When supplying information, agents choose between options that differ both in their contents and in their costs. We establish a “cost-over-content” theorem for a large class of dynamic trading environments where buyers choose from arbitrary sets of processes (experiments) that reveal information to the seller. When all experiments are equally costly, choosing any given experiment is a perfect equilibrium. However, when experiments differ in costs, there is a unique equilibrium: all buyers choose the cheapest experiment, regardless of the information it pro- vides. We explore implications for market performance, privacy, data sale, and defaults in market regulation.