Why does the calculator use 730 hours per month?

AWS, Azure and Google Cloud all price "monthly" estimates as 730 hours (8,760 hours a year divided by 12). Using the same convention makes the numbers match their pricing pages and bills.

What is break-even utilisation?

The share of hours an instance must run for the reservation to cost the same as on-demand. Below it you pay for idle reserved capacity. A typical 1-year reservation with a 30–40% discount breaks even at 60–70% utilisation.

Should I pay upfront?

All-upfront gives the largest discount but locks cash in on day one, so break-even arrives later in the term. No-upfront breaks even immediately at high utilisation but saves less. Compare both by entering each option.

Does this work for savings plans and committed use discounts?

Yes, as long as you express the commitment as an hourly rate (plus any upfront payment) for the capacity you would otherwise run on demand. Flexible plans can move between instance types, which reduces the risk the warnings describe.

Where do the sample prices come from?

They are illustrative values for a small general-purpose VM, not a quote. Prices differ by provider, region, operating system and date, so always enter the current rates for your workload.

Reserved instance break-even calculator

Cloud cost

Enter on-demand and reserved prices from any cloud provider to see how much a commitment saves, how busy the instance must be, and when it breaks even.

Runs entirely in your browser — your input is never uploaded.

Result

Enter the on-demand and reserved prices, or load the sample.

How the comparison works

A reservation (AWS Reserved Instances or Savings Plans, Azure Reservations, Google committed use discounts) trades flexibility for price. You pay for the capacity every hour of the term whether or not you use it, in return for a lower rate.

The calculator compares, per instance:

On-demand cost = on-demand hourly price × 730 hours × utilisation × months
Reserved cost = upfront payment + reserved hourly price × 730 hours × months

From those it derives the total saving, the effective hourly rate (upfront spread over every hour of the term), the break-even utilisation, and the break-even month: the first month in which the on-demand bill you avoided exceeds what the reservation has cost so far.

Worked example

The sample uses illustrative prices for four always-on instances on a 1-year partial-upfront plan:

	Per instance	× 4 instances
On-demand	0.096/hour → 70.08/month	280.32/month
Reservation	280 upfront + 0.032/hour	1,120 upfront + 93.44/month

Over 12 months, on-demand costs 3,363.84 and the reservation 2,241.28, a saving of 1,122.56 (33.4%). The effective rate is 0.064/hour.

The reservation breaks even at 66.6% utilisation. Below about 16 hours a day, on-demand is cheaper. In month 0 you are 1,120 behind because of the upfront payment, and you catch up in month 6.

A decision checklist

Measure utilisation first. Use the past 30–60 days of hourly usage, not peak capacity. Reserve the baseline that never scales down and leave the peaks on demand.
Prefer flexible commitments. Compute Savings Plans (AWS) or spend-based commitments (GCP) follow your workload across instance families and regions. Instance-specific reservations don’t.
Mind the term. A 3-year term has a bigger discount but outlives most hardware generations. Newer instance types are often cheaper per unit of work than your old discounted rate.
Right-size before you commit. Reserving an oversized instance locks in the waste for the whole term.
Revisit quarterly. Coverage and utilisation drift as services change. Most providers can export a utilisation report for existing reservations.

What the calculator does not model

Price changes during the term (on-demand prices usually fall over time, while your reserved rate stays fixed).
Reselling unused reservations (possible for some AWS Standard RIs) or exchanging them.
Tax and currency effects, and the cost of capital tied up in an upfront payment.

Frequently asked questions

Why does the calculator use 730 hours per month?: AWS, Azure and Google Cloud all price "monthly" estimates as 730 hours (8,760 hours a year divided by 12). Using the same convention makes the numbers match their pricing pages and bills.
What is break-even utilisation?: The share of hours an instance must run for the reservation to cost the same as on-demand. Below it you pay for idle reserved capacity. A typical 1-year reservation with a 30–40% discount breaks even at 60–70% utilisation.
Should I pay upfront?: All-upfront gives the largest discount but locks cash in on day one, so break-even arrives later in the term. No-upfront breaks even immediately at high utilisation but saves less. Compare both by entering each option.
Does this work for savings plans and committed use discounts?: Yes, as long as you express the commitment as an hourly rate (plus any upfront payment) for the capacity you would otherwise run on demand. Flexible plans can move between instance types, which reduces the risk the warnings describe.
Where do the sample prices come from?: They are illustrative values for a small general-purpose VM, not a quote. Prices differ by provider, region, operating system and date, so always enter the current rates for your workload.

Kubernetes YAML validator and visualizer
Paste Kubernetes manifests to get a resource diagram and a security and best-practice review: privileged pods, root, probes, limits, deprecated APIs.
KubernetesBeta
CIDR calculator and subnet splitter
Calculate network, broadcast, host range, netmask and wildcard for any IPv4 CIDR, then split it into subnets by prefix, count or host size.
Network

Last updated October 24, 2026 · More Cloud cost tools

Result

How the comparison works

Worked example

A decision checklist

What the calculator does not model

Frequently asked questions

Related tools

Kubernetes YAML validator and visualizer

CIDR calculator and subnet splitter