IPv6 carried half of internet traffic – for one day, according to Google
- Reference: 1776391490
- News link: https://www.theregister.co.uk/2026/04/17/ipv6_50_percent_google/
- Source link:
The search and ads giant tracks the percentage of its users who access its services over IPv6 on a [1]page that The Register has often seen used as a de facto indicator of all IPv6 uptake when we attend networking and internet governance events.
According to the Big G’s stats, on March 28th, 50.1 percent of the traffic the company detected used IPv6, up from 46.33 percent a year earlier. Google’s data records plenty of days over the last year when IPv6 carried over 49.5 percent of traffic, and a slow climb towards greater prevalence of traffic using the protocol.
[2]
Google has a decent view of the internet because its main domain and YouTube are the world’s two most-trafficked websites.
[3]
[4]
However, other sources don’t currently report IPv6 at 50 percent of visible traffic.
Cloudflare’s Radar service, [5]rates IPv6 as the source of 40.1 percent of HTTP requests. APNIC labs [6]found 43.13 percent of networks it can see are IPv6-capable.
[7]
Google’s result is therefore notable, and nice, but not solid proof that IPv6 has finally become dominant.
[8]AI vastly reduced stress of IPv6 migrations in university experiment
[9]Unofficial IETF draft calls for grant of five nonillion IPv6 addresses to ham radio operators
[10]Starlink tells the world it has over 150 sextillion IPv6 addresses
[11]IETF Draft suggests making IPv6 standard on DNS resolvers - partly to destroy IPv4
Internetworking boffins conceived of IPv6 after realizing IPv4’s 4.3 billion available addresses would be insufficient to service the growing number of internet-connected devices. They therefore designed IPv6 around 128-bit addresses, meaning the IPv6 numberspace offers 340 undecillion addresses – 340 followed by 36 zeroes.
That vast quantity of addresses is probably enough to allow humanity to assign a unique identifier to every connected device our species will create between now and the heat-death of the universe – and give an IP address to plenty of the stars that are born and die along the way.
The effectively unlimited availability of IPv6 meant pundits assumed network operators would swiftly adopt the protocol, especially after the pool of available IPv4 addresses dried up in the mid-2010s.
Many therefore regard slow uptake of IPv6 as curious.
[12]
As The Register has [13]reported , two main factors retarded IPv6 adoption.
One was that the protocol didn’t add many useful features, so network operators didn’t rush to adopt it.
The other was the advent of network address translation (NAT), which allows many devices to share a single public IPv4 address. Many organizations with IPv4 holdings used NAT to increase their capacity rather than build a new network on IPv6, slowing uptake of the newer protocol.
Some nations, however, passed 50 percent IPv6 adoption years ago. That’s an artefact of the loose governance that prevailed in the early years of the internet’s growth and allowed developed nations to scoop big pools of IPv4 before other countries could stake a claim. Nations like India and China therefore gained insufficient IPv4 resources to serve their populations and drove for IPv6 adoption instead. The 29 nations in Asia and Oceania served by the Asia Pacific Network Information Centre therefore passed 50 percent IPv6 in [14]2025 . The American Registry for Internet Numbers, which serves 29 nations across North America and the Caribbean, reached 50 percent adoption a couple of years earlier. ®
Get our [15]Tech Resources
[1] https://www.google.com/intl/en/ipv6/statistics.html
[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/networks&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aeGwaLS8WKKsZTkmZ49a0wAAAQ4&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/networks&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aeGwaLS8WKKsZTkmZ49a0wAAAQ4&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/networks&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aeGwaLS8WKKsZTkmZ49a0wAAAQ4&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[5] https://radar.cloudflare.com/adoption-and-usage
[6] https://stats.labs.apnic.net/ipv6/
[7] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/networks&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aeGwaLS8WKKsZTkmZ49a0wAAAQ4&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[8] https://www.theregister.com/2026/02/10/ipv6_generative_ai_experiment/
[9] https://www.theregister.com/2025/12/02/ham_radio_ipv6_ietf_draft/
[10] https://www.theregister.com/2025/10/29/nded/
[11] https://www.theregister.com/2025/08/20/ietf_dnsop_3901bis_ipv4_ipv6/
[12] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/networks&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aeGwaLS8WKKsZTkmZ49a0wAAAQ4&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[13] https://www.theregister.com/2025/12/31/ipv6_at_30
[14] https://www.theregister.com/2025/04/23/apnic_half_ipv6_capable/
[15] https://whitepapers.theregister.com/
Re: International
Also mobile. A lot of the growth in IPv6 usage is just a greater share of mobile devices today. We should expect a plateau effect as the rate of smartphone adoption reaches saturation.
Many of the problems which IPv6 causes for a premises deployment aren't a thing for phones connected to a carrier's network.
We're going to run out of IPv4 addresses soon, the IEFT says
And by soon, I mean... 2005.
Yes, after RFC 1287 was published in 1992, the alarm bells sounded that we could run out of IPv4 by 2005. That was later updated to say that we would run out by... 2011.
It's nice that IPv6 works, and it's good to see the backbone migrating to it. But personally, I suspect the epochalypse (January 19, 2038, mark your calendars) will happen before IPv4 is phased out.
IPv6 introduced new problems, so slow uptake shouldn't be curious or confusing
The designers of IPv6 make a classic mistake in rolling out a new version (something Microsoft does all the time): they arrogantly thought that users would be forced to accept whatever was put in front of them if enough pressure was applied. They thought the word "crisis" was scary enough. It didn't work out that way. I run an IPv4 only network. Dual stacking doesn't solve a single problem for us, but it does cost time and introduce additional points of failure.
Listen to the users. Non-globally-routable local addresses are a feature in many use cases. DHCP4 is a feature. We like reservations, segmentation, DHCP control of DNS and the gateway, and not having to re-address things if a prefix changes. We like local devices being agnostic of addressing on the WAN side of CPE. Our printers don't need to be globally routable and only need to know their local address. NAT is no substitute for a firewall, but we like hiding local network topography behind one public address. Principle of least information. We also like managing a single stack firewall, rather than dual. Fewer points of potential failure is a feature. We can forward a port if needed. If there's a firewall screw-up, then non-routable local stuff benefits from defense in depth.
SLAAC sucks. "Privacy" extensions were a band-aid which should have never been necessary in the first place. Local devices fast fluxing their way through tons of temporary addresses makes logging and auditing a mess. IPv6 logs are harder to grep because addresses can be either expanded or compact. Some software mixes both, making for messed up logs. IPv6 networks can't be scanned as part of a device audit. Google's Android insists on using SLAAC and won't respect DHCP6. I'm not even going to complain about address length or the inability to type IPv6 addresses with one hand on a numpad. Those are annoyances. IPv6 failed in much bigger things than that.
I'm sure the IPv6 evangelists will be by shortly to flame me for my blasphemy. I don't care. None of the things they like about IPv6 solve a single problem for us. Unless our use case changes, the only reason we'd adopt IPv6 is if the exhaustion crisis actually happened, which it didn't. I'll probably be retired before a transition is actually necessary. That'll be one or two upgrade cycles from now. Budgets and time are always finite, and we can spend them on things which deliver useful features in the meantime. Literally the only thing we would get out of running IPv6 is being able to post comments saying, "we run IPv6!"
International
is the key word. Countries that have plenty of IPv4 still are on IPv4 in the main.