What Is IPv6? Structure, Types, Notation & Why It Matters

What Is IPv6? Structure, Types, Notation & Why It Matters (2026)

IPv6 (Internet Protocol version 6) is the current generation of internet addressing, designed to permanently solve the address shortage that IPv4's 4.3 billion addresses could never satisfy. Standardised in 1998 and elevated to full Internet Standard status in 2017, IPv6 provides roughly 340 undecillion unique addresses — a number so large that address exhaustion becomes mathematically impossible.^[1] As of 2026, IPv6 carries the majority of traffic to major platforms like Google, yet most people have never consciously looked at their own IPv6 address.

⚡ Key Takeaways

IPv6 is a 128-bit address — four times longer than IPv4's 32 bits — providing 2¹²⁸ (~340 undecillion) unique addresses.
Written as eight groups of four hexadecimal digits separated by colons, e.g. 2001:0db8:85a3:0000:0000:8a2e:0370:7334.
Two shortening rules — zero compression (::) and leading zero omission — make addresses far shorter in practice.
IPv6 eliminates the need for NAT — every device can have a globally unique, directly routable address.
As of 2026, roughly 45–47% of global traffic to Google uses IPv6 — adoption has been steady but the transition remains incomplete after 28 years.^[2]
If your device connects to the internet today, it almost certainly already has an IPv6 address — you likely just haven't looked.^[2]

What Is IPv6?

IPv6 is the most recent version of the Internet Protocol — the addressing and routing system that identifies every device connected to the internet. Developed by the Internet Engineering Task Force (IETF) starting in 1994, the first IPv6 RFC was published in 1995, with RFC 2460 formalising the standard in 1998. In July 2017, RFC 8200 superseded that original standard, elevating IPv6 to "Internet Standard" — the highest maturity level the IETF assigns to any protocol.^[1]

Functionally, IPv6 does exactly what IPv4 did — it identifies a network interface and determines where data packets should be routed — but with a vastly larger address space and a redesigned packet structure built for the scale of the modern internet, not the small academic network IPv4 was designed for in 1981.

Why IPv6 Exists: Solving IPv4's Hard Ceiling

IPv4's 32-bit design caps the total address space at exactly 4,294,967,296 — a number that seemed limitless to its designers but proved completely inadequate once the internet moved beyond academic networks. By the late 1990s, with commercial internet growth accelerating, IETF engineers recognised that demand would eventually outstrip the available 4.3 billion addresses entirely.^[3]

By 2026, the scale of the problem is stark: there are over 15 billion internet-connected devices globally — smartphones, laptops, smart TVs, IoT sensors, security cameras, gaming consoles — nearly 4x the entire IPv4 address space.^[4] IPv6's 128-bit design solves this permanently: 2¹²⁸ ≈ 340,282,366,920,938,463,463,374,607,431,768,211,456 addresses — informally "340 undecillion." Every person on Earth could be assigned trillions of unique addresses and the space still would not run out.

IPv6 Address Structure

An IPv6 address is 128 bits, represented as eight groups of four hexadecimal digits ("hextets"), separated by colons. Each group represents 16 bits. Unlike IPv4's decimal notation (0–255 per octet), IPv6 uses hexadecimal — base 16, using digits 0–9 and letters A–F.^[5]

2001

Hextet 1

0db8

Hextet 2

85a3

Hextet 3

0000

Hextet 4

0000

Hextet 5

8a2e

Hextet 6

0370

Hextet 7

7334

Hextet 8

8 groups × 16 bits = 128 bits total. Hex digits are case-insensitive, though lowercase is the IETF-recommended convention.^[1]

Shortening Rules: Zero Compression and Leading Zero Omission

Full IPv6 addresses are long and impractical to write by hand. Two rules — defined in RFCs 2373 and 5952 — shorten the representation without changing the underlying 128-bit value:^[6]

Step 1 — Full address (32 hex digits):

2001:0db8:3c4d:0015:0000:0000:1a2f:1a2b

Step 2 — Rule 1: omit leading zeros within each group (not trailing):

→2001:db8:3c4d:15:0:0:1a2f:1a2b

Step 3 — Rule 2: replace one contiguous run of all-zero groups with :: (only once per address):

→2001:db8:3c4d:15::1a2f:1a2b

⚠️ Critical rule: The double-colon (::) compression can be used only once per address. If two separate runs of zero groups exist, you may compress only one of them — using :: twice makes the address ambiguous and invalid, since the parser cannot determine how many zero groups each :: represents.^[6]

There is no equivalent to IPv4's dotted-decimal subnet mask notation (255.255.255.0) in IPv6 — prefix length is always written in CIDR slash notation, e.g. 2001:db8::/32.

Types of IPv6 Addresses

IPv6 defines several distinct address categories — each serving a different scope and purpose. RFC 4291 establishes three fundamental address types: unicast, anycast, and multicast.^[7]

Global Unicast

2000::/3

Globally routable, publicly reachable addresses — the IPv6 equivalent of a public IPv4 address. Structured with a 48-bit global routing prefix + 16-bit subnet ID + 64-bit interface ID.

Link-Local

fe80::/10

Auto-configured on every interface using the device's MAC address. Only valid on the local network segment — routers never forward link-local traffic. You'll see this on your device even without internet access.

Unique Local

fc00::/7

The IPv6 equivalent of IPv4's private ranges (192.168.x.x). Not globally routable, but globally unique in structure — avoiding collisions if networks merge.

Multicast

ff00::/8

Delivers packets to all interfaces in a multicast group. IPv6 has no broadcast addressing at all — multicast (specifically ff02::1 for all-nodes) replaces that role entirely.^[1]

Anycast

(uses unicast format)

Identifies a group of interfaces — but packets are delivered to whichever member is physically closest to the sender. Used heavily by CDNs and DNS root servers for latency optimisation.

IPv4-Mapped

::ffff:0:0/96

Embeds an IPv4 address inside an IPv6 address for transition compatibility — e.g. ::ffff:192.0.2.128 represents IPv4 address 192.0.2.128. Used in dual-stack transition mechanisms.

💡 Practical tip: If you check your network settings right now and see an address starting with fe80::, that's completely normal — it means IPv6 is active on your device via SLAAC (Stateless Address Autoconfiguration), even if your ISP hasn't assigned a full global IPv6 address yet.^[2]

Prefix Length and Subnetting in IPv6

IPv6 uses CIDR-style prefix notation exclusively — the same slash notation introduced for IPv4 in 1993, but applied from IPv6's birth as the only addressing model (no legacy classful system ever existed for IPv6).^[8]

Component	Bits	Purpose
Global Routing Prefix	48 bits	Identifies the network — assigned by IANA → RIR → ISP → customer in a hierarchy
Subnet ID	16 bits	Internal subnetting within an organisation's allocated prefix
Interface ID	64 bits	Identifies the specific device — often auto-generated from the MAC address

A typical ISP customer allocation looks like 2001:db8:3c4d::/48 — leaving 80 bits (16 for subnetting + 64 for interfaces) for the customer to organise internally. This is vastly more generous than IPv4 subnetting, where every bit was scarce.

IPv6 vs IPv4: Direct Comparison

🔵 IPv4

32-bit — 4.3 billion total addresses
Dotted decimal: 192.168.1.1
Requires NAT/CGNAT for address conservation
Uses ARP for MAC address resolution
Supports broadcast addressing
Address space fully exhausted since 2011–2019

🟢 IPv6

128-bit — ~340 undecillion total addresses
Hexadecimal: 2001:db8::8a2e:370:7334
No NAT required — true end-to-end addressing
Uses NDP (Neighbor Discovery Protocol) instead of ARP
No broadcast — multicast handles that role
Address exhaustion mathematically impossible

IPv6 also simplifies the packet header itself, supports optional jumbograms up to 4,294,967,295 octets (versus IPv4's 65,535-octet payload limit), and was designed with extension headers that allow future protocol features without breaking the core packet structure.^[1]

IPv6 Adoption Status in 2026

IPv6 has technically been available since 1998 — 28 years as of 2026 — yet the transition remains incomplete. World IPv6 Day (2011) and World IPv6 Launch (2012) marked formal industry commitments, and adoption has climbed steadily since. Current measurements show:

~45–47% of global traffic to Google uses IPv6 — one of the most reliable public measurements available.^[2]
APNIC reports ~42% global IPv6 capability, with Asia leading at ~50% and the Americas trailing at ~48%.^[9]
Major broadband ISPs in Europe lead deployment: Sky UK (~86%), Deutsche Telekom (~56%), XS4ALL Netherlands (~73%).
Mobile carriers and content providers have moved fastest; enterprise and public-sector networks lag furthest behind due to legacy hardware with 10–20 year replacement cycles.^[9]

The obstacles to full adoption are organisational, not technical: NAT made IPv4 "good enough" for most use cases, hiding the urgency; dual-stack operation (running both protocols simultaneously) adds operational complexity during the long transition; and a chicken-and-egg dynamic persists where ISPs and content providers each wait for broader adoption before fully committing resources.

IPv6 in Proxy and Web Scraping Infrastructure

IPv6's effectively unlimited address space has a direct, practical implication for proxy infrastructure: IPv6 proxies can be provisioned at massive scale without the cost pressure that plagues IPv4 datacenter proxies sourced from an exhausted, increasingly expensive secondary market.

💸

Lower Cost-Per-IP at Scale

Because IPv6 address blocks are not scarce, providers can offer enormous IPv6 proxy pools at a fraction of the cost of equivalent IPv4 datacenter pools — IPv4 addresses now trade for $22–25+ per IP on the secondary market, while IPv6 allocation carries no comparable scarcity premium.

⚠️

Compatibility Gaps with Target Sites

Not every website or anti-bot system processes IPv6 requests correctly — some older infrastructure silently drops IPv6 traffic or falls back to IPv4 via "Happy Eyeballs" connection racing. Before relying on IPv6 proxies for a specific scraping target, test compatibility directly rather than assuming parity with IPv4.

🏠

Residential IPv4 Still Dominates Trust-Sensitive Use Cases

For scraping, account management, and ad verification — where IP trust and ISP-level reputation matter most — residential IPv4 proxies remain the standard because target sites' anti-bot systems have more mature IPv4 reputation models. Nstproxy's residential proxy network uses real consumer IPv4 (and increasingly IPv6) connections for this reason.

🌐

Dual-Stack Scraping Strategy

Sophisticated scraping pipelines increasingly test both IPv4 and IPv6 paths to the same target, using whichever protocol yields better success rates and lower detection risk for that specific site — a pattern that will become more common as IPv6 adoption continues climbing through the late 2020s.

How to Find Your IPv6 Address

# Windows — Command Prompt
ipconfig
# Look for "IPv6 Address" or "Temporary IPv6 Address" under your adapter

# macOS / Linux — Terminal
ifconfig
# or on newer systems:
ip -6 addr show

# Find your PUBLIC IPv6 (as seen by the internet) — any OS
curl -6 https://ifconfig.me
curl -6 https://api6.ipify.org

# Test if IPv6 connectivity is working at all
ping -6 google.com

If you see an address starting with fe80::, that is a link-local address — valid only on your local network, not your public IPv6. Look for an address starting with 2001: or 2600: for a globally routable address, or use the curl -6 commands above to see exactly what the internet sees.

Need IPv4 or IPv6 Residential Proxies for Scraping?

Nstproxy's 110M+ residential proxy network supports both IPv4 and IPv6 connectivity — ethically sourced, with continuous IP health monitoring across both protocols.

Try Nstproxy for Free →

FAQ

Q: What is an IPv6 address in simple terms?

An IPv6 address is a 128-bit unique identifier assigned to a device on a network, written as eight groups of four hexadecimal digits separated by colons, such as 2001:0db8:85a3::8a2e:0370:7334. It is the successor to IPv4, created specifically to provide a vastly larger address space — roughly 340 undecillion unique addresses — that can never run out.

Q: How many IPv6 addresses are there?

Exactly 2¹²⁸, which equals approximately 340,282,366,920,938,463,463,374,607,431,768,211,456 — commonly described as "340 undecillion." This is more than 100 times the estimated number of atoms on the surface of the Earth, making address exhaustion mathematically impossible for any realistic future demand.

Q: Why does my IPv6 address look different sometimes?

The same 128-bit address can be written in multiple valid shortened forms using zero compression (::) and leading zero omission — all representing the identical underlying address. Additionally, many devices use temporary, randomly generated IPv6 addresses (privacy extensions, RFC 4941) that rotate periodically for privacy, separate from the stable link-local (fe80::) address your interface always has.

Q: Do I already have an IPv6 address?

Almost certainly, yes. Every modern operating system (Windows, macOS, Linux, iOS, Android) and most ISPs support IPv6 by default in 2026. Run ipconfig (Windows) or ifconfig / ip -6 addr show (Mac/Linux) — you will very likely see at minimum a link-local address (starting with fe80::), and quite possibly a full global IPv6 address starting with 2001: or 2600: if your ISP has deployed it.

Q: Can IPv6 and IPv4 work together?

Yes — most networks today run "dual-stack," supporting both protocols simultaneously during the long transition period. Devices and applications use "Happy Eyeballs" logic to race IPv4 and IPv6 connections in parallel and use whichever responds first, avoiding the latency penalty of waiting for a failed IPv6 attempt to time out before falling back to IPv4.