A server advertised at €8 a month costs €8 a month. That does not mean running your system on it costs €8 a month.
Infrastructure prices are easy to compare because they are numbers. The costs around infrastructure are harder, because they arrive as staff time, outages, migrations and unpleasant evenings.
The invoice is one line of the cost
A useful comparison includes the machine, the storage, the backups, the traffic, the monitoring, the licences and the support. It should also include the work.
Who patches the operating system? Who responds when the disk fills at 03:00? Who notices that backups stopped three weeks ago? Who replaces the machine when it fails, and how long does recovery take?
If the answer is that a developer will handle it, the work still has a price. It is being paid out of a different budget, usually one that was meant to buy features.
Cheap can be exactly right
There is nothing wrong with inexpensive infrastructure. A development environment, a personal project, a disposable test system or a small low-risk service may belong on the cheapest machine that reliably runs it.
Paying for elaborate redundancy around something that can be rebuilt in twenty minutes is waste.
The mistake is assuming that because the inexpensive option is right for one workload, it is right for all of them.
Reliability has a cost curve
Making a system slightly more reliable is usually cheap. Making it extremely reliable becomes expensive quickly.
A second copy of the data is straightforward. A second site is more complicated. Automatic failover brings orchestration, testing and a set of new failure modes that did not previously exist. Removing every single point of failure can cost more than accepting a short outage would have.
So the design should start from the business consequence of failure, rather than from an abstract preference for high availability.
Migration is part of the price
The cheapest provider today can become the expensive one when the system outgrows it. Proprietary services, unusual deployment models and data-egress charges all make leaving harder.
Even without deliberate lock-in, an undocumented environment becomes expensive simply because nobody remembers how it works.
Standard operating systems, ordinary databases, documented deployments and accessible backups keep future choices cheap.
Compare outcomes, not line items
If two servers do the same job under the same operational model, compare the price.
If one includes monitoring, tested backups, patching and somebody responsible for recovery, and the other is a root password and good luck, they are not the same product and the prices are not comparable.
The cheapest server can absolutely be the right server. Price the system around it as well.