AWS EC2 Pricing Explained: Costs and Savings (2026)

AWS EC2 pricing starts with the instances you choose and how long they run. Your final bill also depends on the purchasing option, storage, networking, and any paid software your workload uses.

For reference, one Linux m5.large instance in US East (N. Virginia) costs $0.096 per hour On-Demand, or $70.08 for 730 running hours, before storage and networking. The examples below show how to build a complete bill and compare purchasing options.

How AWS EC2 pricing works

Amazon EC2 provides compute capacity through servers called instances. Each instance type has a defined combination of virtual CPUs (vCPUs), memory, and other resources. Your rate also depends on the AWS Region (location), operating system, and tenancy, which determines how physical hardware is shared or dedicated. AWS's instance type guide explains the configurations.

Build your estimate in three stages: find the instance's On-Demand rate, add storage and networking, then compare purchasing options for the usage you expect to keep. On-Demand is the baseline here because it has no long-term commitment.

EC2 instance pricing per hour and month

The examples below show how instance choice affects compute cost. T instances are burstable, M are general purpose, C are compute optimized, and R are memory optimized. Compare each instance's CPU, memory, and application fit alongside its price. In m7g.large, m identifies general-purpose compute, 7 is the generation, g indicates AWS Graviton processors, and large is the size. See AWS instance naming conventions.

Each row is for one Linux instance, using shared tenancy and On-Demand pricing in US East (N. Virginia) (us-east-1). Shared tenancy is the default, where physical hardware can serve multiple AWS accounts. Monthly compute cost is the hourly rate × 730 hours, an estimate for continuous running over an average month.

Rates checked October 18, 2026 against AWS EC2 On-Demand pricing. Prices are in USD and exclude storage, networking, taxes, and additional software charges. Actual billed hours vary.

Equal vCPU counts can produce different results across processor generations and architectures. Arm instances require compatible applications and dependencies. Use our EC2 instance types guide to narrow your choices. Burstable CPU charges are explained with the other costs below.

EC2 pricing by Region

For a Linux m5.large with shared tenancy, On-Demand hourly rates are $0.096 in US East (N. Virginia), $0.107 in Europe (Ireland), and $0.101 in Asia Pacific (Mumbai).

At 730 running hours, monthly compute costs are $70.08, $78.11, and $73.73 respectively. Compare regional storage and networking charges alongside latency and data-residency requirements. Rates checked October 18, 2026 against AWS EC2 On-Demand pricing.

Costs beyond the EC2 instance rate

Your EC2 budget should include compute, storage, networking, and any paid software. These costs have different billing rules, so a compute discount will not reduce every line item by the same percentage.

EBS volumes and snapshots

EBS volumes remain billable while provisioned, including when attached instances are stopped. Snapshots have separate storage charges. A gp3 volume includes 3,000 IOPS and 125 MB/s of throughput; additional provisioned performance increases the bill. In the example below, gp3 storage costs $0.08 per GB-month. Check Amazon EBS pricing.

Data transfer, NAT gateways, and public IPv4

Network charges depend on the traffic path. Cross-Availability-Zone transfers, internet egress, and NAT processing can add costs. AWS provides 100 GB of free internet data transfer out each month, aggregated across eligible services and Regions, excluding China and GovCloud. This allowance is shared across the account. AWS data-transfer pricing explains the applicable rates.

Example rates in us-east-1 are $0.045 per hour plus $0.045 per GB processed for a zonal NAT gateway, and $0.005 per hour for a standard AWS-provided public IPv4 address. At 730 hours, those availability charges are $32.85 and $3.65 respectively. Gateway endpoints can avoid NAT processing for suitable S3 and DynamoDB traffic. Review Amazon VPC pricing.

Load balancer charges for EC2 applications

If your EC2 application uses an Application Load Balancer, add its running time and Load Balancer Capacity Unit (LCU) usage to the estimate. In us-east-1, the rates are $0.0225 per ALB-hour plus $0.008 per LCU-hour. One ALB using a steady 1 LCU for 730 hours would add $22.27, before public IPv4 and data-transfer charges. See AWS Elastic Load Balancing pricing.

Operating systems and software

Windows Server, SQL Server, commercial Linux subscriptions, and Marketplace software can change the total rate. Compare the exact machine image and licensing configuration you plan to run. Check which charges are eligible for your chosen discount.

CPU credits: why a burstable instance can cost more

Unlimited mode can add charges when surplus CPU usage becomes billable. A short spike does not automatically create an extra charge: AWS considers earned credits and average CPU use against the baseline over a rolling 24-hour period, or the instance lifetime if shorter. See AWS Unlimited mode rules.

For example, a Linux t3.micro in us-east-1 costs $7.59 for 730 hours at $0.0104/hour, rounded to cents. If AWS also bills 20 surplus vCPU-hours at $0.05 per vCPU-hour, add $1.00. The compute total becomes $8.59, before storage, networking, and tax. The surplus usage is illustrative; check AWS CPU-credit rates for your instance and operating system.

What does EC2-Other mean on your AWS bill?

EC2-Other is a Cost Explorer service category for associated usage rather than instance running hours. AWS lists EBS volumes and snapshots, NAT gateways, elastic IP addresses, and data transfer among its usage types. It is not a separate EC2 instance or a fixed surcharge. See AWS's explanation of EC2-Other.

To investigate it, open Cost Explorer, choose a date range, filter Service to EC2-Other, and group by Usage type. Then narrow the account or Region as needed. This separates charges such as volume storage, snapshots, and NAT processing so you can identify the resources behind them. AWS documents this Cost Explorer report setup.

How to calculate your monthly EC2 cost

The instance rate is only one part of the bill. Combine it with the resources above to estimate a full month of running your workload:

For the Linux On-Demand examples here, billing is per second with a 60-second minimum. A running instance incurs compute charges even when its application is idle. Stopping it ends standard instance usage charges, while retained EBS storage and other resources can keep billing. Operating-system and software rules can differ. AWS instance billing states explain the details.

For a part-time example, one Linux m5.large in us-east-1 running 8 hours a day for 22 days uses 176 hours. At $0.096 per hour, compute costs $16.90, compared with $70.08 for 730 hours. The instance must be stopped outside the planned running hours to reduce On-Demand compute charges. Any existing commitment payments still apply.

Consider ten Linux m5.large instances in us-east-1, each running for 730 hours with a 100 GB gp3 volume at included baseline performance. They use one zonal public NAT gateway in the same Availability Zone, with one standard Elastic IP address and 100 GB of NAT-processed traffic.

This estimate assumes no chargeable cross-AZ traffic and internet egress within the account's remaining free allowance. It excludes snapshots, load balancers, paid monitoring, support, software subscriptions, and taxes. Rates: EC2, EBS, and VPC.

The supporting resources add $121 per month beyond compute in this configuration. Additional gateways, zones, storage performance, or traffic change the estimate. Keep this $821.80 baseline in mind as you compare discounts on the compute portion.

EC2 pricing models and purchasing options compared

With the On-Demand baseline established, choose how to pay for the compute usage. On-Demand, Savings Plans, Reserved Instances, and Spot suit different workload patterns. Match the option to how steady your usage is, how much flexibility you need, and whether the work can recover from interruptions.

Maximum discounts vary by configuration, term, and payment choice. They are not guaranteed for every instance. Sources: AWS Savings Plans and RI comparison and EC2 Spot overview.

On-Demand: start with uncertain usage

Use On-Demand for new, short-lived, or variable workloads and usage above existing commitments. There is no long-term purchase obligation. For a comparison, hold the instance, Region, operating system, and tenancy constant.

Savings Plans: commit to stable hourly spend

Savings Plans exchange a commitment to eligible usage, measured in dollars per hour, for lower rates over one or three years. Compute Savings Plans provide broader flexibility across eligible EC2 usage; EC2 Instance Savings Plans apply within one family and Region, with flexibility for size, operating system, and tenancy in that scope.

Size the commitment against usage likely to remain after rightsizing, scheduling, and planned migrations. Review AWS Savings Plans pricing and our Savings Plans guide for the configuration you intend to buy.

Reserved Instances: match the configuration and scope

RIs provide discounts when usage matches their attributes and applicable flexibility rules. Some Standard RI attributes can be modified, but Standard RIs cannot be exchanged like Convertible RIs. Convertible exchanges can change configuration, subject to AWS rules including equal or greater value.

Zonal RIs can reserve capacity in a specified Availability Zone. Regional RIs provide billing benefits without reserving capacity. Savings Plans do not reserve capacity. See AWS RI offering types and our Reserved Instances guide.

Spot: use for work that can recover from interruptions

Spot uses spare EC2 capacity and can suit batch processing, build jobs, and distributed applications designed to recover. AWS generally gives a two-minute notice before stopping or terminating a Spot instance, but notices are best effort. Hibernation begins immediately after notification. Design for the interruption rules.

Spot rates vary by instance type and Availability Zone as capacity supply and demand change. Check current AWS Spot pricing and pricing history for your configuration when estimating costs; the maximum discount in the comparison table is not a fixed rate.

Dedicated Hosts and capacity reservations

Dedicated Hosts allocate a physical server to your account for licensing or isolation requirements. Their purchase options include On-Demand, reservations, and eligible Savings Plans. On-Demand Capacity Reservations secure capacity in an Availability Zone and can work with eligible discounts; unused reserved capacity can still incur charges. AWS purchasing options describe the distinctions.

For scheduled GPU workloads, EC2 Capacity Blocks for ML use offer pricing for a reserved period. AWS charges the block price upfront and fixes it once reserved. Operating-system charges may apply when instances run; Savings Plans and RI discounts do not apply to the block. Review Capacity Blocks pricing and billing.

EC2 payment options: No Upfront, Partial Upfront, and All Upfront

Compute and EC2 Instance Savings Plans, plus Standard or Convertible RIs, offer three payment options for their one- or three-year terms. Compare quotes for the same configuration and term, including the initial payment and the full cost of the commitment.

No Upfront still creates a term commitment. Unused Savings Plans and RI charges remain payable. All Upfront prepays the commitment, not storage, networking, or usage outside its coverage. Review the AWS RI payment rules and Savings Plans payment options.

Average monthly commitment cost = (upfront payment + recurring commitment payments over the term) ÷ term months.

For illustration, an EC2 commitment with $120 upfront and $10 per month for 12 months costs $240 over the term, averaging $20 per month. Cash payments are still $120 initially plus $10 each month. This is a calculation example, not an AWS quote.

How coverage and utilization affect EC2 savings

The discount you choose is only part of the savings calculation. Eligible usage must consume the commitment for it to pay off. Coverage measures the share of eligible usage receiving discounted rates. Commitment utilization measures how much of the purchased commitment is consumed. These are billing measures, separate from CPU or memory activity.

Return to the $821.80 monthly example. Assume an illustrative 35% discount on covered compute, full commitment utilization, and unchanged storage and networking costs of $121.

This is a calculation example, not a current Savings Plan quote. At 80% coverage, total-example savings are approximately 23.9%, because storage and networking remain unchanged.

If you also bought $100 of commitments that went unused, the same workload would cost $725.58 and savings against the original bill would fall to approximately 11.7%. Usage can fall or move outside a commitment's eligibility while the purchase remains payable.

For your own calculations, use the On-Demand-equivalent cost of covered usage divided by total eligible On-Demand-equivalent cost for coverage. Calculate utilization as used commitment divided by total purchased commitment, multiplied by 100. AWS documents the coverage formula and utilization definition.

Set commitments around expected stable usage and existing purchases. Account for seasonal demand, migrations, and workloads due for retirement. Our Savings Plans vs Reserved Instances guide covers the trade-offs.

Estimate your own EC2 bill with AWS Pricing Calculator

Now apply the same process to your workload in AWS Pricing Calculator. Use the instance requirements, supporting resources, and purchasing options you have identified:

1. Create an estimate. Find Amazon EC2 in the service list and choose Configure.
2. Set the configuration. Select the Region, operating system, tenancy, instance type, instance count, and expected usage. Use your planned running pattern rather than assuming every instance runs all month.
3. Choose a purchasing option. Start with On-Demand for a baseline. Compare commitment terms and payment options where relevant, checking both upfront and recurring costs.
4. Add supporting resources. Configure EBS volumes and expected data transfer. Add separate service estimates for supporting resources such as NAT gateways and load balancers.
5. Review the summary. Choose Save and add service, then View summary. Recheck quantities, usage assumptions, and which resources the estimate includes.

The result is an estimate, not a final invoice. Check account-specific Free Tier eligibility separately. Follow the AWS calculator instructions.

EC2 Free Tier in 2026

Eligible new AWS customers receive $100 in sign-up credits and can earn up to $100 more through qualifying activities. The Free plan ends after six months or when credits run out, whichever occurs first. Check which EC2 resources are eligible under your account's plan.

Check your credit balance and expiry dates in AWS Settings or Billing and Cost Management. Expiry depends on the sign-up experience and credit type. Remaining eligible credits can apply after upgrading to a Paid plan, subject to their expiry and AWS terms. See the AWS Free Tier FAQs.

The legacy 12-month EC2 offer applied to accounts created before July 15, 2025. Those eligibility windows had ended by October 2026. For ongoing paid usage, budget for the instance rate, storage, and networking separately from any credits. See AWS EC2 Free Tier guidance.

Six steps to reduce your EC2 bill

Use your estimate to find costs you can reduce. Work through usage and configuration changes before buying commitments, so the discounted rate applies to resources you still need:

1. Remove waste. Review unused resources and unnecessary running hours. Check whether existing commitments remain payable and whether other eligible usage can consume them.
2. Rightsize against workload needs. Use CPU, memory, disk, network, and application-performance data, including normal peaks. Low average CPU alone does not establish that an instance is oversized. Use AWS Compute Optimizer to compare recommended instance types and hourly prices, then validate workload performance before changing size.
3. Compare newer families and Arm options. Test application compatibility and cost per request or completed job before changing instance families.
4. Review storage and network paths. Check EBS provisioned performance, snapshot retention, NAT processing, and cross-AZ transfers.
5. Choose purchasing options for the usage that remains. Buy commitments against stable demand, leave uncertain usage On-Demand, and consider Spot for interruptible work.
6. Track net savings. Reassess coverage and utilization after infrastructure changes. Include fees and unused commitment costs.

EC2 memory monitoring requires the CloudWatch agent or another monitoring system; memory metrics are additional to default EC2 metrics. See CloudWatch agent metrics.

Manage EC2 commitment costs with Usage.ai

Usage.ai can help analyze eligible EC2 usage and manage commitments as requirements change. Compare its recommendations with your current EC2 usage, planned changes, and existing commitments.

Fees under Usage.ai's commitment program are a percentage of realized savings. Compare the net reduction in EC2 costs after service fees and any unused commitment charges.