IPv4 Exhaustion: What Happened and What's Next

In February 2011, IANA — the Internet Assigned Numbers Authority — allocated the last blocks of IPv4 addresses to the five regional registries. The global pool was empty. Regional registries continued allocating from their existing reserves for a few more years, with ARIN (the North American registry) running out of its free pool in September 2015. IPv4 exhaustion is real, it happened, and the internet has been in a managed crisis state ever since.

The Numbers Behind the Crisis

IPv4 uses 32-bit addresses, giving a theoretical maximum of 4,294,967,296 unique addresses. In 1981 when the address space was designed, that seemed enormous. By the early 1990s, it was obvious it wouldn't be enough. The internet was growing faster than anyone had predicted, and the address allocation had been wasteful in the early years — Class A blocks of 16 million addresses were handed to organisations that needed a fraction of that.

NAT and CIDR were invented specifically to buy time. NAT lets millions of devices share one public IP. CIDR allowed more efficient allocation of address space, replacing the old classful system that wasted huge blocks. These measures pushed exhaustion back by roughly 10 to 15 years. Without them, the crisis would have hit in the late 1990s.

The Secondary Market

IPv4 addresses are now traded as commodities. Amazon, Microsoft, and other cloud giants have purchased large blocks from organisations that no longer need them. The going rate fluctuated around $45 to $60 per IPv4 address in 2024 — meaning a /8 block (16 million addresses) was worth somewhere around $700 million at market rates. MIT sold 8 million addresses it had from the early internet days for a reported $50 per address in 2017.

And that's the real story behind IPv4 'exhaustion' — the addresses don't go away; they just get concentrated in the hands of entities that can pay for them. Cloud providers and CDNs can outbid everyone else, which means they control an increasing share of the IPv4 space.

Why IPv6 Hasn't Fixed It Yet

IPv6 has been the official solution since 1998. Adoption is real and growing — but the transition requires every piece of network infrastructure to be updated simultaneously, and legacy systems resist change. Corporate networks, industrial equipment, and embedded devices often can't be easily updated. The dual-stack approach (running both protocols) works but doesn't reduce the demand for IPv4 addresses — it just adds IPv6 alongside them.

What the Transition Actually Looks Like Day to Day

For most end users, the IPv4-to-IPv6 transition is invisible. Your device picks whichever protocol works. Websites support both. The Happy Eyeballs algorithm in your OS tries IPv6 first and falls back to IPv4 within milliseconds if needed. You'll never see an error message saying 'IPv4 exhausted, please upgrade.' The internet keeps working, just with more CGNAT and more creative address management happening behind the scenes.

For developers and system administrators, the transition is ongoing operational work. New services should support IPv6 from day one. Existing services need auditing — many applications still have hardcoded assumptions about IPv4 address formats, IPv4-only socket bindings, or inadequate IPv6 logging. The IETF declared IPv4 exhaustion a crisis in the early 2000s, and while they were right, the internet's resilience has meant the crisis has been managed rather than catastrophic. IPv6 adoption will probably reach full coverage sometime in the 2030s.