The article summarizes the experience gained in empirical research and theoretical analysis of information costs conducted by the author. The main conclusion is that information costs are systematically underestimated. That is why the high national economic efficiency of digitalization projects often turns out to be illusory. In support of the conclusion, it substantiates the regularity of the increase in per-unit direct information costs as the volume of data collected by an economic entity grows and explains why microprocessor technologies cannot reverse this trend. Without denying the possible benefits of digitalization, the article draws attention to selected examples of situations where underestimating or neglecting information costs cause serious problems. Among them are negative information externalities, i.e., forcing information providers to bear uncompensated information costs; limited ability to apply the cost accounting principle to assess direct and especially alternative information costs; the limited competitive niche of the agricultural extension service as an intermediary in the implementation of digitalization projects; obstacles to the improvement of operational management based on the use of economic mathematical models (using field cultivation as a case) due to the prohibitively high costs of collecting and presenting data. Case studies are accompanied by suggestions that can potentially mitigate the problem.