You have found a fundamental value problem, which has been discussed here at length earlier. The fact that bitcoin has value, contradicts Mises' regression theorem, which basically states that the value must be based in part on yesterdays value, and ultimately, from the time deep in history, when some commodity got exchange value because it was the preferred commodity to hold for later. (Discounting debt as money, which could have existed before or after or at the same time).
Some believe that this means that bitcoin must have some value for direct use, in addition to it's money (exchange) value. I do not agree, I think the vision of the future usage of bitcoin as money, was enough to give it initial value, and that the regression theorem has to be shelved.
What is your thought to the argument that Bitcoin does, in fact, comply with the regression theorem since an established value for "yesterday" was assumed as soon as the earliest exchanges had pegged a USD valuation for means of exchange and trade became possible? Continuing to trace back you will find the pre-monetary commodity that makes the point of reference all that's necessary, yes?
Secondly, users might have found further value in bitcoin's inherent properties, such as its low costs, deflationary nature, pseudo-anonymity, lack of central authority, etc, as opposed to an initially established value being a purely speculative one.