Cloud Migration TCO: How to Calculate Total Cost of Ownership

"Move to the cloud and save money" is a slogan, not a business case. Whether cloud is cheaper depends entirely on your workloads, and the only honest way to know is a total-cost-of-ownership model that counts the one-time cost to get there, the recurring cost to run there, and the on-prem baseline you're comparing against. This guide gives you that framework.

What TCO means for cloud

Total cost of ownership is every dollar a workload consumes over a defined horizon — typically three years — not just the monthly cloud bill. A credible cloud TCO has three parts:

1. One-time migration cost — the investment to get there.
2. Ongoing run cost — what it costs to operate each year once migrated.
3. Baseline cost — what staying on-premises would cost over the same horizon.

The decision isn't "is cloud cheap?" — it's "over three years, does (migration + cloud run) beat (on-prem run), and if so, when does it break even?"

One-time migration costs

These are frequently underestimated because they're spread across teams and don't appear on a cloud invoice:

• Assessment & planning — discovery, dependency mapping, migration design.
• Re-architecture / refactoring — the "6 Rs" spectrum: rehost (lift-and-shift) is cheap up front but often expensive to run; replatform and refactor cost more now and less later.
• Data transfer — moving terabytes in (often free) but the effort, tooling, and cutover windows are not.
• Parallel run — you usually pay for both environments during transition.
• Testing & validation — performance, security, and DR testing in the new environment.
• Training & change management — cloud skills, new runbooks, new on-call.

Ongoing run costs

The recurring bill is more than instances. Model each of these:

The hidden costs that wreck the model

Establishing the on-prem baseline

Cloud only looks expensive or cheap relative to something. Build the fully-loaded on-prem cost over the same horizon:

The trap is comparing a cloud bill to only hardware cost. Data center facilities, power, and the staff time to rack, patch, and replace gear are real — and they're exactly the costs cloud is meant to absorb.

Break-even analysis

With migration cost, annual cloud run, and annual on-prem run in hand, break-even is when cumulative cloud spend (including the up-front migration) crosses below cumulative on-prem spend:

In this illustration cloud is more expensive until sometime in year 3, then pulls ahead. If your horizon were two years, on-prem would win — which is exactly the kind of conclusion a TCO model exists to surface before you commit.

Optimization levers (model them, don't assume them)

• Reserved / savings plans & spot — large discounts for committed or interruptible workloads.
• Right-sizing & autoscaling — match capacity to real demand instead of peak.
• Storage tiering & lifecycle policies — move cold data to cheaper tiers automatically.
• Architecture changes — reduce cross-region and egress traffic.

Model these as scenarios, not as a foregone 40% saving. The honest number is a range, and the assumptions behind each lever should be written down.

Common mistakes

• Comparing the cloud bill to on-prem hardware only (ignoring facilities and staff).
• Forgetting egress and inter-region traffic.
• Assuming lift-and-shift is the cheap option — cheap to migrate, often costly to run.
• Using a 1-year horizon when migration costs need 2–3 years to amortize.
• Treating optimization savings as guaranteed rather than a scenario.