This paper has two goals. First, we demonstrate that standard arguments and methods from production and duality analysis can be used to provide a comprehensive and general treatment of the value of information for a risk-averse firm with expected-utility (linear-in-probabilities) preferences and a general stochastic technology. Second, we place bounds on the value of information for a risk-averse firm and relate these bounds to characteristics of the technology and the producer's preferences. A particularly striking observation that emerges from this representation is that the most common representation of production uncertainty corresponds to a polar case that trivializes the role that information can play in economic decision making under risk.