What is a public IP address?

Every device that talks to the Internet needs an address so replies can find their way back. The address the rest of the Internet actually sees is your public IP address, and it usually belongs to your network rather than to your device.

Public and private addresses

IP addresses come in two broad flavours. Private addresses are reserved for use inside a local network and are never routed across the public Internet. The reserved IPv4 ranges are defined in RFC 1918, and you have almost certainly seen them: 10.0.0.0/8, 172.16.0.0/12 and 192.168.0.0/16. Your laptop, phone, printer and smart speaker typically each hold one of these.

A public address is globally routable. It is allocated to a network operator by a regional Internet registry, and the operator assigns it to a customer connection. When your router connects, it receives one — or a small block — of these, and that is the address a website records when you visit it.

Private address: visible only inside your home or office network.
Public address: visible to every server you connect to.
Shared address space (100.64.0.0/10, RFC 6598): used by some ISPs and mobile networks for carrier-grade NAT, so a share of subscribers see an address that is neither truly private nor uniquely theirs.

How NAT ties the two together

Network address translation, described in RFC 3022, is the mechanism that lets many private addresses share a single public one. Your router rewrites the source address of outgoing packets to the public address, remembers which internal device each conversation belongs to, and reverses the translation for the replies.

That is why a household of ten devices commonly presents one IP address to the world. It is also why an IP address rarely identifies a device. On carrier-grade NAT, mobile networks and large campus networks, a single public address can represent hundreds or thousands of subscribers at once.

What a website can infer from it

A server always sees the source address of the connection, because without it there is nowhere to send a response. From that address it can look up publicly registered facts: which network operator holds the address block, the autonomous system number that announces it, and any hostname published in reverse DNS.

Commercial databases add an estimated location, usually at country or city level and often less precise than that. Those estimates come from registration and routing data, not from any measurement of where you physically are.

What it does not reveal

An IP address is not a name, an account, or a street address. On its own it does not disclose who you are, what device you are using, what else you have browsed, or your exact location. Those things become linkable only when combined with other information — an account you signed into, a cookie, or records held by your provider.

It is also not permanent for most home connections. Many ISPs reassign addresses periodically, so the address you have today may belong to a different subscriber next month. Treating an IP address as a stable identity is a common and consequential mistake.

Why your address changes

Most home addresses are handed out by the provider rather than owned by the customer, so they can change. What actually triggers a change is a provider policy decision, not a fixed rule of the protocol. DHCP, defined in RFC 2131, gives each address a lease with a limited lifetime; the client renews it well before it expires, and a client that already has an address asks for that same address again. A server that still holds the binding is expected to grant it, which is why renewing a lease normally leaves the address unchanged.

Restarting a router does not reliably change anything either. If the equipment comes back before the lease and the server's record expire, the same address is usually reissued. An address typically changes when the provider's binding is released or expires, when the provider deliberately rotates allocations, when line equipment or the account's network segment changes, or when the connection moves onto carrier-grade NAT and shared address space. DHCPv6 (RFC 8415) works the same way for IPv6, though IPv6 networks more often change the prefix delegated to your router.

Renewing a lease: usually keeps the same address.
Router restart: often keeps the same address; it depends on whether the binding survived.
Provider reassignment, equipment change, or a move to shared address space: the address changes.
Mobile data and VPN servers: a different address on most reconnections, by design.

Why it still matters

Your public address is the first thing every service you contact learns about you, and it is the doorstep through which anything exposed on your network is reachable. Knowing what it is, which network it belongs to, and whether anything is listening behind it is a reasonable baseline for understanding your own exposure.

Check your own connection

Everything above describes how this works in general. These tools show what your own connection currently looks like.

My IP — Shows the public address this connection presents, with its network operator, ASN and reverse DNS hostname when one exists.
Internet Exposure — Checks a short list of common ports on your public address to see which ones answer from the outside.