What an MVP actually costs in 2026

A few weeks ago a friend showed me a quote for a data management app for the agriculture market: €20k, 17 weeks, one developer plus 10% for project management. He wanted to know if it was fair. He expected me to say it was too expensive.

I told him it was too cheap.

€20k over 17 weeks is about €1.2k a week for a turn-key project where the developer carries all the risk. That barely covers cost. And margin is exactly what absorbs the unexpected. On a project this loosely defined, slipping from 17 weeks to 24 is easy — it happens on the first real scope surprise — and a quote priced at cost has nowhere to go. Either the developer eats the overrun or the project stalls when the money runs out. The founder usually blames the builder for something the brief set in motion.

That's the whole reason for this article. Every founder wants a number, but the honest answer to “what does an MVP cost” is another question: who's building it, and how well do you know what you want? Those two variables move the price by an order of magnitude — and a suspiciously low number can be a bigger warning sign than a high one, because it usually means the risk hasn't been priced in, just deferred to you.

Here's the honest version, with real numbers, including the paths we don't sell.

The four ways to build an MVP

Freelancer — roughly €3k to €35k

One person, or two (usually one for frontend, one for backend). Cheapest hourly, fastest to start. The catch is structural: a freelancer is a single point of failure. If they get sick, get a better offer, or stop replying, you're holding half-finished code nobody else wants to inherit. Quotes also tend to cover only design and code — hosting, domains, and API costs land on you separately.

No-code / low-code — €2k to €30k+ to build, plus platform fees

Build it cheaply, maybe without engineers at all. For a genuinely simple product, it can be exactly right. But no-code has a ceiling, and you tend to discover where it is at the worst possible moment — when you've got users and you're hitting a limit you can't engineer around. More on this below.

Agency — €15k to €250k+, months not weeks

Real project management, a team that doesn't vanish, a contract. You're paying for reliability — and for process, which runs 6 to 28 weeks with the meter on scope creep the whole time. A US agency quote for a “validation MVP” commonly lands between $30k and $80k; fintech and healthcare go far past that.

AI-assisted fixed-price — €6,995, fixed

The newer category, and the one we build in. One number for a defined scope, no hourly meter. AI genuinely compresses parts of the build — scaffolding, boilerplate, first-draft APIs — which is what lets a small operation commit to a fixed price and timeline instead of billing by the hour. But only if the requirements are clear enough to scope up front. That “if” is the whole game, and it governs every option above too.

What actually drives the number

Look past the delivery model and the same factors set the price every time. A software engineer on Reddit recently posted real MVPs they'd shipped, with prices — the most honest thing I've read on this topic all year. A few that stuck with me:

• A two-sided marketplace where they faked the supply side with an admin spreadsheet at launch — because the founder needed to test demand, not build real supply yet. Cheaper, faster, smarter.
• A €19k SaaS dashboard where two integrations were half the budget.
• A real-time collaboration feature set that "roughly doubles whatever you're imagining" — real-time always does.
• A €27k rebuild of a vibe-coded v1 with paying users, where a third of the cost was just untangling the original.

The pattern: integrations, real-time, compliance, and native device features are where budgets go to die. Not the core feature — the plumbing around it. A booking app is cheap until it syncs with three calendars. A fintech tool is cheap until compliance adds €6k over the features alone.

And the biggest multiplier isn't a feature at all. It's how well you've defined what you want. Ambiguity is the most expensive thing in software. Every builder pads their estimate to cover the decisions you haven't made. A vague brief doesn't get you a cheap build. It gets you a padded quote and a fight later.

A cautionary tale about no-code's “cheaper and faster” promise

Here's the clearest thing I've seen about how “save money, skip the engineers” plays out when a project pushes past what the platform was built for.

A major telecom in Chile built their customer self-care app on a low-code platform. The pitch was the classic one: hire non-technical people instead of engineers, save on implementation — three months, a couple of hackathons, a free academy to spin up local trainees. The app was essentially a frontend over existing APIs, but it had to serve 8 million users. It took a year to build. Six months more to stabilize. For a stretch it was the worst-rated app in the stores, and the platform vendor's own CTO had to fly in and debug it.

Same company, another app: order-entry with native fingerprint recognition, estimated at three months, delayed by six because the low-code layer couldn't handle the range of fingerprint devices. The fix was a native Java app with the low-code app embedded inside it — a hack, to make the “no-code” solution do the one thing it couldn't.

We're not dunking on low-code. We've shipped a conference registration app on the same platform and it went great — simple, self-contained, exactly its sweet spot. That's the point: the tool is genuinely the best choice inside its range. The problem is the gap between where it works and where it gets sold. “Cheaper and faster because you skip the hard part” is a promise that comes due eventually. The hard part doesn't disappear; it just moves to where you'll notice it later.

So which path is right for you?

Testing whether anyone wants this at all? Don't build the real thing yet. A landing page and a manual “product” behind the button validates demand for a few thousand euros or less. Half of you could skip the build for another month.

Genuinely simple, and likely to stay that way? No-code, honestly — just know where the ceiling is first.

Is your product the transactions — payments, payouts, compliance, real-time? Budget properly and pick a path with real engineering. This is not where you save money.

Know roughly what you need and want it built properly on a predictable budget? This is where fixed-price, AI-assisted delivery fits, and the gap we built FSGen to fill: defined scope, fixed price, two-week delivery, full source code you own.

Which raises a fair question, given how this opened: if a low quote is a warning sign, isn't a fixed price that undercuts an agency the same red flag? It's the opposite. That opening quote was cheap because the risk was ignored — priced at cost, nothing set aside for the decisions still to make. A fixed price works because the risk is controlled — scope locked before a line of code, so there's little left to overrun. One is cheap because someone forgot to price the uncertainty. The other is affordable because the uncertainty has been eliminated.