IPv6 is 30 years old. It still carries less than half of global internet traffic. 91.4% of IoT devices break on IPv6-only networks. And the industry keeps pretending the migration is "right around the corner."
What if we just... skipped it?
The IPv6 Problem Nobody Wants to Admit
In 1994, the IETF started designing IPv6 to fix IPv4 address exhaustion. Thirty years later:
- Global adoption is ~49%
- The US is barely above 50%
- Only 8.6% of IoT devices function on IPv6-only networks
- Dual-stack (running both IPv4 and IPv6) doubles operational complexity
- IPv6 addresses are unreadable:
2001:0db8:85a3:0000:0000:8a2e:0370:7334
The dual-stack transition model — requiring every device, application, and network to simultaneously support both protocols — proved commercially unacceptable. That's not my opinion. That's from the IPv8 IETF draft.
Enter IPv8: A 64-bit Protocol That Actually Makes Sense
On April 14, 2026, an IETF Internet-Draft called draft-thain-ipv8 proposed something radical: stop trying to force IPv6 migration and build a protocol that works with IPv4, not against it.
IPv8 is a 64-bit protocol. The address is simple:
[32-bit ASN prefix].[32-bit host address]
Here's the genius: IPv4 is a proper subset of IPv8. When the ASN prefix is zero, an IPv8 address IS an IPv4 address:
IPv4: 192.168.1.1
IPv8: 0.0.0.0.192.168.1.1 ← same thing
IPv8: 0.0.251.240.192.168.1.1 ← now on ASN 64496
No dual-stack. No flag day. No migration pain. Your existing IPv4 network already speaks IPv8 — it just doesn't know it yet.
I Built It. It Runs. For ₹0.
I didn't just read the draft. I implemented it. The world's first IPv8 firmware stack, running on my laptop with zero hardware cost.
Here's what the stack implements:
- IPv8 Addressing — 64-bit addresses with ASN dot notation
- DHCP8 — One discover request gives you EVERYTHING (address, DNS, NTP, auth, telemetry, routing). Compare that with traditional networking where you configure DHCP, then DNS, then NTP, then auth separately.
- DNS8 — Resolver that requests both A (IPv4) and A8 (IPv8) records
- OAuth8 — Every device authenticates via JWT tokens from the Zone Server
- NetLog8 — Real-time telemetry pipeline
- XLATE8 — IPv4 backward compatibility translation
- Zone Server — The brain that runs all services on a single platform
The boot sequence is beautiful in its simplicity:
Device sends DHCP8 Discover
↓
Zone Server responds with ONE message:
├── IPv8 address
├── DNS8 endpoints
├── NTP8 endpoint
├── OAuth8 server + JWT token
├── NetLog8 telemetry endpoint
├── WHOIS8 route validator
└── XLATE8 translation gateway
↓
Device is fully operational. One request. Done.
Here's real output from my simulator running 3 IPv8 devices:
◆ DHCP8 DISCOVER from AA:BB:CC:00:00:01
✓ DHCP8 ACK → 0.0.251.240.10.0.0.100 assigned
All Zone Server endpoints delivered in single response
✓ Authenticated as: dev-aabbcc000001
◆ DNS8 QUERY for "api.example.com"
A8 → 0.0.251.240.93.184.216.34 (IPv8) ← preferred
A → 93.184.216.34 (IPv4 fallback)
IPv8 Packet: Version 8 | Header 24 bytes | Valid ✓
Device #1 fully operational on IPv8!
Why IPv4 → IPv8 Makes More Sense Than IPv4 → IPv6
| IPv6 | IPv8 | |
|---|---|---|
| Address size | 128-bit (overkill) | 64-bit (enough) |
| Backward compatible | No | Yes — IPv4 is a subset |
| Dual-stack required | Yes | No — single stack |
| Address format | 2001:0db8:85a3::8a2e:0370:7334 |
64496.10.0.0.100 |
| Header size | 40 bytes | 24 bytes |
| Management | Fragmented (separate DHCP, DNS, NTP, auth) | Unified Zone Server |
| Hardware change | Often required | Software-only update |
| Years of effort | 30 years, <50% adoption | Draft published April 2026 |
The IoT Argument
This is where it gets real. There are going to be 60+ billion IoT devices soon. IPv4 can't handle them. IPv6 was supposed to be the answer, but 91.4% of IoT devices don't work on IPv6-only networks.
IPv8's 64-bit space gives every ASN holder 4.3 billion host addresses. That's enough. And because IPv4 is a subset, existing devices don't need to change. New devices get native IPv8. The transition happens naturally, not by force.
Try It Yourself (2 Minutes, Zero Cost)
git clone https://github.com/santhanuss/ipv8-firmware.git
cd ipv8-firmware
mkdir -p build
gcc -Wall -std=c11 -O2 -I./include sim/zone-server.c -o build/zone-server
gcc -Wall -std=c11 -O2 -I./include sim/ipv8-device.c -o build/ipv8-device
./build/zone-server &
sleep 1
./build/ipv8-device --id=1
That's it. You'll see a device boot on IPv8 in your terminal. No hardware. No emulator. Just gcc and 2 minutes.
GitHub: github.com/santhanuss/ipv8-firmware
What's Next
This is a reference implementation of a week-old IETF draft. It's not production-ready. But it proves the concept works. Here's the roadmap:
- ESP32 port via QEMU emulator (free, no hardware)
- lwIP fork with AF_INET8 socket support
- Device-to-device IPv8 packet routing
- Web dashboard for Zone Server telemetry
- Real hardware testbed (2x ESP32-S3, ~₹1600)
The Bottom Line
IPv6 tried to replace IPv4. After 30 years, it hasn't. IPv8 doesn't try to replace IPv4 — it embraces it. IPv4 is literally a subset of IPv8. That's not a workaround. That's the design.
The question isn't whether IPv6 will eventually win. The question is whether a simpler path exists. I think it does. And now there's working code to prove it.
The IPv8 firmware stack is open source under MIT license. Built based on draft-thain-ipv8-02 submitted to the IETF on April 14, 2026.
Star the repo if this interests you: github.com/santhanuss/ipv8-firmware
Top comments (1)
Great to discover IPv8, never heard before.