Because you do not know ahead of time which devices may have such a need, and by allowing for the possibility you open up more flexibility.
> [Residential customers] don't care about engineering, but they sure do create support tickets about broken P2P applications, such as Xbox/PS gaming applications, broken VoIP in gaming lobbies, failure of SIP client to punch through etc. All these problems don't exist on native routed (and static) IPv6.
> In order for P2P to work as close as possible to routed IPv6 in NATted IPv4, we had to deploy a bunch of workarounds such as EIM-NAT to allow TCP/UDP P2P punching to work both ways, we had to allow hairpinning on the CGNAT device to allow intra-CGNAT traffic to work between to CGNAT clients, as TURN can only detect the public-facing IP:Port, hairpinning allow 100.64.0.0/10 clients to talk to each other over the CGNATted public IP:Port.
* https://blog.ipspace.net/2025/03/response-end-to-end-connect...
By having (a) a public address, and (b) a CPE that supports PCP/IGD hole punching, you eliminate a whole swath of infrastructure (ICE/TURN/etc) and kludges.
When it was first released, Skype was peer-to-peer, but because of NAT "super nodes" had to be invented in their architecture so that the clients/peers could have someone to 'bounce' off of to connect. But because of the prevalence of NAT, central servers are now the norm.
A lot of folks on HN complain about centralization and concentration on the Internet, but how can it be otherwise when folks push back against technologies that would allow more peer-to-peer architectures?