Often, it's better to see the IPv6 address space as 64+a few bits instead of 128 bits, because a lot of the common ways to choose an IP address for a device def

Often, it's better to see the IPv6 address space as 64+a few bits instead of 128 bits, because a lot of the common ways to choose an IP address for a device defines the entire last 64 bit, so subnets smaller than /64 (i.e. prefixes larger than /64) are often impractical. Before privacy extensions and temporary addresses, one common way was to take the 48 bit Ethernet MAC address of the device/interface and pad it up to 64 bit. Nowadays it's all more random-ish. 2^64 is still a very large number, and the article already correctly points out that this allocation is still just a drop in the ocean.