I love finding parallel analogies between BTC and TCP/IP.

There are many areas in the USA that did not have access to high speed broadband until after 2020, many areas still do not have fiber infrastructure and use Starlink or 4G instead. It will take a similar scale of infrastructure building for a new payment system to go mainstream.

Early internet used HTTP for everything, sending passwords in cleartext over the wire. BTC hasn't had its HTTPS moment yet. Maybe something like confidential txns added to base layer, or coinjoin becoming economically cheaper than non-collaborative txns will be the HTTPS moment.

Scaling IPv4 required subnets on top of subnets, making P2P connections a NAT/firewall configuration nightmare. Scaling BTC's UTXOs requires layers on top of layers, makes self custody of small amounts a force-closed channel nightmare.

I also like to draw parallels between the personal computer revolution and AI.

Computers used to be room-sized mainframes. LLMs require entire datacenters to train and run. There's only a handful of flagship models and those models are trying to be as generalized as possible.

Computers today are tiny and highly specialized. A toothbrush today might have a computer inside, it might not be a general computer, but its a highly specialized one that makes the toothbrushing experience better in some way.

Likewise, AI models of the future might be running everywhere, not just in data centers. Highly specialized models integrated into tiny packages that make the tool better in some way.