How often does a person create a website and never update it? I could be off base here, but it seems like your picture of new data creation/uploading mirrors that of a static museum–all the exhibits are put in place once, and no one touches it again. Even this we know to be untrue. While many museums have exhibits that do not change, every museum brings in new, fresh, rotating exhibits to keep things interesting and (re)attract the audience.
On a global scale, we are increasingly creating excess data storage capacity with every passing moment, so running out of storage space isn’t a concern. It’s more a question of distribution and allocation. Now, I could be reading too much in between the lines, but one could say your concern is less about the sustainability and more about profitability. That is to say, will farming SAFE prove as profitable to farmers as mining Bitcoin, for example, is to miners.
On that note A) I think the SAFE developers are more concerned with viability than profitability. If equilibria for the network leaves a razor thin margin (or 0% margin), well then, one might say supply and demand are in perfect balance with no consumer/producer surplus or deadweight loss. However, to the extent that an old microeconomics professor of mine is correct in stating that “economics is about learning how to exploit people for private gain” then “price is never based on worth but rather willingness to pay.” One of the things that intrigues me most about Bitcoin is that it shows how decentralization (and anarchy) are (as of yet) unsustainable because the supplier’s (or in this case, miner’s) drive to leverage economics as a tool of exploitation will lead to increasing centralization. One could say that another name for Smith’s invisible hand is “greed,” and consumers fall for it every time, which brings me to…
B) I think the network is (being) designed to function in such a way that farmers will not farm if they do not find it profitable (enough) to do so, nor will users participate if they do not perceive they are receiving enough value to justify the expense. In light of that assertion, I think the network (and external exchanges) will modify the price such that farmers are willing to farm and users are willing to participate (even if that results in price > marginal cost or a state of consumer surplus). On that note, the current let’s-turn-everything-into-a-subscription pricing model is a perfect example of economics as willfully applied and willingly accepted exploitation. From software (e.g., adobe) to cars (ride-sharing), we are seeing ownership devolve into renting. Why? Not because price skims so close to marginal cost that suppliers must do so in orders to stay afloat, but rather because it is an easy way to guarantee that profitability moves up and to the right (and why we think that an already highly profitable business must always be increasing in profitability is something I’ve never been able to understand except for to chalk it up to some shade of greed. For example, what’s wrong with maintaining a 30% margin in-perpetuity?). The thought that we must continually pay for data storage (rather than pay once; store forever) is to fall prey to the notion that the subscription model is both natural and structurally necessary.
TL;DR-I think the network will be dynamic enough to keep price at such a level that farming remains sustainable so long as users feel that there is inherent value in (and therefore use) SAFE. Subscription pricing models are not definitively natural or structurally necessary, so pay once/store forever is not inherently unviable.