The internet has been “running out of addresses” for decades. The fix has existed since 1998. So why are we still not done?
The Problem That Was Supposed to Be Solved by Now
IPv4, the addressing protocol that underpins the modern internet, was designed in 1981 with a pool of roughly 4.3 billion addresses. At the time, that seemed infinite. By the early 1990s, it was clear it wasn’t.
IPv6 was standardized by the IETF in 1998 (RFC 2460), offering a staggering 340 undecillion addresses — enough for every atom on Earth’s surface to have its own IP. The plan was straightforward: migrate, deprecate IPv4, move on.
It is now 2026. IPv4 is still the dominant protocol. The transition is, generously, about halfway done.
This is a story about technical debt, economic incentives, human inertia, and one of the most complex infrastructure migrations in computing history.
What Changed Between IPv4 and IPv6
Before exploring why the transition stalled, it helps to understand what actually changed.
| Feature | IPv4 | IPv6 |
|---|---|---|
| Address size | 32-bit (~4.3B addresses) | 128-bit (~3.4×10³⁸ addresses) |
| Header size | Variable (20–60 bytes) | Fixed (40 bytes) |
| NAT required | Yes, extensively | No — native end-to-end |
| Auto-configuration | Limited (DHCP) | Built-in (SLAAC) |
| IPSec | Optional | Mandatory (originally) |
| Broadcast | Yes | Replaced by multicast |
| Checksum in header | Yes | Removed |
IPv6 isn’t just “more addresses.” It was designed to simplify routing, eliminate NAT, improve security at the network layer, and reduce the complexity that decades of IPv4 workarounds had introduced.
The Numbers: Where We Actually Stand
Global IPv6 adoption has grown significantly but unevenly:
- Google’s IPv6 statistics consistently show around 40–45% of users accessing Google over IPv6 as of 2025–2026.
- Countries like India, the United States, Germany, and Malaysia lead adoption, often exceeding 60–70%.
- Large parts of Africa, Southeast Asia, and Latin America remain below 10%.
- Among ISPs, adoption ranges from near-100% at forward-thinking carriers to effectively zero at legacy operators.
The numbers look better than they did in 2010, when global adoption was under 1%. But the transition is clearly not a story of steady, inevitable progress.
Why the Transition Stalled
1. NAT Bought Time — and Killed Urgency
Network Address Translation (NAT) was introduced as a temporary stopgap in the 1990s to extend the lifespan of IPv4. It allows thousands of devices to share a single public IP address by mapping private internal addresses.
It worked. Too well.
NAT made the IP address exhaustion problem invisible to end users, enterprise networks, and — critically — to the CFOs who control infrastructure budgets. The “crisis” that was supposed to force migration never arrived as a visible emergency. It arrived as a slow, bureaucratic allocation problem managed by regional registries.
NAT is also deeply embedded in firewalls, load balancers, VPNs, and security policies. Removing it isn’t just a network change — it requires rethinking security architectures that have been built around it for 30 years.
2. The Chicken-and-Egg Problem
IPv6 deployment suffers from a classic two-sided network effect failure:
- ISPs won’t invest in IPv6 infrastructure until customers demand it.
- Customers don’t demand it because their connections work fine.
- Content providers didn’t rush to offer IPv6 until significant traffic could use it.
- Hardware vendors didn’t prioritize IPv6 until operators required it.
This deadlock persisted for years. It only broke when major players — Google, Facebook, Comcast, AT&T — coordinated a public push around World IPv6 Launch Day in June 2012, permanently enabling IPv6 on their services.
3. IPv4 Has a Price, Not a Cliff
When the IANA exhausted its free IPv4 pool in February 2011, many predicted a rapid transition. Instead, a secondary market emerged.
IPv4 addresses are now bought and sold. Prices peaked around $50–60 per address in 2022–2023. Large organizations with legacy allocations — universities, early internet companies, government agencies — sit on enormous blocks of address space they can monetize.
This market removes one of the strongest economic incentives to migrate. Why invest in an IPv6 rollout when you can simply buy addresses?
4. Dual-Stack Complexity Is Real
The transition strategy endorsed by most standards bodies is dual-stack: running IPv4 and IPv6 simultaneously during migration. Every device, every router, every application must support both protocols. Troubleshooting becomes significantly more complex. Network teams must understand two protocol stacks. Monitoring tools must track both.
For resource-constrained organizations — small ISPs, municipal networks, enterprises without dedicated network engineers — the operational overhead is a genuine barrier.
5. Legacy Systems That Cannot Be Touched
Banks running 1980s mainframes. Industrial control systems embedded in factory floors. Medical devices certified to specific software configurations. Embedded systems in power grids.
Many of these systems speak IPv4 only and cannot be updated without regulatory re-certification, hardware replacement, or prohibitive cost. They will remain on IPv4 for years or decades, making a full sunset of IPv4 a distant prospect.
The Hidden Cost of Staying on IPv4
IPv6 skeptics often ask: if NAT works, why bother? The answer lies in what NAT breaks.
End-to-End Connectivity
The original internet design assumed every device has a globally routable address. NAT shatters this assumption. Applications that need direct device-to-device communication — VoIP, video conferencing, gaming, IoT sensors — must use increasingly baroque workarounds: STUN, TURN, ICE, UPnP hole-punching.
WebRTC, one of the more sophisticated real-time communication frameworks, spends significant complexity budget on NAT traversal. This complexity exists entirely because of IPv4 exhaustion.
Carrier-Grade NAT and the Death of IP Identity
As IPv4 addresses grew scarce, ISPs began deploying Carrier-Grade NAT (CGN), placing entire customer neighborhoods behind a single IP. This introduces a new layer of NAT on top of home routers — double NAT.
The consequences:
- Port forwarding becomes impossible or unreliable.
- Peer-to-peer applications break.
- IP-based abuse tracking becomes ambiguous — one “IP address” may represent thousands of users.
- Connection quality for gaming, video streaming, and VoIP degrades.
CGN customers are, in a very real sense, receiving a degraded internet experience compared to those with native IPv6 or even native IPv4.
Routing Table Bloat
The global BGP routing table has grown from under 100,000 routes in 2000 to over 1 million routes today, driven largely by IPv4 address fragmentation as organizations carve up legacy allocations and advertise increasingly small prefixes. This strains router memory and increases convergence times across the entire internet.
The Forces Pushing IPv6 Forward
Despite the headwinds, the transition is progressing. Several forces are accelerating it:
Mobile Networks
Mobile carriers are natural IPv6 adopters. When 4G LTE was designed, engineers knew they’d need to connect hundreds of millions of devices. IPv6 was built into LTE specifications from the start.
In many countries, mobile networks are the primary driver of IPv6 adoption. India’s dramatic IPv6 growth is largely attributable to Reliance Jio deploying an IPv6-first mobile network at massive scale.
Hyperscale Cloud Providers
AWS, Azure, and Google Cloud have all moved aggressively toward IPv6. AWS began charging for public IPv4 addresses in 2024 — a direct economic signal. When cloud providers make IPv4 cost money and IPv6 free, the calculus for new deployments shifts immediately.
IoT Explosion
The Internet of Things — billions of sensors, cameras, smart meters, industrial devices — cannot practically be deployed on a fragmented IPv4 infrastructure. IPv6’s address space is essentially mandatory for large-scale IoT deployments. Every smart city project, every industrial automation initiative, every connected vehicle platform drives IPv6 adoption.
Regulatory Pressure
Governments and standards bodies are beginning to mandate IPv6 support. The US federal government has required IPv6 compliance from agencies for years. The European Union has published IPv6 action plans. China’s government has aggressively promoted IPv6 deployment as part of national infrastructure policy, resulting in hundreds of millions of IPv6 users.
What the Transition Actually Looks Like in Practice
The transition has never been a single event. It’s a patchwork of migration approaches:
- Dual-stack remains the most common strategy for organizations mid-transition.
- IPv6-only with NAT64/DNS64 allows IPv6-only clients to reach IPv4 content through translation gateways — increasingly used in mobile networks.
- 464XLAT is a hybrid approach widely deployed by mobile carriers, translating between IPv4 clients and IPv6 infrastructure.
- Tunnel brokers allow IPv6 traffic to traverse IPv4-only networks, used as a bridge during partial rollouts.
None of these are elegant. All of them are operational realities.
What Lies Ahead
The Next Decade
IPv4 will not disappear in the next ten years. The secondary market will keep it viable. Legacy systems will keep it necessary. But the balance will continue to shift.
Predictions with reasonable confidence:
- IPv6-only new deployments will become the norm for mobile networks, cloud infrastructure, and IoT platforms.
- IPv4 address prices will likely plateau or decline as IPv6 reduces demand, though large legacy holders may sit on addresses indefinitely.
- CGN will expand in developing markets where IPv4 addresses are unaffordable, accelerating the quality divide between IPv4 and IPv6 users.
- Enterprise holdouts will face increasing pressure as software vendors drop IPv4 support in new features.
The Long Tail
The hardest part of the IPv6 transition isn’t the big players — it’s the long tail. Thousands of small ISPs, regional carriers, mid-sized enterprises, legacy application vendors, embedded systems manufacturers. These organizations don’t have dedicated network teams. They have one IT generalist and a support contract.
For this segment, the transition will be slow, messy, and driven more by vendor forcing functions (end-of-support for IPv4-only equipment, cloud pricing) than by proactive planning.
Will IPv4 Ever Truly Sunset?
The honest answer: not within any planning horizon most organizations operate on. The IETF has no sunset date. No major government has mandated an IPv4 deprecation deadline. The economic infrastructure around IPv4 address trading is too established.
What’s more likely is a gradual irrelevance: IPv4 becomes a legacy compatibility layer, maintained but not developed, increasingly invisible to end users who operate entirely on IPv6 infrastructure even if they don’t know it. Like analog phone lines persisting alongside digital infrastructure for decades, IPv4 may simply fade rather than terminate.
Lessons for Technology Transitions
The IPv6 transition offers uncomfortable lessons for anyone working on large-scale infrastructure migration:
- Temporary fixes become permanent. NAT was never meant to last 30 years.
- Economic incentives matter more than technical superiority. IPv6 is objectively better in many ways. It didn’t matter until the economics aligned.
- Coordination problems require coordination solutions. World IPv6 Launch Day worked because major players acted together. Unilateral migration is much harder.
- Upgrade paths must be smooth. The dual-stack period has been long and painful. Transitions that require a hard cutover rarely happen cleanly.
- Legacy systems are real constraints, not excuses. Planning that ignores them fails.
Conclusion
IPv6 adoption is not a failure — it’s a slow success playing out across three decades of infrastructure renewal. The protocol works. The address space is adequate for any conceivable future. The ecosystem is maturing.
But the transition reveals something important about how the internet actually works: it is not a designed system that can be upgraded on a schedule. It is an emergent infrastructure operated by thousands of independent actors with different capabilities, incentives, and timelines.
The question was never whether IPv6 would eventually dominate. It will. The question was always how long the overlap period would last, and who would bear the costs of maintaining two parallel protocol stacks across an increasingly complex global network.
The answer, it turns out, is: a very long time, and everyone.