I couldn't explain it better than a member of the Slovenian Bitcoin Association, so I am quoting his post.
The conclusion is that price will stay unstable due to increasing adoption? I'm not so sure. At least it need not rise proportionally to adoption... One quite simple way to look at this is Fisher's formula for velocity of money: M * V = P * T.
M is money supply (fixed in case of bitcoin), T is num of transactions (rising in orders of magnitude). So to keep the equation balanced, the price (P) can rise but so can V (velocity of money, "frequency at which one unit of currency is used to purchase domestically-produced goods and services within a given time period"). Velocity has so far in history been rather stable, but given the usefulness of bitcoin for transactions (not very disputed even among critics), I wouldn't dare say V for bitcoin is going to be stable or at the same approximate level as for regular fiat money.
That's why I'm uneasy about projections like "the use of bitcoin will increase 1000x, therefore the price will rise 1000x". I believe the first part, but I don't think the second part follows necessarily.
Not a completely bad point, but in order to compensate for growing adoption, V would need to keep on rising. Exponentially, to compensate for exponentially rising adoption/usage. Not likely. Probably impossible in fact, considering the technical limitations and tradeoffs of the network.
In a thread a while ago, I made a (relatively simplistic) argument that, if Bitcoin is used as a method for Internet payments (and only for that) on the order of yearly Paypal volume today - a comparably modest goal, I'm sure you'll agree - and making a few extremely conservative estimates for the necessary parameters involved (including V), we arrive at a valuation of around $2000 per unit. The single biggest not-so-conservative assumption I am making is that those transactions are actually taking place
on chain, not
off chain, which could be an (unfortunate, because trust requiring) development.
(
Link to my post, the calculation is under Addendum #2)
Back to the topic: I set V_year = 10, which is a conservative value considering that actual velocity of Bitcoin is probably a lot lower currently. M1's V_year seems to be able to reach into the range of 10 (
WP link), but usually is lower, so if the Bitcoin network would be stable anywhere 10 it'd be a) pretty efficient, and b) the argument you mention, that increased adoption and usage doesn't lead to proportionally higher valuation, doesn't hold.