It is the year 2026. The IPv6 protocol has existed since 1998. It has been an RFC standard for over a quarter century. Every modern operating system supports it. Every modern network component can handle it. Every textbook, every course, every certification covers it. And still millions of people browse exclusively over IPv4 — not because they want to, but because others neglected it.

This is no longer merely a technical oversight. It is an embarrassment.

The conspicuous shame

Let’s start with the obvious. GitHub — the world’s largest software development platform, owned by Microsoft, one of the richest technology companies on earth — does not support IPv6. Not a single byte. Whoever has an IPv6-only connection cannot reach GitHub. No repositories, no issues, no pull requests, no documentation. Microsoft buys GitHub for 7.5 billion dollars, but rolling out IPv6 — apparently that exceeds their means.

Reddit, which bills itself as the Front Page of the Internet, equally lacks IPv6. A platform styling itself as the internet’s homepage doesn’t master the internet’s foundational protocol.

The list goes on. Docker Hub. Cloudflare Workers (partially). Numerous SaaS providers. The list of companies failing outright on this basic topic is long. And it is embarrassing.

A quarter century of ignorance

IPv6 isn’t news. It isn’t experimental technology. It isn’t even controversial. IPv4 addresses have been exhausted since 2011 — IANA allocated the last free blocks. Since then the internet keeps growing only because NAT, carrier-grade NAT and other crutches torture the IPv4 network beyond its construction.

IPv6 solves this problem. Has solved it since 1998. The address space is so vast that every person, every device and every toaster could have their own address — and the supply wouldn’t deplete. It simplifies network architecture, eliminates NAT hacks, improves end-to-end connectivity, reduces latency in mobile networks, enables genuine peer-to-peer communication.

And still, in 2026, we debate adoption. As if it were an optional extension. As if it were a nice-to-have.

It is a must. Whoever doesn’t have it has understood nothing.

Dual Stack: the permanent condition nobody wants

Because some won’t play along, everyone else has to run Dual Stack. Dual Stack — operating IPv4 and IPv6 in parallel — was meant as a transition. A bridge you cross and then demolish. Instead they built houses on the bridge, settled in, and now call it home.

Dual Stack means doubled effort. In every dimension.

Firewall rules. Every rule that exists for IPv4 needs a counterpart for IPv6. And IPv6 rules aren’t copies — ICMPv6 is critical for functionality, neighbour discovery needs special treatment, extension headers require their own filtering logic. Whoever configures IPv6 firewalls like IPv4 firewalls ends up with either holes or broken connections.

Service bindings. Every service must listen on both protocols. Web servers, mail servers, databases, message brokers, monitoring agents. Some software does it automatically. Some doesn’t. Some claims to and still doesn’t.

VPN connections. IPSec, WireGuard, OpenVPN — all support IPv6. But tunnel configurations transporting only IPv4 are endemic. Whoever sets up a VPN tunnel routing only IPv4 has halved their network without noticing.

DNS resolution. AAAA records have existed since 1999. Nevertheless there are hosting providers that don’t offer them. Registrar APIs that don’t support them. CDN configurations that ignore them. DNS is the internet’s phone book — and a phone book listing only half the numbers is broken.

Debugging. A connection timeout under Dual Stack can have ten different causes. Is it an IPv4 problem? An IPv6 problem? A Happy Eyeballs race condition where the client tries IPv6 first, fails, then attempts IPv4 — but so slowly that the timeout triggers? Whoever has diagnosed an intermittent IPv6 failure in a hybrid network knows how much lifetime is wasted in the process.

Compliance and certification. Audits must examine both protocols. Penetration tests must attack both protocols. Documentation must cover both protocols. Certifications testing only IPv4 are worthless — and unfortunately still the norm.

Were fewer people walking around with blinkers on, we could have finished the migration long ago and moved to IPv6-only operation. Instead, administrators and network teams will probably have to lug this Dual Stack nonsense around for another five to ten years. Five to ten years of doubled effort, doubled failure sources, doubled complexity. Because some deemed it unnecessary.

The T-Online example

The university hospital of RWTH Aachen offers free Wi-Fi. Provided via T-Online. Internet access: IPv4 only. No IPv6.

That is more than embarrassing. This is a German flagship company — Deutsche Telekom, styling itself as technological pioneer, investing billions in infrastructure, boasting in advertising campaigns about gigabit networks and 5G — and it cannot manage IPv6 in its Wi-Fi solution. At a university. At a hospital. At an institution conducting research, teaching, healing. IPv4 only. In 2026.

What can you say? The same as always: it isn’t the technology that’s lacking. It’s the will. Telekom can do IPv6. It rolls it out in fixed-line connections. It has the infrastructure. But at this point, in this solution, somebody decides it isn’t necessary. That what’s there is sufficient. That nobody notices half the internet is missing.

But you notice. Everyone notices who has a modern connection and suddenly finds that pages don’t load, apps don’t connect, services are unreachable — because the client wants IPv6, the network offers none, and the fallback to IPv4 is so slow that timeouts trigger.

In a recently published article — All or Nothing — we discussed companies that work half-heartedly. That pretty-label rubbish heaps. That claim to deliver while merely collecting payment. T-Online, in this case, supplies the illustration free of charge: if you can’t do IPv6, better do nothing at all — but stop calling yourself a provider.

The consequence

It is time we stopped sugar-coating this topic.

A provider that doesn’t offer IPv6 or implements it inadequately should lose the right to call itself a provider, period. Whoever doesn’t master the internet’s foundational protocol isn’t a provider. They’re an IPv4 relic heading for irrelevance on a foreseeable schedule.

Whoever brings a network-capable device to market that isn’t fully IPv6-compatible should not receive EU market approval. Period. No exceptions, no transitional regulations, no grandfather clauses. Whoever sells devices without IPv6 in 2026 sells defective goods.

Whoever doubts IPv6 has understood nothing. Not the technology, not the necessity, not the responsibility. IPv6 isn’t a holy war. It’s a standard. Whoever doubts standards because they’re inconvenient doesn’t belong in an industry built on standards.

Whoever fails belongs in the pillory. Not behind closed doors, not in internal post-mortems, but publicly. Because public embarrassment is the only language some companies understand.

What libcom.de means

libcom.de plans, builds and operates networks — for over two decades. IPv6 isn’t a project you introduce. It’s a prerequisite you assume.

Whoever comes to me gets an honest answer: Do you have IPv6? If not, why? If yes, fully or partially? Are your services dual-stack or IPv4-only? Are your DNS records complete? Are your firewall rules symmetric? Are your VPN tunnels IPv6-aware?

The answer to these questions isn’t academic. It decides whether your infrastructure is future-proof or whether it falls victim in five years to a migration that should have happened long ago.

If you wonder whether your IT is IPv6-compatible — genuinely compatible, not theoretically —, or if you seek someone who means it honestly: write to contact@libcom.de. We take an honest inventory. Without sales pressure. With a view to what lasts long-term.

IPv6 is not optional. Whoever still treats it as optional is the problem.