Of course English is spoken in other countries, and other countries have high numbers of internet users, but it does not follow that English is a commonly used language for internet users in other countries. Most Chinese are probably speaking Chinese, most Indians are probably speaking Hindi.

The IPv6 graph you linked shows that adoption is still less than 50%, and I'm not clear on their methodology... does "users that access Google" mean users with Google accounts? or individual users that use google.com? or does it include all of their cloud services? do web servers linking content from Google Ads count? does this data represent mostly end users, or also infrastructure connections?

These graphs do not give an indication of how many users per country there are. There are in fact statistics on that which expectedly show China and India on top.

Well sure, but people from those countries are far less likely to be speaking English, which is why I said:

It is entirely rational to assume that an English-speaking person on the Internet is from the US, given no other information.

The prevalence of internet use in countries with primary languages other than English has no bearing on this statement.

The point of using the IP address statistics is to show that the vast majority of websites on the Internet were created in the US for the US market, and that is still true today.

On a side note, the distribution of addresses is unbalanced but it isn't "bad". It is a consequence of a system growing over time. Communications infrastructure cannot pop into existence everywhere all at once, and realistically not many people outside the US had any interest in the internet in 1983.

I hate to say it, but, these things don't change a lot, or quickly.

IP blocks, or "large groups of IP addresses" are assigned to regional internet registries, or RIRs which then hand them out from there. There's a couple RIRs. I think five in total? ARIN covers North America, and has, by far, the most IP addresses given out.

There's also RIPE, in Europe, APNIC, for Asia and the Pacific areas, including China and Australia, AfriNIC, which is basically all of Africa... If that wasn't obvious. And lacnic, which is South America.

Large IP blocks can, but rarely ever do, get transferred between RIRs.

But wait, it gets more complicated. IP addresses allocated in one region could be used anywhere in the world. The vast majority are not, and it's important to note that because of global routing, you can't have a block smaller than 256 addresses allocated in the default free zone (DFZ). The DFZ is the part of the internet that doesn't have a "default gateway". All routes are advertised, and by those advertisements are learned by others. The routers in the DFZ only have so much memory, and there was a crisis a while ago when the memory of most of the routers in the DFZ were dangerously close to being full.... That was around when ipv6 was first switched on. The routing memory is extremely fast, because it needs to be. Looking up a route in a table with a million+ entries takes time, but that time needs to be so short that latency is effectively mitigated. So that memory is some of the fastest used in tech at times, notwithstanding newer technologies.

I'm off topic. Anyways, my point is, ARIN is big. They have a lot of IPs. However allocation doesn't and shouldn't imply usage. A large number of addresses are allocated for US military use that are basically unseen in the internet. There's a few infamous /8 blocks of around 16.7 million addresses that don't get advertised and can't be used by anyone besides the US military. I forget which branch of military owns it. They've owned it since the internet started giving out allocations (more or less) and today one of those /8 blocks is worth billions, with a cost of about $50 per IP.

So yeah, the US has a lot of IP allocation, they also have a large amount of unused IP addresses.

I would love to see a more recent source if you have one.

Regardless, possession of IP addresses doesn't change all that much. In the early days a company could buy an entire Class A (1.X.X.X) address space comprising 16million+ addresses for their private use. There are still many companies holding large blocks of addresses, and most of those companies are in the US, and they don't just give up those addresses.

The point being, there's significant resistance to redistributing addresses once they've been allocated. They don't change hands terribly often (and keep in mind we're talking about actual internet addresses, not local network addresses that are being dynamically assigned and NATed across router domains).