three.

Maintenance time. Security updates, version upgrades that occasionally break things, disk space that fills up, and the odd outage you have to debug at an inconvenient hour. This is the cost that compounds. It never goes to zero, and it scales with the number of tools you run.

A model you can actually use

Put a number on your hourly time. If you are a professional developer, a conservative figure is what your employer pays for an hour of your work, or what you would charge a client. Even if you self-host on weekends "for fun," your time has an opportunity cost. Use something between $30 and $100 per hour.

Now compare over a three-year window, because that is roughly how long a tool stays in your stack before you re-evaluate it.

SaaS 3-year cost   = monthly price × 36

Self-host 3-year   = (server + storage) × 36
                   + setup hours × hourly rate
                   + maintenance hours per month × 36 × hourly rate

A worked example. Say the SaaS costs $25 per month, so $900 over three years. You can self-host the same tool on a $12 per month server with $3 per month of backup storage, which is $540. Setup takes you 6 hours. Maintenance is about 1 hour per month. At a $50 hourly rate:

In this example, self-hosting costs roughly three times as much once your time is counted. The subscription is the bargain.

Flip the numbers and the answer flips too. If the SaaS costs $200 per month per seat and you have ten seats, that is $ 72,000 over three years. Suddenly 6 hours of setup and an hour a month of babysitting is trivial by comparison, and self-hosting saves you tens of thousands of dollars.

Where self-hosting clearly wins

• Per-seat pricing at scale. Anything charged per user gets expensive fast. Self-hosted tools usually do not care how many people log in.
• High-volume usage. Transactional email, analytics events, CI minutes, storage. SaaS meters these; a server does not.
• Data you cannot send to a third party. Sometimes the compliance or trust requirement makes the cost comparison irrelevant.
• Tools you would run anyway. If you already operate a server and a backup routine, the marginal cost of one more container is small.

Where SaaS clearly wins

• Low-seat, low-volume tools. A $10 per month service you use lightly is almost never worth self-hosting.
• Anything where downtime hurts. If an outage costs you customers or revenue, you are now paying for on-call, not just a server.
• Tools with hard operational demands. Databases at scale, email deliverability, video processing. These are their own specialties.
• When your time is the bottleneck. If you are a small team shipping a product, hours spent on infrastructure are hours not spent on the thing customers pay for.

A middle path

You do not have to choose globally. A reasonable pattern:

1. Self-host the boring, stable, high-value stuff. A wiki, a bookmark manager, an analytics tool, a status page. These change slowly and rarely break.
2. Pay for the operationally hard stuff. Email, error tracking at scale, anything customer-facing where an outage is expensive.
3. Re-evaluate yearly. Prices change, your team size changes, and a tool that was not worth self-hosting at three seats might be very worth it at thirty.

The honest summary

Self-hosting is not a way to get software for free. It is a way to trade money for control and, often, to trade a predictable subscription for an unpredictable time commitment. For a solo developer running a handful of stable tools, that trade is frequently worth it. For a growing team paying per seat, it is worth it by a wide margin. For a small team trying to ship a product with a $15 per month tool, it usually is not.

Run the three-year numbers with your own hourly rate before you decide. The answer is more often "it depends" than either camp will admit.