Cloud computing was supposed to save money. And it can, but only if you understand what you're actually paying for. Too many small businesses either overspend on cloud they don't need, or avoid it entirely based on misconceptions.

Here are the myths I hear most often, and what's actually true.

The myths

"Cloud is always cheaper than on-premises." Not always. For stable, predictable workloads that run 24/7, a well-managed server can be cheaper than the equivalent cloud instance. Cloud wins on flexibility, not always on raw cost. If your usage is bursty or growing, cloud usually makes sense. If it's flat and predictable, do the maths first.

"We need the biggest instance available." Most apps run fine on smaller instances. I regularly see businesses paying for 8-core, 32GB instances to run applications that use 10% of the CPU. Start small, monitor, and scale up only when you need to.

"Serverless is free." Serverless (Lambda, Azure Functions) can be very cheap at low volumes. But costs scale linearly with usage, and at high volumes it's often more expensive than a dedicated instance. It's free up to a point. Know where that point is.

"We'll save money if we move everything to the cloud." Lift-and-shift migrations (taking your existing setup and putting it on AWS or Azure without redesigning it) often cost more, not less. You end up paying cloud premiums for infrastructure patterns designed for on-premises.

"Cloud providers handle all the security." Cloud providers secure the infrastructure. You secure everything you put on it: your data, your access controls, your application code. This "shared responsibility" model trips up a lot of businesses.

The reality

Cloud costs are manageable if you treat them as ongoing optimisation, not a set-and-forget expense. The businesses that spend well on cloud share some habits:

  • They review their cloud bill monthly and know what each line item is for
  • They right-size instances based on actual usage, not estimated peaks
  • They use reserved instances or savings plans for predictable workloads
  • They shut down development environments outside business hours
  • They delete orphaned resources: old snapshots, unused load balancers, forgotten storage

Right-sizing your cloud spend

  1. Audit current usage: Look at actual CPU, memory, and network utilisation for every instance. Most are over-provisioned.
  2. Use the right pricing model: On-demand for bursty workloads. Reserved or savings plans for stable ones. Spot instances for batch processing.
  3. Architect for cloud: Containerised apps with auto-scaling cost less than oversized VMs sitting idle.
  4. Set billing alerts: AWS and Azure both let you set alerts when spending exceeds a threshold. Use them.
  5. Review quarterly: Cloud pricing changes, your needs change. Schedule quarterly reviews.

Cloud isn't inherently expensive or cheap. It's a tool with a meter running. The businesses that manage it well save money. The ones that don't, don't.

Kasun Wijayamanna Founder & Lead Developer Postgraduate Researcher (AI & RAG), Curtin University - Western Australia
Cloud CostsBusiness OwnersIT Leaders