I agree with you. These are not the right ways to look at the value.
Cost savings is too amorphous - you can't value anything based on what you may or may not save people down the road. Basing value on possible cash flows would make more sense, but it's still not correct. The decentralized Bitcoin network will not have the equivalent cash flows of a company. Plus the cash flows in any case would only be a fraction of those of the companies since in theory costs will be lower.
The value of Bitcoin comes from its ability to pay for a valuable service. That valuable service is what miners do: processing transactions that can be global, fast and done with minimal friction. That is actually a useful thing to be able to do - in fact, you cannot do it with the traditional companies out there. So it's worth paying for it. As it happens, the only way to pay for it is via Bitcoins. Therefore the Bitcoins have value.
But how do you set a number on that value? It's too uncertain today. Per your point, it will depend on the demand for Bitcoins. Which will depend on the number of transactions that are eventually being pumped through the entire system.
I write about that in a set of blog posts, starting with this one:
http://www.thebusinessofbitcoin.com/2014/01/the-nature-of-bitcoin.html