When evaluating coal and mineral deposits for mining, it is important to keep profit in mind. But the mine will eventually cease production because (1) the mineral is exhausted or (2) the prices have dropped or costs risen so much that what's left is not worth the effort.
Similarly, data mining extracts large quantities of data with the main goal of profit. Because data mining is the extraction of knowledge from large amounts of data, not the extraction of data itself, it seems this is an endless resource.
However, with machine-learning and statistical models improving exponentially, even the abundant resource of data will eventually become worthless, as our reality becomes the mirror of what we have told the system we wish it to be.