That's just a unit of account and a very poor one at that. If everything was priced in BTC then how much would it cost to secure the network?
My argument does not rely on any exchange rates, and is about waste of real resources. Substitute "1000 ounces of gold" or "200,000 blocks of cheese" as the price of a bitcoin and it applies just the same.
BTC is not going to be worth 1 million a coin in 2 years. That is where you are wrong so the rest of your argument makes no sense.
Try to look at the general point. If it goes to $50k, the same logic applies.
Let's assume it goes to 1 billion. Just for fun.
For your computing , that would mean 2000% of the wealth produced globally.
And now , it seems there is a problem with this , right?
Spot it and you have the answer to your question in the first place.
The error in your analogy is that it's impossible for us to spend 2000% of our current GDP on anything in two years, whereas it wouldn't be impossible for us to spend 2%.
Nice analysis! I think this shows that the bitcoin network is perhaps using more resources than necessary even right now. It is using more power than Facebook, but doing a lot less useful processing (look at the number of requests/transactions Facebook is processing). One could argue that the Bitcoin network needs to be extra secure in the early stages so as to avoid an attack that could kill it before it develops. On the other hand, one could argue that a 51% attack might kill Bitcoin (or might not -- it could easily recover after some manual editing of the blockchain), but that another cryptocurrency would just take its place, and that protecting Bitcoin itself is not that critical.
I also think in the future, people will spend up to almost the entire value of block rewards to get them (once ASIC manufacturers create the technology to make their latest ASICS, producing extra ones is very cheap). So IMO your arguments based on difficulty are red herrings in the long run. The cost to run the Bitcoin network will be about the value of the block rewards.