What Is IPv4? Structure, Classes, Exhaustion & How It Works (2026)
IPv4 (Internet Protocol version 4) is the addressing system that has identified devices on the internet since the early 1980s — a 32-bit number, written as four decimal numbers separated by dots, that tells data packets where to go. Despite being designed decades before smartphones, cloud computing, or IoT existed, IPv4 still routes the majority of internet traffic in 2026, kept alive well past its intended lifespan by a series of engineering workarounds.
⚡ Key Takeaways
- IPv4 is a 32-bit address, providing exactly 4,294,967,296 (≈4.3 billion) unique addresses — a hard mathematical ceiling.[1]
- Written in dotted decimal notation: four numbers (0–255) separated by dots, e.g.
192.168.1.1. - The original classful system (Classes A–E) was replaced by CIDR in 1993 — but classes still appear in documentation, certifications, and legacy systems.
- The global free pool of IPv4 addresses was fully exhausted by January 2011; all five Regional Internet Registries had depleted their pools by 2019.[2]
- As of early 2026, only ~45-47% of global internet traffic uses IPv6 — IPv4 still carries the majority of traffic, sustained by NAT and CGNAT.[3]
- IPv4 address scarcity created a real secondary market: a
/24block (256 addresses) sold for $5,600–$6,500 in February 2026 auctions.[4]
What Is IPv4?
IPv4 is the fourth version of the Internet Protocol — the foundational addressing system that gives every device on a network a unique numerical identifier so data knows where to go. Developed in the late 1970s and standardised in 1981 (RFC 791), IPv4 uses a 32-bit binary number for every address. Computers process this number in binary; humans read it in dotted decimal notation — four groups of digits (0–255) separated by periods, such as 172.16.254.1.[5]
Every IPv4 address serves two purposes simultaneously: it identifies which network a device belongs to, and which specific device on that network is being addressed — similar to how a street address identifies both the street and the house number. This is the network/host split, and it is the foundation for almost every other concept in IPv4 networking: subnetting, routing, NAT, and the original class system.
IPv4 Address Structure: Octets and Binary
An IPv4 address is 32 bits, split into four 8-bit segments called octets. Each octet can represent a decimal value from 0 to 255 (2⁸ = 256 possible values per octet). The example below shows 192.168.43.1 broken into its binary form:
192.168.43.1 → binary: 11000000.10101000.00101011.00000001
Key structural rules: each octet's value ranges from 0–255 only; leading zeros are invalid (054 is wrong, 54 is correct); and the total address space is exactly 2³² = 4,294,967,296 unique addresses — a number that seemed limitless in 1981 and proved completely insufficient by the 2010s.[1]
IPv4 Address Classes (A–E)
From 1981 to 1993, IPv4 used a classful addressing system that divided the entire address space into five fixed classes — A through E — identified by the binary pattern of the first octet. This let routers instantly determine network size without any extra configuration, which mattered when processing power was scarce.[6]
Class A
126 networks, ~16.7M hosts each. Mask: 255.0.0.0. Used by ISPs and massive organisations (early internet allocations to Google, Apple, IBM).
Class B
16,384 networks, 65,534 hosts each. Mask: 255.255.0.0. Used by universities and mid-large corporations.
Class C
2M+ networks, 254 hosts each. Mask: 255.255.255.0. The workhorse class — small businesses, home routers (192.168.x.x).
Class D
Reserved for multicast — one-to-many communication. No network/host split; not assigned to individual devices.
Class E
Reserved for experimental use by IANA. Never deployed publicly. Still unroutable on the internet today.
Special: Loopback
Reserved for local testing. 127.0.0.1 ("localhost") always refers to the device itself, regardless of network.
Public vs Private IPv4 Addresses
Not every IPv4 address is reachable from the public internet. Three blocks are reserved (RFC 1918) exclusively for private, internal networks:
| Private Range | Class | Typical Use |
|---|---|---|
10.0.0.0 – 10.255.255.255 | A | Large corporate networks, cloud VPCs |
172.16.0.0 – 172.31.255.255 | B | Medium business networks, Docker default range |
192.168.0.0 – 192.168.255.255 | C | Home routers — almost every home network uses this |
Devices on these ranges cannot be reached directly from the internet — they communicate externally through Network Address Translation (NAT), which maps many private addresses to one shared public address. Every other IPv4 address outside these reserved blocks is potentially public and routable.
CIDR and Subnetting
CIDR (Classless Inter-Domain Routing) replaced rigid class boundaries with a flexible prefix notation: 192.168.1.0/24. The number after the slash indicates how many bits represent the network portion — the remaining bits represent hosts.
| CIDR Prefix | Subnet Mask | Usable Hosts | Typical Use |
|---|---|---|---|
| /8 | 255.0.0.0 | 16,777,214 | Equivalent to old Class A |
| /16 | 255.255.0.0 | 65,534 | Equivalent to old Class B |
| /24 | 255.255.255.0 | 254 | Equivalent to old Class C — most common home/office subnet |
| /27 | 255.255.255.224 | 30 | Small department or VLAN |
| /30 | 255.255.255.252 | 2 | Point-to-point links |
CIDR sizing reference per Meridian Outpost IPv4 classes guide.
IPv4 Exhaustion: How 4.3 Billion Addresses Ran Out
IPv4's 32-bit design provides exactly 4,294,967,296 addresses — a number Vint Cerf, co-creator of TCP/IP, has admitted he assumed would be more than sufficient for what he believed was an experimental protocol.[1] It was not. The timeline of exhaustion unfolded as follows:
/24 block (256 addresses) sells for $5,632–$6,528 at auction — roughly $22–25 per address.[4]The shortage didn't break the internet — it forced a series of workarounds (NAT, CIDR, address trading) that have kept IPv4 functional for 15 years past its theoretical exhaustion date, at the cost of growing complexity and a thriving secondary market for legacy address blocks.
NAT and CGNAT: How IPv4 Survives the Shortage
Network Address Translation (NAT) is the single biggest reason IPv4 exhaustion didn't cause a visible internet-wide failure. NAT allows many devices on a private network to share one public IPv4 address — your home router does this for every device on your Wi-Fi network right now.
Carrier-Grade NAT (CGNAT) takes this further: ISPs place multiple customers (not just devices) behind one shared public IPv4 address, conserving the ISP's limited address pool. This comes with real costs:
- Broken port forwarding — you cannot reliably forward ports for self-hosted services or gaming servers behind CGNAT.
- Gaming NAT-type issues — multiplayer games report "Strict NAT," limiting peer-to-peer connections.
- Shared IP reputation risk — if another customer sharing your CGNAT address sends spam or gets flagged, your traffic inherits that reputation.
- VoIP quality degradation — double NAT can introduce latency and connection issues for real-time communication.[3]
IPv4 vs IPv6: The Comparison
🔵 IPv4
- 32-bit address — 4.3 billion total addresses
- Dotted decimal notation:
192.168.1.1 - Requires NAT for address conservation
- Classful legacy system (now CIDR-based)
- Still carries the majority of global traffic in 2026
- Address space fully exhausted since 2011–2019
🟢 IPv6
- 128-bit address — ~340 trillion trillion trillion addresses
- Hexadecimal notation:
2001:0db8:85a3::8a2e:0370:7334 - No NAT required — true end-to-end addressing
- Built-in multicast support, simplified headers
- ~45–47% of global traffic to Google as of 2026[3]
- Address exhaustion mathematically impossible
Despite being available since 1998 — 28 years as of 2026 — IPv6 has not fully replaced IPv4. The obstacles are not technical: legacy hardware with 10–20 year lifecycles, dual-stack complexity during transition, and a chicken-and-egg problem where ISPs and content providers each wait for the other to move first.[3] Most networks today run dual-stack — supporting both protocols simultaneously during the (very long) transition.
IPv4 Addresses in Proxy and Scraping Infrastructure
IPv4 scarcity has a direct, practical impact on proxy infrastructure and web scraping. Because the public IPv4 pool is fixed and fully allocated, every IPv4-based proxy — datacenter, residential, ISP, or mobile — draws from a finite, increasingly expensive resource.
Rising Datacenter Proxy Costs
Datacenter proxy providers purchase or lease IPv4 blocks on the open secondary market — where prices reached $22–25 per address in early 2026. This cost pressure is part of why datacenter proxy pricing has remained relatively stable rather than falling, despite increased competition.
Why Residential Proxies Don't Face the Same Scarcity
Residential proxy networks don't purchase IPv4 blocks outright — they route traffic through real consumer devices that already have an ISP-assigned IP (often already behind CGNAT). This is structurally different from datacenter IPv4 acquisition and explains why residential proxy pools can scale to 100M+ IPs without buying address blocks. See Nstproxy's residential proxy sourcing guide.
CGNAT and Shared Residential IP Reputation
Because many ISPs use CGNAT, a single public IPv4 address may serve hundreds of customers simultaneously. For residential proxy use, this means a "residential IP" might already be shared infrastructure at the ISP level — which is why provider-side IP health monitoring (tracking abuse signals per session, not just per IP) matters more than raw IP exclusivity.
IPv6 Proxies: An Emerging Low-Cost Alternative
Because IPv6 address space is virtually unlimited, IPv6 proxies are significantly cheaper to provision at scale than IPv4 datacenter proxies. The tradeoff: many older target websites and anti-bot systems still don't process IPv6 requests correctly, requiring compatibility testing per target before relying on IPv6 proxy pools.
How to Find Your IPv4 Address
# Windows — Command Prompt ipconfig # Look for "IPv4 Address" under your active network adapter # macOS / Linux — Terminal ifconfig # or on newer systems: ip addr show # Find your PUBLIC IPv4 (as seen by the internet) — any OS curl -4 https://ifconfig.me curl -4 https://api.ipify.org
Your local IPv4 (from ipconfig/ifconfig) is typically a private address (192.168.x.x or 10.x.x.x). Your public IPv4 — what websites actually see — is different and is shown by the curl -4 commands above or any "what is my IP" service.
Need Clean IPv4 Addresses for Scraping or Automation?
Nstproxy's 110M+ residential IPv4 addresses are sourced ethically — not purchased from the scarce secondary market — giving you scale without the cost pressure of datacenter IPv4 acquisition.
Try Nstproxy for Free →FAQ
An IPv4 address is a unique numerical label assigned to a device on a network, written as four numbers (0–255) separated by dots, such as 192.168.1.1. It identifies both the network a device belongs to and the specific device on that network, allowing data to be routed to the correct destination across the internet.
Exactly 4,294,967,296 (2³²) — approximately 4.3 billion. This is a hard mathematical limit set by IPv4's 32-bit address format. The global free pool of these addresses was fully allocated by 2011 at the IANA level, and all five Regional Internet Registries had exhausted their individual pools by 2019.
No, not for actual address allocation. Classful addressing (Classes A–E) was replaced by CIDR (Classless Inter-Domain Routing) in 1993 because the fixed class sizes wasted enormous numbers of addresses. Classes still appear in networking certifications, legacy documentation, and as a foundational teaching concept, but no IP addresses are allocated by class today — modern networks use CIDR prefix notation like /24 or /27.
A private IPv4 address (from reserved ranges like 192.168.x.x, 10.x.x.x, or 172.16-31.x.x) is only valid within a local network and cannot be reached directly from the internet. A public IPv4 address is globally unique and routable across the internet. Most devices use a private address locally and share a public address via NAT when communicating externally.
IPv6 has been available since 1998, but as of 2026 only about 45–47% of global traffic to major services uses it. The barriers are organisational and economic rather than technical: legacy hardware with long replacement cycles, the cost and complexity of running dual-stack networks during transition, and a chicken-and-egg dynamic where ISPs and content providers each wait for broader adoption before fully committing. NAT also made IPv4 "good enough" for most use cases, reducing urgency.
Further Reading
Sources
- Wikipedia — IPv4 Address Exhaustion (May 2026)
- Wikipedia — RIR Exhaustion Timeline
- TrustMyIP — IPv4 Exhaustion: Why IPv6 Matters (April 2026)
- Grokipedia — IPv4 Address Exhaustion Market Data (April 2026)
- ClouDNS — What Is IPv4? Everything You Need to Know
- RouterHax — IPv4 Address Classes Explained (April 2026)
- GeeksforGeeks — Classful IP Addressing (April 2026)
- CircleID — The Internet's Address Crisis (Jan 2026)
