Summary form only given. We consider a discrete-time dynamic programming formulation of a technology adoption problem where a firm is offered a technology whose value is uncertain and possibly non-stationary. The firm can choose to adopt, reject or gather more information. After receiving the information, the firm updates beliefs in a Bayesian manner. We show that if the firm is more optimistic about the value of the technology, the optimal action moves towards adoption. Furthermore, if the precision of information increases, the firm obtains more value. These results hold without restricting the beliefs of the firm to a specific probability distribution; the ordering of distributions on "optimism" requires likelihood ratio dominance and the notions of "precision" uses Blackwell's notion of sufficiency