Global IPv6 Adoption in 2026: A Research Summary
Updated Published by Kishan Prajapat, SEO & Content Lead
This is a summary of public sources, listed under Data Sources at the end, not original research. Figures are only given where we can link to where they came from.
Summary
IPv6 crossed a milestone in 2026: on 28 March, more than half of users reaching Google did so over IPv6 (50.10%), according to APNIC. APNIC’s own measurement, which uses a different method, puts worldwide IPv6 capability at around 42%, and the two figures together bracket the likely range. Adoption is very uneven. The Internet Society lists France (73%) and India (72%) among the leaders, while countries such as Italy (17%) and Spain (10%) lag far behind. This summary looks at why adoption varies and at the routing and configuration problems that can make IPv6 connections perform worse than IPv4 even where both are available.
Key Findings
- 1A majority milestone: over 50% of Google users reached it over IPv6 by March 2026, though measurements differ by method, and APNIC’s capability figure is lower (about 42%).
- 2Mobile networks lead: large mobile operators have moved to IPv6-only cores and translate to IPv4 at the edge (NAT64/464XLAT), which drives much of the growth in countries like India.
- 3Enterprise networks lag: corporate networks move more slowly because of legacy applications, firewalls and monitoring that assume IPv4.
Methodology
This summary draws on published IPv6 statistics from Google and APNIC and on Internet Society Pulse reporting (see Data Sources). It doesn’t include an original measurement. You can compare IPv4 and IPv6 paths to a service yourself with our traceroute tool.
Analysis
Where IPv6 performs worse than IPv4, the cause is usually routing rather than the protocol. Decades of traffic engineering have produced short, well-peered IPv4 paths, while some networks still don’t peer IPv6 locally and send it to a distant transit hub instead. Traffic between two cities in the same country can then leave the country over IPv6 while staying local over IPv4. A second common problem is path MTU discovery: IPv6 routers don’t fragment packets, so if a firewall blocks ICMPv6 “Packet Too Big” messages, larger packets silently disappear and connections stall in ways that look like outages. Browsers limit the user impact with Happy Eyeballs, falling back to IPv4 when IPv6 is slow, which also means these problems can go unnoticed. Our guide to static and dynamic IP addresses covers how ISPs assign addresses.
Industry Insights
Mobile operators and large ISPs have the strongest incentive to move, because IPv4 addresses are scarce and expensive to buy or lease. Banking, healthcare and other sectors with long-lived internal systems move more slowly, often keeping IPv4 internally and translating at the edge, which delays full end-to-end IPv6.
Actionable Recommendations
- ✓Audit IPv6 routing: compare IPv4 and IPv6 paths to your key services and prefer local peering for IPv6 as you would for IPv4.
- ✓Peer IPv6 locally: set up native IPv6 peering at the internet exchanges where you already peer IPv4.
- ✓Check dual-stack binding: make sure server software listens on both IPv4 and IPv6 so clients don’t wait for connection fallbacks.
- ✓Allow ICMPv6 Packet Too Big: permit these messages through firewalls so path MTU discovery works.
- ✓Update firewalls and monitoring: make sure security tools inspect and log IPv6 as thoroughly as IPv4, and use a subnet calculator when planning IPv4 ranges alongside.