Skip to main content

Most people assume the internet just sort of works. Plug in a device, get an address, connect. But the system handing out those addresses has been running on fumes since about 2015, and the consequences are showing up in places you wouldn’t expect.

IPv4 gives us roughly 4.3 billion unique addresses. Sounds like a lot until you count the devices: somewhere around 15 billion right now, and climbing.

The Free Pool Is Gone

ARIN, the registry responsible for North American IP allocations, exhausted its last available IPv4 block in September 2015. RIPE NCC in Europe followed in November 2019. Asia-Pacific’s APNIC has been rationing tiny /22 blocks (just 1,024 addresses each) since 2011.

There’s no more free IPv4. Full stop.

What replaced it is a trading market that looks a lot like real estate speculation. Brokers like Hilco Streambank and IPv4.Global facilitate deals where a single address can fetch $50 or more. A /16 block, which is 65,536 addresses, has sold for upwards of $3.2 million in recent transactions. Five years ago those same blocks went for maybe half that.

Amazon quietly acquired over 100 million IPv4 addresses over the years. Microsoft grabbed 666,624 addresses in a single purchase from Nortel’s bankruptcy estate back in 2011 for $7.5 million. At today’s prices, that lot would be worth closer to $33 million. Smart move on their part.

Proxy Infrastructure Pays the Price

This is where IPv4 scarcity gets tangible for a lot of businesses. Proxy services need IP addresses the way taxis need license plates. No address, no connection.

Datacenter proxies can stretch their supply through virtualization. One beefy server spins up 200 or 300 proxy instances sharing a pool of IPs. But residential and ISP proxy providers don’t have that luxury. Their addresses come from real internet service providers, assigned to real connections. An ipv4 residential proxy pool is only as big as the provider’s ability to source legitimate, ISP-verified addresses from a market that gets tighter every year.

The cost pressure is real. Proxy providers who built their address inventories before 2019 locked in prices that new competitors can’t touch. That cost gap gets passed to customers, which means businesses relying on proxies for price monitoring, ad verification, or market research are all paying more than they were three years ago.

IPv6 Adoption Is Stuck in the Middle

Google publishes real-time IPv6 stats, and the chart tells an interesting story. Adoption crossed 45% globally, but it’s wildly uneven. France and Germany sit above 70%. China hovers under 30%. Large swaths of Africa and Southeast Asia barely register.

The gap isn’t about awareness. Network engineers know IPv6 is the future. The problem is money and inertia. Replacing routers, updating firewall rules, testing every application against a new protocol stack, training staff on a different addressing scheme: none of it is free, and none of it generates revenue. So it keeps getting pushed to next quarter.

And here’s the awkward part: even companies that have migrated internally still need IPv4 for external-facing services. Too many third-party APIs, payment gateways, and legacy partners only speak IPv4. Running both protocols at once isn’t a transition. It’s just more work.

CDNs, Hosting, and the Geography Tax

Content delivery networks have their own version of this headache. Cloudflare, Akamai, and Fastly need local IP addresses in every region they serve, and some regions are far more expensive than others.

Try spinning up infrastructure in Jakarta or Lagos. IPv4 addresses in APNIC and AFRINIC territories carry a premium because supply is thinnest there. ARIN’s own resource guide lays out how the transfer market works in North America, and other regions follow similar patterns. The Internet Society documented these regional disparities years ago, and the gaps have only widened since.

Carrier-Grade NAT is the band-aid most ISPs use: stuffing hundreds of customers behind one public IP. It keeps things running but creates real problems. WebRTC breaks. Geolocation accuracy drops. Rate-limiting by IP becomes useless when 500 different households share one address. For anyone doing fraud detection or abuse prevention, it’s a mess.

No Clean Fix Coming

The honest answer is that this situation doesn’t resolve neatly. IPv6 will keep growing, slowly. IPv4 prices will keep climbing, also slowly. And the companies caught in between will keep paying the tax.

Businesses that stocked up on address space or committed to dual-stack engineering early are fine. For the rest, the options are pretty limited: pay secondary market rates, layer on more NAT complexity, or scale back plans that require large blocks of clean IPv4 addresses. None of those are great. But that’s where things stand.

Leave a Reply