How much does a SaaS MVP cost in 2027?

If your number came out scary, good: you found out here, for free, and not in week nine. Go back and untick something. The first version almost always has an extra that can wait for real users to ask for it.

For the record, my own prices are on my services page.

AI made typing code cheap. It did not make deciding, checking or running software cheap.

The ranges above are already smaller than they would have been a few years ago, because coding agents write most of the boilerplate. A form, an API endpoint, a test, a migration: minutes instead of hours. If someone quotes you 2021 timelines for a standard CRUD screen, ask why.

What didn't shrink:

• Deciding what to build. An agent will build the wrong thing very fast and very confidently.
• Checking what it built. Generated code compiles and looks fine. Whether it leaks one customer's data to another is a question someone has to ask, on purpose.
• Running it. Backups, monitoring, security updates, the payment webhook that fails at 3 a.m. AI doesn't get paged.

This is also why "I'll just build it myself with an AI app builder" is a real option now, not a joke. Tools like Lovable start at about $25 a month and get you a working prototype in an afternoon. Where it gets expensive is the last part: payments, permissions, data that must not be lost. I wrote a whole post on what to do when an AI-built app needs a developer, because it happens a lot.

There's no single right answer. Here's an honest comparison, with prices checked on 13 November 2026.

• No-code (Bubble): from about $30 a month for a web plan, plus usage, plus your own time. Best when you want to test demand and you're happy to learn the tool. Watch out: you can't take the code with you if you outgrow it.
• AI app builder (Lovable): about $25 a month and your evenings. Best for a prototype to show investors or first users. Watch out for the last 20%: payments, security, data.
• Freelancer: their rate × days; the calculator above uses $35, $75 and $150 an hour. Best with a clear scope and one person you trust. Watch out: one person is one point of failure, so check who owns the code.
• Agency: most on Clutch charge $24 to $49 an hour; onshore agencies charge a lot more. Best for a big scope with design and a project manager included. Watch out: you pay for the project manager too, and you may never meet the developer.
• Hire an employee: a US software developer's median pay was $135,980 a year in May 2025, before taxes and benefits. Best when you're past the MVP and building for years. Watch out: hiring takes months, and one person rarely covers design, backend and servers.

Note what's missing from that list: "the cheapest option". Every one of them is the cheapest for somebody.

The build is a one-time bill. These come every month:

• A server. A small cloud server at Hetzner starts at €5.49 a month before VAT. With a staging copy and backups, my rule of thumb is €20 to €50 for an early product. If someone suggests a setup billed per request with no spending cap, ask what happens when a bot finds your sign-up form.
• Payment fees. Stripe takes 2.9% + 30¢ per US card payment, plus 0.7% for subscriptions through Stripe Billing. On a $20 subscription, that's about $1 a month per customer. Not a reason to skip payments; a reason to price above $5.
• Email. Most sending services have a free tier for the first few thousand emails a month. You'll pay once you have users, which is a nice problem.
• AI usage, if your product calls an LLM. Set a hard monthly cap with the provider on day one, not after the first surprising bill.
• Someone fixing things. Software breaks: libraries need updates, a browser changes something, a user does something nobody imagined. My rule of thumb is 15 to 20% of the build cost per year, or a few hours a month, for a product that's live but not growing fast.

Most MVP budgets don't blow up because someone lied. They blow up because the scope grew one "small thing" at a time. A few habits stop that:

1. Write the cut list first. Next to the feature list, write what's not in version one. It's the most useful page in the whole project.
2. Pay for a small first milestone. One or two weeks of work you keep either way. You learn how this person works before you bet the whole budget.
3. See working software every week. Not a status report. Something you can click.
4. Get a price for every change before the work starts. "We'll sort it out at the end" is how $20,000 becomes $35,000.
5. Own your accounts from day one. The domain, the server, the Stripe account and the code repository are in your name. If the developer disappears, your product doesn't.

Whoever you ask, me included, send this and you'll get a quote you can actually compare:

• Who pays, and for what? One sentence about the customer and the action they pay for.
• The core feature, described as steps a user takes, not as a vision.
• The extras from the table above that version one can't launch without.
• Your cut list, so nobody quotes features you don't want.
• The tools it must talk to on day one.
• Your budget and your date. You're not weakening your position. You're stopping someone from quoting a spaceship when you need a bicycle.

Two developers who get this page will give you quotes within shouting distance of each other. Two who get "it's like Uber, but for dog walkers" will not.

The cheapest MVP is the one you cut in half before anyone writes a line of code.