Why even Ethereum 2.0 will not be as scalable as you think

When reading this keep in mind that I'm not talking about the next few years but about a much much wider timeframe. I also expect a wave of downvotes but if you have constructive criticism I encourage you to write it.

A lot of people throw around the 100k TPS of Ethereum 2.0 and how it'll solve scalability forever... but the reality is not that simple. Let me expand on this:

Ethereum 2.0 will be split into 64 different chains (all kind of equal to the current Ethereum) that connect to the beacon chain to improve scalability. The high TPS numbers that are thrown around are in theory a combination of all the shards running asynchronously + layer 2 solutions BUT the assumption here for 100k TPS is that all the load on Ethereum will be split to fill equally all shards, which is not realistic, let me explain why...

The biggest issue with this architecture is that interoperability is kind of thrown out the window, for two shards to interact with each other they first have to go through the beacon chain and that takes about 6-12 minutes due to security which means that everything that needs synchronous and instant interoperability with everything else will be forced to go into the same shard which means that they will be bottlenecked by the shard's slow speed (less than 40 tps on chain, maybe 1-2k tps WITH layer 2 scaling).

At this point you'll say that "2k tps is super scalable, Visa is doing that much on a daily basis". That's wrong. We shouldn't compare smart contract platforms that are supposed to be the backbone of the Web3 with simple payment processors like Visa. Smart contract transactions can be triggered automatically and they can in turn trigger tens or maybe hundreds of other transactions, a smart contract platform is insanely more demanding than a value transfer protocol like Bitcoin, Monero, Nano etc.

A better example are the Nasdaq Markets which ACTUALLY handle at least 75000 TPS of orders, cancellations and trades continuously and scales to 250000 tps if needed. How can we decentralize JUST this one exchange on an Ethereum shard in a future where Web3 will have a billion users?

I'm not here to break your bubble, Ethereum 2.0 will hold up for a many years without big issues and it will onboard millions of new people into Web3, but it will not be able to be the backbone of the new internet by design.

Ethereum 3.0?