Fargate vs EC2 Calculator
Compare Fargate against the EC2 capacity you would actually have to buy — whole instances, agent overhead, deploy headroom, and the hours a cluster runs when the tasks do not. The per-vCPU comparison everyone makes is the misleading one.
| vCPU40.00 vCPU × $0.04048/hr | $1,182.02 |
| Memory80.0 GB × $0.004445/GB-hr | $259.59 |
| Ephemeral storageAbove the 20 GB included per task | $0.00 |
| Total | $1,441.60 |
What the EC2 side actually costs
The EC2 figure charges the cluster for the whole month, because a cluster sized for this workload runs continuously even when the tasks do not. That is the asymmetry that decides most of these comparisons: Fargate stops billing when a task stops, and an instance does not.
What this model assumes
- EC2 is charged for capacity, not for tasks. You cannot buy 5.5 vCPU; you buy instances and pack tasks onto them. The model sizes the cluster at your stated achievable utilisation and bills it continuously, which is what actually happens.
- The utilisation input is doing the real work. 60% is a reasonable default for a mixed cluster with headroom for deploys. Teams running bin-packed batch workloads reach 80%+; teams with a handful of large tasks per instance rarely pass 50%. Measure yours before trusting the verdict.
- Fargate Spot is a band, not a rate. Spot pricing floats with capacity and AWS does not publish it in the price list, so the headline uses the conservative end. It also carries a two-minute termination notice, which rules it out for anything that cannot be interrupted.
- Not modelled: the ECS control plane (free), load balancers, NAT and egress, EBS on the EC2 side beyond the root volume, or the operational cost of patching and scaling an instance fleet — which is the reason many teams choose Fargate at a price premium.
- Tasks bill a one-minute minimum. For very short-lived tasks the minimum, not the rate, dominates.
Frequently asked questions
Is Fargate more expensive than EC2?
Per vCPU-hour, yes — roughly twice the vCPU component of an equivalent EC2 instance. Per cluster, frequently not, because you cannot buy exactly the EC2 capacity you need. You buy whole instances, lose a slice to the OS and the ECS agent, keep headroom for deploys, and pay for the cluster during hours the tasks are not running. Charge those and the premium shrinks sharply; for spiky or low-duty-cycle workloads it inverts.
How much does AWS Fargate cost?
$0.04048 per vCPU-hour and $0.004445 per GB-hour in US East (N. Virginia), plus $0.000111 per GB-hour of ephemeral storage above the 20 GB included. A task with 1 vCPU and 2 GB running continuously costs about $36 a month.
When is Fargate cheaper than EC2?
Whenever your achievable EC2 packing density is low or your duty cycle is short. Clusters with a few large tasks per instance, bursty batch work, or anything that runs well under 24/7 tend to favour Fargate. Densely bin-packed always-on clusters favour EC2 — and this calculator reports the exact utilisation at which the two cross over.
Does Fargate Spot save money?
Substantially — commonly 50-70% off — but AWS does not publish Spot rates in its price list because they float with capacity, so this calculator shows a band and uses the conservative end for the headline. Spot tasks also get a two-minute termination notice, which rules them out for anything that cannot be interrupted.
What task sizes does Fargate allow?
Only specific vCPU and memory combinations: 0.25 vCPU with 0.5-2 GB, up to 16 vCPU with 32-120 GB. Ask for a memory size outside the legal range for your vCPU and the task will not start, which this calculator flags rather than silently pricing.
One workload compared. A cluster has hundreds.
Finitizer profiles every task definition and every instance across your accounts, finds the services where the compute model is the wrong one, and shows the packing density you are actually achieving rather than the one you assumed.