The Real Cost of AWS: What Nobody Tells You When You Sign Up

The first AWS bill is almost always a pleasant surprise.

You’ve spun up a server, maybe a database, a storage bucket. The cost is low — sometimes startlingly so. The dashboard is impressive. The documentation is thorough. It feels like the obvious choice, and for a moment, you wonder why anyone would do things any other way.

Then the organisation grows. The usage grows with it. And the bills start to look different.

Why cloud pricing is hard to predict

AWS, Azure, and Google Cloud all use consumption-based pricing. You pay for what you use — compute time, storage, data transfer, API calls, and dozens of other dimensions that interact with each other in ways that are genuinely difficult to model in advance.

This isn’t a flaw in the system. It’s a feature for organisations with highly variable workloads who want to scale up and down on demand. For them, it makes sense.

For a school, a clinic, or a small manufacturer with relatively stable and predictable usage, it’s a different story. You end up paying for flexibility you don’t need, priced in a way you can’t easily forecast.

The costs that don’t show up on day one

Beyond the base compute costs, a few things tend to surprise organisations that haven’t been through this before.

Data transfer fees. Moving data out of a cloud platform — to another provider, to your own systems, or even just to your users at scale — costs money. This is sometimes called egress pricing, and it’s one of the most effective mechanisms for keeping organisations inside a platform once they’re in.

Service dependencies. Once you start using managed databases, queuing services, authentication tools, and monitoring systems offered by the same platform, switching becomes exponentially harder. Each service you adopt is another thread tying you in.

Support costs. The free support tier on most platforms is limited. Meaningful response times and technical support require paid plans that add meaningfully to the monthly total.

The cost of expertise. Running infrastructure on AWS or Azure requires either hiring someone who knows the platform well or paying someone external who does. That knowledge is specific to the platform — it doesn’t transfer cleanly elsewhere.

The lock-in nobody talks about openly

The biggest cost of big cloud isn’t on any invoice.

It’s the moment you realise that leaving would be so disruptive, so expensive, and so technically complex that you simply don’t. So when the next price increase arrives, you absorb it. When the service you depend on changes its terms, you adapt. When a better option exists elsewhere, you stay anyway.

That’s vendor lock-in. And for regulated organisations — schools handling student records, clinics managing patient data, manufacturers running operational systems — it carries an additional dimension: you’re dependent on a vendor’s decisions for the security and availability of data that isn’t yours to risk.

What predictable infrastructure actually looks like

The alternative isn’t going back to a server room in the basement. Private cloud infrastructure today can be hosted in professional data centres, with proper redundancy, security, and uptime — without the consumption-based pricing model and without the vendor dependency.

The organisations I work with that have made this shift share a common experience: the first year feels like a larger investment, and every year after that feels increasingly like the right decision. Fixed costs. Predictable budgets. Data that stays where you put it.

For a clinic or a school, that predictability isn’t just financially useful. It’s operationally essential.

This isn’t an argument against cloud for everyone

Large organisations with genuinely variable workloads, global distribution requirements, and dedicated cloud engineering teams get real value from AWS and its equivalents. The platforms are genuinely impressive at the scale they were designed for.

But if you’re a school with 500 students, a clinic with three locations, or a manufacturer running a webshop and an operations system — you’re not that organisation. And you’re paying for infrastructure designed for one that is.

If your cloud bills feel harder to justify than they used to, or if you’re starting to ask questions about where your data lives and who controls it, I’m happy to have that conversation.

Connect on LinkedIn or reach out directly at hi@madalin.me.