A public IP address is an IP address intended to be globally routable or otherwise usable for communication across the public Internet, subject to routing policy and network configuration. In IPv4, public addresses are addresses that are not reserved for private or other special-purpose uses. In IPv6, most globally routable unicast addresses are globally unique and can serve as public Internet addresses, although IPv6 has its own categories of special-purpose and non-global addresses.
A public IP address identifies an Internet-facing network interface or endpoint at the IP layer and allows routers across independently administered networks to forward packets towards it when appropriate routes exist. “Public” does not mean that a particular service is necessarily accessible from every Internet host: firewalls, access-control lists, NAT, service configuration and routing policies can prevent inbound access even when an address is globally routable.
Public IPv4 addresses are allocated through a hierarchical system involving the Internet Assigned Numbers Authority and Regional Internet Registries, which allocate address resources within their respective regions and policy frameworks. Internet service providers, hosting providers, cloud platforms, enterprises and other organisations can receive or lease public address space through these arrangements. Public addresses must be unique within the relevant global routing system to avoid ambiguity.
IPv4 scarcity has made public addresses valuable resources, contributing to widespread use of private addressing and NAT. IPv6 was designed with a much larger address space, making globally unique addressing practical on a vastly larger scale, although address allocation and routing policies still determine how addresses are used. An address being globally unique does not by itself mean that it is permanently assigned to a particular person, device or service.
A public IP address can be static or dynamic. A static public address remains assigned according to a provider's or organisation's configuration and is useful for services that need a stable endpoint, such as certain VPN gateways, mail servers, hosted websites or remote-access systems. A dynamic public address can change when an ISP reconnects a customer, renews an allocation or otherwise changes its addressing configuration.
Dynamic DNS services can associate a changing public address with a stable hostname, allowing services to be located without requiring a permanently fixed address. Cloud environments commonly provide several distinct concepts, such as automatically assigned public addresses, reserved or static addresses, load-balancer addresses and service-specific endpoints. Consequently, “having a public IP” does not necessarily mean that the address is permanently tied to one physical machine.
In a typical home IPv4 network, the public IP address is usually assigned to the Internet-facing interface of the household router. Internal devices receive private addresses such as 192.168.1.20 from DHCP, while the router performs NAT when they communicate with Internet destinations. External servers generally see the router's public IPv4 address rather than the internal private address. This explains why several computers and phones can appear to Internet services to originate from the same public IPv4 address. With IPv6, a device may instead receive a globally unique address and communicate without conventional IPv4-style NAT, although a firewall can still prevent unsolicited inbound connections. The distinction between public and private addressing therefore depends strongly on the IP version and network architecture.
Public IP addresses are relevant to routing because Internet routers need route information describing where globally reachable prefixes can be found. The global routing system, principally using the Border Gateway Protocol, or BGP, exchanges reachability information between autonomous systems. A public IP address is therefore not simply “an address that can go online”; its prefix must be appropriately routed and the relevant network must accept and forward traffic.
Route announcements can be filtered, withdrawn or changed, and organisations can use provider-independent or provider-assigned address space under different arrangements. Even a correctly allocated public address can become unreachable because of routing problems, upstream outages or policy filters. Global reachability is consequently a property of addressing plus routing and network policy rather than of the numeric address alone.
Public addresses can expose useful information without necessarily identifying a specific individual. An IP address may be associated through databases with an ISP, autonomous system, country or approximate region, but IP geolocation is not inherently precise enough to establish a person's exact physical location. Public addresses can also be shared by multiple users through NAT or CGNAT, meaning that an Internet service observing one address may be seeing many customers.
Conversely, a dedicated public address can be associated with a specific server, business connection or other network endpoint. Logs often record source IP addresses for security, troubleshooting and abuse investigation, but interpreting those logs requires contextual information such as timestamps, ports, NAT mappings and provider records. An IP address should therefore not automatically be treated as a unique human identity.
Public IP addresses are not synonymous with websites, servers or Internet connections. A website can be reachable through multiple public IP addresses, for example through load balancing, anycast, content delivery networks or multiple network interfaces. Multiple domains can share the same public IP address through virtual hosting. A home Internet connection can have a public address without hosting any publicly accessible service because its firewall blocks unsolicited inbound traffic.
Cloud services can also expose public addresses while placing application instances behind load balancers or reverse proxies. These examples show that the relationship between a public IP address and an application endpoint is architectural rather than one-to-one. DNS normally provides the human-friendly mapping between domain names and IP addresses, but DNS itself does not make an address public.
The key distinction is that a public IP address is intended to participate in globally routable addressing, whereas private and special-purpose addresses are intended for restricted scopes. Public does not mean unrestricted, permanent, confidential or inherently secure. Public addressing enables direct Internet routing where routing and security policy permit it, while firewalls and other controls determine which traffic is actually accepted.
IPv4 public-address scarcity has encouraged private addressing, NAT and address sharing, whereas IPv6 makes globally unique addressing much more practical. Understanding these distinctions is important when configuring servers, firewalls, VPNs, DNS, cloud infrastructure and remote-access services because an endpoint's public addressing status is only one component of whether it can actually communicate with the Internet.
