Why IPv4 Still Dominates the Internet Despite IPv6 Growth

Sometime around 2011, the internet was supposed to run out of IPv4 addresses. The warnings were everywhere. IANA handed out its last blocks, regional registries started rationing, and the tech press ran breathless headlines about the coming address apocalypse.

That was fourteen years ago. IPv4 still carries more than half of global internet traffic today.

The Migration That Never Quite Happened

Google’s transparency reports put IPv6 adoption at roughly 45% of its traffic. Sounds like real progress, right? It is, but that number hides a lot. Google’s audience leans heavily toward mobile users in the US, India, and Western Europe, exactly the markets where carriers have actually bothered to deploy IPv6.

Go check adoption rates in Peru or Nigeria or Indonesia. You’ll find IPv4 running practically everything. Telecoms in those countries poured money into IPv4 infrastructure over the past 20 years, and “rip it all out for IPv6” doesn’t show up on anyone’s capital expenditure plan.

The pattern holds in corporate networks too. An IT team at a mid-sized company faces a genuinely miserable project if they want to migrate: updating firewalls, rewriting application configs, testing load balancers, retraining staff. All for a protocol that, honestly, their current setup doesn’t need yet. Businesses that still require dedicated IPv4 addresses for proxy infrastructure, multi-account management, or web scraping can proxy ipv4 buy through specialty providers without breaking a sweat. The supply hasn’t dried up the way people expected.

NAT Changed the Math Completely

The single biggest reason IPv4 refused to die has a name: Network Address Translation. NAT was invented in the mid-1990s as a stopgap. It lets hundreds or thousands of devices hide behind one public IPv4 address. Your Wi-Fi router at home does exactly this.

It was never meant to last. But ISPs saw an opportunity and ran with it. Carrier-grade NAT (CGNAT) lets companies like Comcast and Deutsche Telekom serve entire neighborhoods with a handful of public IPs. The technique has real downsides (good luck hosting a game server from your apartment), but for browsing, streaming, and email? Works fine.

And “works fine” is enough for most decision-makers to shelve the IPv6 conversation indefinitely.

The Bizarre IPv4 Aftermarket

Here’s where things get genuinely interesting. IPv4 addresses now trade like real estate. A single address goes for $40 to $50 on the secondary market. Full /16 blocks (65,536 addresses) change hands for millions.

Microsoft spent $7.5 million buying Nortel’s address space back in 2011. At today’s prices, that looks like a bargain. ARIN overseas transfer records in North America, and the volume of trades keeps climbing year after year. Universities and government agencies that grabbed huge allocations in the 1980s are sitting on genuinely valuable assets now.

This market creates a perverse incentive. Why rush to IPv6 when your unused IPv4 blocks appreciate 10% annually? Some organizations treat their address space as an investment portfolio rather than networking infrastructure. It’s bizarre, but it’s real.

IPv6 Wins Are Real, Just Concentrated

None of this means IPv6 is failing. T-Mobile routes over 90% of US subscriber traffic through IPv6. Reliance Jio in India launched as an IPv6-native network, which single-handedly made India the global leader in adoption. Cloud providers like AWS and Azure push dual-stack by default on new deployments.

The Internet Society documents these gains carefully, and the trend line points in the right direction. Content delivery networks at Cloudflare and Akamai serve most requests over IPv6 already.

But the wins cluster in mobile and cloud. Corporate LANs, legacy applications, embedded systems, older VPN concentrators: all still overwhelmingly IPv4. Plenty of enterprise software written ten or fifteen years ago hardcoded four-octet addresses in ways that would take months to untangle. Nobody’s volunteering for that project.

Living With Both Protocols

Most networks that claim IPv6 support actually run dual-stack, meaning both protocols operate simultaneously. It’s practical, but it also means maintaining two of everything: firewall policies, DNS records, monitoring rules, troubleshooting playbooks.

Network engineers expected the dual-stack to be a five-year bridge. We’re well past that. At this point, planning for a single-protocol future feels naive. Organizations budgeting for new infrastructure should assume they’ll carry both IPv4 and IPv6 through at least the 2030s, probably longer.

The internet has always been held together with duct tape and good intentions. IPv4 sticking around longer than expected is perfectly on brand.