IPv6 Adoption in 2026
17 points by binjip978
17 points by binjip978
I think there's an economic angle here that isn't talked about much. If clients are all on ip4 behind NAT, they can't talk to each other. The means of production for web content are no longer by default in every household. There's lots of control and money to be made from this situation.
There are so many assumptions folk have about networking with their exposure to IPv4 that trash their ability to internalize the IPv6 model. It doesn't help that IPv6 is a technology that is far more relevant in its appeal to network operators, where folk at home are likely immediately flustered by the idea that they have to properly firewall their network instead of pretending that's what NAT does.
It's also unfair how IPv4 fans will overlook how IPv6 has been key to expanding end-to-end routing over the Internet to regions which are seemingly "not important enough" to get proper allocations from the internet number registries. It's nice that most of us here (I assume) are running ISPs drowning in IPv4 addresses, and don't even have to imagine not having proper connectivity with the Internet in the first place. From what I've seen there are also solutions being built around IPv6 that are allowing large metro and backbone scale networks to push their network utilization higher than they've been able to push IPv4 + QoS.
In my experience, actually running a dual-stack IPv6 network is why IPv6 frustrates people. I personally run my network directly on the global unicast IPv6 prefixes I'm assigned by the ISP at each of my sites. This is not a security issue, it doesn't reduce confidentiality or integrity, and makes addressing and address planning easy. When I want to assign an IPv4 address to a given service, the host establishes a point-to-point tunnel to the egress router with said IPv4 address. Thus, I entirely avoid running a dual-stack network.
IPv6 is really not that hard, yet we see many people -- usually the same kind of people who say self-hosting email is too difficult -- saying it's too much work for no gain. They often blame IPv6 for their ISP's or their own issues, too.
I've literally had people who have the same sentiment say that IPv6 is irrelevant because their logs don't show any attempts to use IPv6 (and they didn't realize that if you don't offer it and don't use it, it can't possibly be in the logs).
What surprises me is how many large providers aren't IPv6 enabled by default. Pretty much every cell phone uses IPv6, and it's the only way to get proper connections when behind CG-NAT is IPv6, so dual stacking should already be in most places.
I see those afraid of IPv6 like I see Windows users - the majority can be stuck with a bad thing, but that doesn't mean the minority isn't benefitting by doing something different.
Honestly I cannot be bothered to self-host email, seems incredibly annoying. But IPv6 is easy. So easy, much easier than IPv4.
I do both self-host mail and run IPv4/IPv6 dual-stack services but honestly, neither is easy. My setup would be roughly half as complex if I went with pure IPv4. I do eat the cost because some part of me still believes in IPv6 but it's definitely not a no-brainer.
The lack of IPv6 adoption is frustrating
The world in many ways moved past this issue. There are some people who want to see a full switch to IPv6, but at this point it's unlikely to happen, and there does not seem to be a big push towards that potential future.
It can be frustrating if you have vested interests in it, but for most people IPv4 v. IPv6 is just not really a particularly relevant issue in 2026.
The world in many ways moved past this issue.
Not at all. CGNAT can't scale forever, you run out of ports eventually (and it's illegal in many places to have more than like 16 customers behind the same CGNAT IP). Staying on IPv4 forever is straight up impossible as more and more people get connected to the net. IPv6 adoption was at 0% in 2010, and has been growing linearly since 2012. That's not super fast, sure, but it's happening.
Maybe! We’ll see in a few years where it’s going. At the moment the split is remarkably stable.
There are some people who want to see a full switch to IPv6
I've never come across this at all, aside from people trying to create a strawman. People who've claimed they've heard this can only point to other instances where people were trying to create a strawman.
Nobody seriously ever thought or thinks that the world should or even can switch entirely from IPv4 to IPv6.
I've never come across this at all, aside from people trying to create a strawman.
Just to quote the internet society:
IPv6, the latest version of the Internet Protocol (IP) standard, is currently supplementing and will eventually replace IPv4
I think right now the general mood around IPv6 is quite a less enthusiastic than it was before. There is a growing consensus that we will end up with a ~equal deployment of IPv4 and IPv6 for a very long time.
I think it's safe to say they're talking about eventually, in the abstract, as in many decades from now. People who act like that's something that's being discussed now really should know better.
I think right now the general mood around IPv6 is quite a less enthusiastic than it was before.
Than when? The general mood has never been one of enthusiasm about IPv6.
What I see that's different is in the past people said things like, "perhaps I should learn IPv6", "perhaps I should implement dual stack" because they hadn't, and these days it's more of, "of course it should be dual stack".
My experience with IPv6 is that I want to like, let stuff from my LAN through the firewall, so I search for "IPv6 local addresses", and then you have to scroll past a bunch of StackOverflow "answers" that are multiple paragraphs were IPv6 maximalists herp a derp about how akkshually LAN IPs with their NAT are an IPv4ism and in the fantasy future of IPv6 every device will have a publicly routed unique IPv6 delegated to them, until I get to the post "oh yeah you want fc00::/7", which answers the question for any real IPv6 deployment one has in their home.
Should've just increased the address space and gotten rid of fragmentation but no, we need to think about a planet covered in individually-addressible nanobots that dynamically decide on a routing table through local neighbour discovery.
It's not "maximalism" to point out the entire fucking point of the protocol upgrade.
It's not a fantasy future either.
Literally if you just get a rented modem-router box from an ISP that has IPv6 and don't even think about it, your box would get a /64 from the ISP and when you connect your phone to the WiFi it'll get a publicly routed address.
This is how the protocol is intended to be deployed and it's working like that already for millions of customers who don't even know what IP is.
If we got started setting up the Internet today, this would maybe not be a problem, but now we have a decade or more of IoT junk with paper towel grade security hardening that flat out assumes they're behind a NAT.
Everyone and their grandma has juiceros, chromecasts, roombas and whatever the heck else on their home network. It's optimistic to think half these things have seen a software update this decade. More likely their manufacturers have already either moved on or gone broke.
I'm security conscious enough to not put these category of devices on the same vlan as the rest of my stuff, it would be insanely irresponsible to make them publicly accessible.
The category of people who do not know what an IP is wouldn't understand a word in this post. Unless this state of affair somehow changes, we're stuck with NAT as the only sane default.
None of those devices assume that they're behind a NAT, they assume that they're behind a firewall. Which they are. None of them are publicly accessible.
Good for you, but my ISP router hands out IPv6 addresses in the fc00::/7 prefix and has no globally routable IPv6. And in either case, I'd still need to know about fc00::/7 because that's probably the addresses of what my boxes would speak to each other from. Or maybe not, because I don't actually know what mDNS would resolve to there.
So just say the equivalent to your 10.0.0.0/8 is fc00::/7 instead of overcomplicating it, because I can talk to a publicly routed address on IPv4 as well.
I understand your frustration –your ISP is giving you the pains of IPv4 on IPv6. That's meaningless. But rest assured this is not how IPv6 was intended, and many ISPs do do the correct thing. It seems your gripe is misdirected when aimed at IPv6.
Even if they gave me a globally routed IPv6, I still wouldn't know what address to whitelist, all of which is an IPv6 problem. Unless you think people's devices should all have the privacy nightmare MAC-based IPv6 addresses with a customer static prefix, in which case: yikes, another stupid thing IPv6 tried and thankfully failed.
Holy cow, I have heard of ISPs doing some very broken things with IPv6 but that absolutely takes the cake. That's like, "why even bother" level of stupidity.
My experience with IPv6 is that I want to like, let stuff from my LAN through the firewall, so I search for "IPv6 local addresses", and then you have to scroll past a bunch of StackOverflow "answers" that are multiple paragraphs were IPv6 maximalists herp a derp about how akkshually LAN IPs with their NAT are an IPv4ism and in the fantasy future of IPv6 every device will have a publicly routed unique IPv6 delegated to them, until I get to the post "oh yeah you want fc00::/7", which answers the question for any real IPv6 deployment one has in their home.
Funny how different experiences can be. When I first got IPv6 from my ISP, I did the same confused search you did. My first result was the Wikipedia page on ULAs, which described what they are perfectly well. Then I started searching for how I'm supposed to route between those and the Internet, at which point the SO posts you deride very clearly and succinctly told me that I probably don't want something that's close to reintroducing NAT, because my ISP has probably already handed me a whole globally routable /56 to play with. And what do you know – they had! So that shook my brain for a bit, and then it settled, and now it's awesome. I think one shakeup like that every 30 years is survivable by the small segment of home users that care about these things (for those who don't, things Just Work much more than with IPv4).
Yeah, unfortunately it's one of these topics that attract the useless zealots and they're not easily distinguished from the the normal (and probably helpful) people.
I've also mostly given up. All my stuff on servers has been dualstack for at least 10 years (unless company says no) but I had it completely disabled in my home network until last year. Now I've not really noticed any problems either.
let stuff from my LAN through the firewall
Can you explain what you mean by this?
IPv6 maximalists herp a derp about how akkshually LAN IPs with their NAT are an IPv4ism and in the fantasy future of IPv6 every device will have a publicly routed unique IPv6 delegated to them, until I get to the post "oh yeah you want fc00::/7", which answers the question for any real IPv6 deployment one has in their home.
Typically you'd just have both types of address. Local addresses for local stuff, global addresses for global stuff. So there's no NAT. IPv6 was designed with every interface having multiple addresses from the start, so it works well. If the destination address is in 2000::/3, it'll prefer to select a source address that's also in 2000::/3, and if it's in fc00::/7, it'll select a source address in that range.
dynamically decide on a routing table through local neighbour discovery.
That's basically the same as in IPv4 though, just you get a NDP Router Advertisement instead of a DHCP doodad to tell you about the local network.
Can you explain what you mean by this?
Letting two devices which are in the same house communicate through each others respective firewalls.