---
title: "AWS Public IPv4 Charge Explained, and How to Cut It"
description: "AWS bills $0.005 an hour for every public IPv4 address, idle or in use. What it covers, how to find it on your bill, and what it means for a static IP."
canonical: https://fixedmark.com/blog/aws-public-ipv4-charge-explained
updated: 2026-11-10
source: Fixedmark (https://fixedmark.com), BitMask LLP
---

Cost · 11 min read

# AWS Public IPv4 Charge Explained, and How to Cut It

Since February 1, 2024, AWS charges $0.005 per hour, about $3.65 a month, for every public IPv4 address in your account, whether it is attached to a resource or idle. That covers Elastic IPs, auto-assigned EC2 addresses, and the addresses AWS services such as load balancers, NAT gateways, and RDS create for you. Addresses you bring through BYOIP are not charged.

## What AWS charges for a public IPv4 address

The rate is $0.005 per address per hour. The [Amazon VPC pricing page](https://aws.amazon.com/vpc/pricing/) lists two lines, one for in-use addresses and one for idle addresses, and both are $0.005. AWS bills in one-second increments with a 60-second minimum.

An in-use address is attached to a resource in a VPC, or assigned to a Global Accelerator or a Site-to-Site VPN tunnel endpoint. An idle address sits in your account and is attached to nothing. Since both cost the same, attaching an address no longer makes it free.

Over a month of 730 hours, one address costs $3.65. Over a year it costs $43.80. The rate is the same in every commercial Region, including Mumbai (`ap-south-1`), and in GovCloud, the China Regions, and Local Zones.

### What is not charged

- Addresses you own and bring to AWS through BYOIP (Bring Your Own IP).
- Customer-owned IP addresses on AWS Outposts.
- Public addresses that are not dedicated to your resource, such as the shared addresses in front of Amazon S3.
- IPv6 addresses. The charge applies to IPv4 only.

The free tier changed in 2025. The [launch announcement](https://aws.amazon.com/blogs/aws/new-aws-public-ipv4-address-charge-public-ip-insights/) gave the EC2 free tier 750 hours of public IPv4 a month for an account's first 12 months. That 12-month tier applies to accounts created before July 15, 2025. Accounts created on or after that date get a credit-based free plan instead, so check your own Billing console rather than assuming an allowance.

## What changed in February 2024

AWS [announced the charge](https://aws.amazon.com/blogs/aws/new-aws-public-ipv4-address-charge-public-ip-insights/) on July 28, 2023, and it took effect on February 1, 2024. Before that date, AWS already billed $0.005 an hour for an idle Elastic IP and for a second Elastic IP on a running instance. What changed is that the first public address on a resource, which used to be free, now costs the same.

_Public IPv4 pricing before and after February 1, 2024, per address per hour_

| Address                                                             | Before | From Feb 1, 2024 |
| ------------------------------------------------------------------- | ------ | ---------------- |
| In-use address on a VPC resource, Global Accelerator, or VPN tunnel | Free   | $0.005           |
| Additional Elastic IP on a running instance                         | $0.005 | $0.005           |
| Idle Elastic IP                                                     | $0.005 | $0.005           |

AWS gave scarcity as the reason. The announcement says the cost to acquire a single public IPv4 address rose more than 300% over the five years before it. AWS framed the charge as an incentive to use fewer IPv4 addresses and to move to IPv6.

In practice, most bills grew by a few dollars per resource, multiplied by every resource that had a public address it did not need. Default VPCs assign a public address to every instance at launch. Container tasks in public subnets often get one each. Those addresses were invisible on the bill before 2024 and now show up as a line item.

## Which resources hold a public IPv4 address

Some of these are obvious, because you allocated them. Others are created by a service on your behalf. AWS calls those service-managed addresses, and they are billed the same way. The [pricing page](https://aws.amazon.com/vpc/pricing/) names Amazon EKS, EMR, ECS, RDS, and WorkSpaces as services that may create public IPv4 addresses in your VPC.

_Common sources of billed public IPv4 addresses_

| Resource                                     | Addresses                          | Notes                                                                                      |
| -------------------------------------------- | ---------------------------------- | ------------------------------------------------------------------------------------------ |
| Elastic IP                                   | 1 each                             | Billed while allocated, attached or not. Release it to stop the charge.                    |
| EC2 instance with an auto-assigned public IP | 1 per instance                     | Default subnets and subnets with auto-assign on. Released when the instance stops.         |
| Public NAT gateway                           | 1 or more per gateway              | Each gateway needs an Elastic IP. Gateway hours and data processing are billed separately. |
| Internet-facing ALB or NLB                   | At least 1 per enabled AZ          | A two-AZ load balancer holds two or more addresses.                                        |
| RDS instance with public access on           | 1                                  | Turn public access off if nothing outside the VPC connects.                                |
| ECS or Fargate task with a public IP         | 1 per task                         | Set by the service's network configuration.                                                |
| Global Accelerator, Site-to-Site VPN         | Per accelerator or tunnel endpoint | Listed as in-use addresses on the pricing page.                                            |

The stop behaviour matters for cost. An auto-assigned EC2 address is released when you stop the instance, so a stopped instance stops paying for it. You also lose the address, and the instance gets a new one when it starts. An Elastic IP stays in your account while the instance is stopped, and it keeps billing as an idle address.

Lambda functions outside a VPC do not hold an address in your account. They egress from AWS-owned addresses that change, which is why they need a NAT gateway or a proxy when a partner asks for a fixed IP. See [why serverless egress IPs change](https://fixedmark.com/blog/why-serverless-egress-ips-change).

## What the charge adds up to

Count addresses, not resources. A small production stack in one Region usually has more public addresses than people expect. Here is one at 730 hours a month.

_Public IPv4 cost of a small stack in one Region, at $0.005 per hour_

| Item                               | Addresses | Per month |
| ---------------------------------- | --------- | --------- |
| Internet-facing ALB in two AZs     | 2         | $7.30     |
| One NAT gateway per AZ, two AZs    | 2         | $7.30     |
| Bastion host with an Elastic IP    | 1         | $3.65     |
| RDS instance with public access on | 1         | $3.65     |
| An Elastic IP nobody released      | 1         | $3.65     |
| Total                              | 7         | $25.55    |

That is about $307 a year for addresses alone. The NAT gateway rows here count only their Elastic IPs. The gateways themselves cost far more in hourly and per-GB charges, which [the hidden costs of a NAT gateway](https://fixedmark.com/blog/nat-gateway-hidden-costs) works through in detail.

Containers scale the number fast. A Fargate service of 20 tasks in public subnets, each with a public IP, holds 20 addresses. That is $73 a month, the same as two NAT gateways and their Elastic IPs for a month before any traffic.

## Find public IPv4 charges on your bill

The charge has two usage types, `PublicIPv4:InUseAddress` and `PublicIPv4:IdleAddress`. They appear under the Amazon Virtual Private Cloud service, not EC2. In Cost Explorer, open the Usage type filter, type `ipv4`, and select both. Expect a delay of a day or two before new usage shows.

The same numbers are available from the CLI. This command groups last month's VPC charges by usage type and keeps only the public IPv4 rows.

**Terminal**

```bash
aws ce get-cost-and-usage \
  --time-period Start=2026-10-01,End=2026-11-01 \
  --granularity MONTHLY \
  --metrics UsageQuantity UnblendedCost \
  --filter '{"Dimensions":{"Key":"SERVICE","Values":["Amazon Virtual Private Cloud"]}}' \
  --group-by Type=DIMENSION,Key=USAGE_TYPE \
  --query "ResultsByTime[].Groups[?contains(Keys[0], 'PublicIPv4')][].[Keys[0], Metrics.UsageQuantity.Amount, Metrics.UnblendedCost.Amount]" \
  --output table
```

Usage quantity is in address-hours. Divide by 730 to get the average number of addresses you held that month. Usage type names usually carry a Region prefix, so expect one row per Region.

### Per resource, from the Cost and Usage Report

Cost Explorer stops at the usage type. To see which resource each charge belongs to, use the Cost and Usage Report with resource IDs turned on. AWS's [guide to identifying public IPv4 usage](https://aws.amazon.com/blogs/networking-and-content-delivery/identify-and-optimize-public-ipv4-address-usage-on-aws/) maps the operation column to the source. `AllocateAddressVPC` is an idle Elastic IP, `AssociateAddressVPC` is an in-use Elastic IP, `RunInstances` is an EC2 public IP, and `DescribeNetworkInterfaces` is a service-managed address.

**Athena query on the Cost and Usage Report**

```sql
SELECT
  line_item_resource_id,
  line_item_operation,
  line_item_usage_type,
  SUM(line_item_usage_amount) AS address_hours,
  SUM(line_item_unblended_cost) AS usd
FROM cur_table
WHERE line_item_usage_type LIKE '%PublicIPv4%'
  AND line_item_usage_start_date >= TIMESTAMP '2026-10-01 00:00:00'
  AND line_item_usage_start_date < TIMESTAMP '2026-11-01 00:00:00'
GROUP BY 1, 2, 3
ORDER BY usd DESC;
```

Replace `cur_table` with your report's Athena table. The column names match both the legacy report and CUR 2.0 from Data Exports.

### Public IP Insights

[Public IP Insights](https://docs.aws.amazon.com/vpc/latest/ipam/view-public-ip-insights.html) is a free view in Amazon VPC IP Address Manager. You create an IPAM first, then open Public IP insights in the IPAM console. It lists every public IPv4 address across Regions, its type (Elastic IP, EC2 public IP, service managed, BYOIP), and the service that owns it, such as a load balancer, NAT gateway, or RDS. It also flags unassociated Elastic IPs so you can release them.

## List every address with the CLI

If you would rather not set up IPAM, two EC2 calls cover most of it. This script walks every enabled Region and prints idle Elastic IPs, then every network interface with a public IPv4 address.

**list-public-ipv4.sh**

```bash
#!/usr/bin/env bash
set -euo pipefail

for region in $(aws ec2 describe-regions --query 'Regions[].RegionName' --output text); do
  echo "== $region"

  echo "Idle Elastic IPs:"
  aws ec2 describe-addresses --region "$region" \
    --query 'Addresses[?AssociationId==null].[PublicIp,AllocationId]' \
    --output text

  echo "Network interfaces with a public IPv4:"
  aws ec2 describe-network-interfaces --region "$region" \
    --query 'NetworkInterfaces[?Association.PublicIp!=null].[Association.PublicIp,InterfaceType,NetworkInterfaceId,Description]' \
    --output text
done
```

The interface type and description tell you who owns each address. NAT gateways show the `nat_gateway` type, and the description of a service-managed interface names the load balancer, gateway, or database behind it. The script reads the primary address on each interface. An interface with extra Elastic IPs on secondary private addresses holds more than one, so cross-check the count with Public IP Insights.

### Stop paying for the ones you do not need

1. 1

   **Release idle Elastic IPs** Run `aws ec2 release-address --allocation-id eipalloc-0123456789abcdef0` for each one. Check first that no partner allowlist names the address, because you cannot get a released Elastic IP back.
2. 2

   **Turn off auto-assign on subnets** Find subnets that hand out public IPs with `aws ec2 describe-subnets --filters Name=map-public-ip-on-launch,Values=true`, then run `aws ec2 modify-subnet-attribute --subnet-id subnet-0123456789abcdef0 --no-map-public-ip-on-launch`. New instances stop getting addresses. Existing ones keep theirs.
3. 3

   **Turn off public access on databases** Set RDS instances that only serve the VPC to not publicly accessible. Do it during a maintenance window and confirm no external client connects first.
4. 4

   **Move tasks to private subnets** ECS and Fargate tasks that only make outbound calls can run in private subnets behind a NAT gateway. Run the cost check in the next section first, because the NAT gateway can cost more than the addresses it replaces.

## Cut the address count without breaking things

AWS's [EC2 addressing guide](https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/using-instance-addressing.html) lists the main levers. Each one trades the charge for some other cost or constraint.

- Put instances behind a load balancer and turn off their public addresses. One load balancer replaces a public address per instance. Consolidating several load balancers into one cuts more.
- Use EC2 Instance Connect Endpoint for SSH to private instances instead of a bastion with an Elastic IP.
- Use PrivateLink or VPC endpoints for traffic to AWS services, so private instances do not need an internet path for it.
- Use Global Accelerator endpoints to reach instances in private subnets, or bring your own range to it.
- Bring your own IPv4 range through BYOIP if you already own one. Those addresses carry no charge.

Moving instances to private subnets behind a NAT gateway is not automatically cheaper. The EC2 guide gives a formula. Add NAT gateway hours (`$0.045 x 730 x AZs`), NAT gateway addresses (`$0.005 x 730 x IPs`), and data processing (`$0.045 x GB`), then divide by what your current public addresses cost (`$0.005 x 730 x addresses`). Below 1, the NAT gateway wins. At the guide's $0.045 hourly rate, one NAT gateway costs about $32.85 a month in hours alone, as much as nine public addresses. The [NAT gateway cost calculator](https://fixedmark.com/tools/nat-gateway-cost-calculator) runs the numbers for your traffic.

### IPv6-only load balancers

Since May 2024 an internet-facing Application Load Balancer can use the `dualstack-without-public-ipv4` address type. [AWS's launch post](https://aws.amazon.com/blogs/networking-and-content-delivery/introducing-dual-stack-without-public-ipv4-application-load-balancer) explains that the load balancer then has no public IPv4 addresses. Clients reach it over IPv6 only, and it still uses private IPv4 to talk to targets. The subnets must be dual-stack with an IPv6 route to an internet gateway.

**Terminal**

```bash
aws elbv2 set-ip-address-type \
  --load-balancer-arn arn:aws:elasticloadbalancing:ap-south-1:111122223333:loadbalancer/app/my-alb/0123456789abcdef \
  --ip-address-type dualstack-without-public-ipv4
```

This fits when every client has IPv6, such as a CDN that supports IPv6 origins in front of the load balancer. It breaks any client that only has IPv4, and many office, cloud, and partner networks still are.

## What the charge means for a static IP

On AWS, a static public IPv4 address is an Elastic IP. It is the address you give a bank, a broker, or a partner API to put on an allowlist. Since 2024 it costs $3.65 a month from the moment you allocate it until you release it, attached or not. That is a small number, and it is the floor, not the total.

The address is rarely the expensive part. An EC2 instance with an Elastic IP pays for the instance too. Lambda, Fargate, or anything else in private subnets reaches the internet from a fixed address only through a NAT gateway, and that costs hourly and per GB on top of its Elastic IP. A partner that wants two addresses for failover doubles the address cost and usually means a gateway in each of two AZs. See [why one static IP is a single point of failure](https://fixedmark.com/blog/static-ip-pairs-and-failover).

If partners keep asking for neat ranges, IPAM can give you an Amazon-provided contiguous IPv4 block and allocate Elastic IPs from it. That is billed differently. The whole block costs $0.008 per address per hour from the moment it is provisioned, whether each address is used or not. The pricing page's example puts a /30 block of four addresses at $23.04 a month.

Do not respond to the charge by allowlisting AWS's published ranges instead of your own address. Those ranges admit every other AWS customer. See [why cloud IP ranges are not an allowlist](https://fixedmark.com/blog/cloud-ip-ranges-are-not-an-allowlist).

IPv6 avoids the charge, but most allowlists you need a static IP for still take IPv4 only. Broker consoles, bank onboarding forms, and payment payout dashboards usually ask for an IPv4 address. If your platform prefers IPv6 and the partner wants IPv4, read [IPv6 egress and IPv4-only allowlists](https://fixedmark.com/blog/ipv6-egress-and-ipv4-only-allowlists).

## Doing this with Fixedmark

Fixedmark is a static IP proxy, in early access. Instead of an Elastic IP and a NAT gateway in your VPC, your code sends the calls that need a fixed address through a proxy URL, set as `FIXEDMARK_PROXY_URL`. The partner allowlists a pair of IPv4 addresses that Fixedmark holds, so they do not appear as public IPv4 lines on your AWS bill.

This suits workloads that only need a fixed IP for a few outbound APIs, such as Lambda calling a payout API. The function can stay outside a VPC with no NAT gateway. The planned launch price for a dedicated IP pair is $29 a month on [Pro](https://fixedmark.com/pricing), with 100 GB of bandwidth, in [Mumbai, Singapore, Newark, and Ashburn](https://fixedmark.com/regions). Prices are planned and may change. The [AWS NAT gateway comparison](https://fixedmark.com/compare/aws-nat-gateway) puts both options side by side.

It does not replace the Elastic IP on a server that must accept inbound connections, or a NAT gateway for compute that needs general internet access. For those, the AWS charge is the cost of the address, and the cleanup steps above are what bring it down. After you switch, [verify your egress IP](https://fixedmark.com/docs/verify-egress-ip) before you tell a partner to update their allowlist.

## Sources

Checked November 2026. Confirm current details on the vendor's site before you decide. Prices and limits change.

- [AWS: Amazon VPC pricing, Public IPv4 Address tab](https://aws.amazon.com/vpc/pricing/)
- [AWS News Blog: New AWS public IPv4 address charge and Public IP Insights](https://aws.amazon.com/blogs/aws/new-aws-public-ipv4-address-charge-public-ip-insights/)
- [AWS Networking blog: Identify and optimize public IPv4 address usage on AWS](https://aws.amazon.com/blogs/networking-and-content-delivery/identify-and-optimize-public-ipv4-address-usage-on-aws/)
- [Amazon EC2 User Guide: EC2 instance IP addressing](https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/using-instance-addressing.html)
- [Amazon VPC IPAM User Guide: View public IP insights](https://docs.aws.amazon.com/vpc/latest/ipam/view-public-ip-insights.html)
- [AWS Networking blog: Dual-stack without public IPv4 Application Load Balancer](https://aws.amazon.com/blogs/networking-and-content-delivery/introducing-dual-stack-without-public-ipv4-application-load-balancer)
- [ELB User Guide: Update the IP address types for your Application Load Balancer](https://docs.aws.amazon.com/elasticloadbalancing/latest/application/load-balancer-ip-address-type.html)
- [Amazon EC2 User Guide: Track your Free Tier usage](https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-free-tier-usage.html)
- [Amazon VPC User Guide: IP addressing for your VPCs and subnets](https://docs.aws.amazon.com/vpc/latest/userguide/vpc-ip-addressing.html)
- [Vantage: How to see AWS public IPv4 address costs](https://vantage.sh/blog/aws-public-ipv4-cost-and-how-to-see)

Keep reading

## Related guides

- [Comparison AWS NAT Gateway The real monthly cost of a highly available AWS NAT Gateway, and when a static IP proxy is simpler.](https://fixedmark.com/compare/aws-nat-gateway)
- [Integration AWS Lambda A fixed IP for Lambda without VPC, subnets, and two NAT Gateways.](https://fixedmark.com/integrations/aws-lambda)
- [Tool NAT gateway cost calculator Monthly NAT gateway, data processing, and public IPv4 cost for your traffic.](https://fixedmark.com/tools/nat-gateway-cost-calculator)
- [Use case Partner API allowlisting Give Salesforce, SOAP vendors, and payment gateways one dedicated IP pair to allowlist, wherever your code runs.](https://fixedmark.com/use-cases/partner-api-allowlisting)
- [Guide Pricing Planned launch prices for shared and dedicated IP pairs.](https://fixedmark.com/pricing)

## Frequently asked questions

### How much does a public IPv4 address cost on AWS?

$0.005 per hour, which is about $3.65 a month or $43.80 a year per address. The rate is the same for in-use and idle addresses and the same in every commercial Region.

### Is an Elastic IP attached to a running instance still free?

No. Since February 1, 2024, every public IPv4 address is charged, including the first Elastic IP on a running instance. Before that date only idle and additional Elastic IPs were charged.

### Does a stopped EC2 instance keep paying for its public IP?

An auto-assigned public IP is released when the instance stops, so the charge stops too, and the instance gets a new address on start. An Elastic IP stays allocated while the instance is stopped and keeps billing as an idle address.

### Where do public IPv4 charges appear on the AWS bill?

Under Amazon Virtual Private Cloud, as the usage types \`PublicIPv4:InUseAddress\` and \`PublicIPv4:IdleAddress\`, with a Region prefix. Use the Cost and Usage Report with resource IDs to see which resource each charge belongs to.

### Are IPv6 addresses or BYOIP addresses charged?

No. The charge covers Amazon-provided public IPv4 addresses only. IPv6 addresses and IPv4 addresses you bring through BYOIP are not charged.
