IPv4 addresses trade at $35-52 each in 2026. Actual commodity instruments with brokers, escrow services, and standardized contracts. Organizations lease them at $0.38-0.45 per month. Over 40 million transferred last year alone.

IPv6 has been "the future" for 25 years. Yet here we are, trading addresses from a protocol designed when the internet was a research project connecting universities.

IANA's global pool ran dry on January 31, 2011. Regional registries followed over the next four years. But the scarcity itself traces back to decisions made when scarcity seemed impossible.

32-bit addressing gave the internet 4.3 billion unique addresses—limitless in the 1980s. The allocation methodology was "request and receive." MIT, IBM, and Ford each got 16.7 million addresses simply by asking. Free. No justification required.

Those Class A blocks now represent $600 million to $900 million in market value. Each one.

Classful allocation divided the address space into rigid blocks. Fifty percent of all addresses went to just 128 Class A networks. Need 2,000 servers? You got a Class B block of 65,536 addresses and wasted 63,000. Structural waste, built into how addresses could be allocated.

By the early 1990s, exhaustion looked inevitable. CIDR arrived in 1993, allowing flexible allocation unconstrained by class boundaries. Organizations needing 2,000 addresses could get 2,048 instead of 65,536. IPv4 got a reprieve.

Then RFC 1918 in 1996 defined private address ranges any organization could use internally—10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16. NAT let multiple devices share a single public IP. Suddenly you didn't need public addresses for every device. Just the edge.

Temporary bridges to IPv6, or so the thinking went.

NAT restructured how networks operated. Organizations built entire infrastructures around private addressing. ISPs designed their networks assuming NAT at every customer premise. When carrier-grade NAT arrived, ISPs could share small pools of public addresses among thousands of users. The workaround became the architecture.

Four months after IANA exhaustion, Microsoft bought 666,624 addresses from bankrupt Nortel for $7.5 million. A U.S. bankruptcy court recognized them as property of the estate. The commodity market was born at $11.25 per address. Prices climbed to $60 by 2021 before settling into today's $35-52 range.

IPv4 still functions, so the market exists. And IPv4 functions because those 1990s workarounds removed the urgency to move.

IPv6 adoption sits at 43-45% globally. France and Germany are above 75%. The U.S. just crossed 50%. But enterprise adoption lags far behind. Organizations with legacy IPv4 systems see little immediate benefit in changing what still works. Transition requires hardware upgrades, software changes, dual-stack complexity—running both protocols simultaneously, which means double the configuration, double the security surface, double the operational overhead.

As long as IPv4 functions through NAT and private addressing, urgency stays abstract. And the longer it functions, the more infrastructure gets built assuming it will continue functioning. Dependencies compound.

Addresses designed for an experimental network, extended through temporary fixes that became foundation, now support a commodity market where they trade like real estate. The 32-bit decision created a path dependency powerful enough to trap an entire internet. The workarounds succeeded, and that success is exactly what made them permanent.