Are we on ipv6 yet? (simple answer: no)

In December 1995, RFC 1883 defined IPv6 to solve address exhaustion, we are in 2026 and still a major part of the Internet is still running on IPv4. This is how doesgithub.onipv6.work was born.

A long-waited upgrade #

I finally had the option to upgrade my Internet connection and I choose a provider that, finally supported IPv6 (Aruba if you are in Italy).

I started experimenting with services, and while I was trying to finally cut out AWS costs for NAT Gateways or Public IPv4 addresses associated to instances, I found out that GitHub, still, does not offer its services on IPv6.

And it wasn’t the only big tech giant.

Here are my thougts about that, the “doesgithub.onipv6.work” tool is only for irony (still, it makes real checks behind the scenes).

Innovation at the Speed of 1995 #

We live in the era of Hyper-Growth, Agile Transformation, and AI-Driven Leadership. We attend conferences where entrepreneurs talk about “breaking things and moving fast.” We have weekly “Retrospectives” to improve our velocity by 2%.

Yet, here we are in 2026, still using a networking protocol designed long before Coolio’s Gangsta’ Paradise was at the top of the charts.

We’ve mastered the art of Continuous Delivery for our CSS buttons, but we’ve been “Iterating” on IPv6 adoption for 30 years.

If this were a Jira ticket, it would be the most expensive technical debt in human history.

We talk about “Sustainability” and “Resource Optimization,” while we burn millions of dollars (and megawatts) on:

• NAT Gateways: Expensive heaters designed to translate 1980s addresses into the modern world, now adding $32.40/month in multiples to add HA
• Address Auctions: Buying IPv4 blocks at the price of fine art.
• IPv4 Address Scarcity Surcharge +$3.60 / IP / monht
• Legacy Complexity: Wasting engineer hours debugging “split-horizon DNS”, “CIDR overlaps” and double NATs that shouldn’t exist

Project Retrospective: The 30-Year Sprint #

Welcome to the ultimate tech debt. While we spent the last three decades innovating frameworks, technologies and processes, in the meantime the literal plumbing of the internet has been clogged since the mid-90s.

We build hyper-scale Kubernetes clusters, then suffocate them behind legacy translation gateways and we’ve been running the same migration sprint for 11,000+ days.

In 2026, we have AI that can simulate consciousness, write poetry, generate realistic images and videos and, yet, we are still paying “The IPv4 Tax”:

GitHub is not the only one, but it is the most used. My brain hurts when I still read on HuggingFace help forums “disable IPv6” as a troubleshooting step.

It is time to evolve and stop living in the past. IPv6 is not anymore a “nice to have” feature.

If you’re reading this and you’re still thinking that migrating is not worth the hassle, I hope that you’ll never have to deal with organizations that have overlapping CIDRs or, simply, hosting your own service at home without having to deal with carrier-grade double NATs or VPNs (even if TailScale does an excellent job to overcome these issues)

Drop a comment below and let me know your thoughts!