In Section 5, we give explicit examples of production, inventory, and acquisition cost functions that satisfy the three, rather general, assumptions.The proofs of the results in this section are given in the Appendix.The first result characterizes the optimal cost functions, given in (2) and (A).For any interest-technology pair s, the optimal cost incurred from period n through period for any a, < a < 1.With the structure of the optimal cost function developed in Proposition 1 and the structure of the unconstrained solution given in Lemma 1, we can now describe the structure of the optimal acquisition policy. Generally, period n's optimal level of automation is independent of the current level of automation (unless the level of automation in place at the beginning of period n exceeds the desired level).Proposition 2. For all a, s, and n, p (s,a) = max{p (s,a),a).It follows from Lemma 1 and Proposition 2 that p (s,a) is nondecreasing in a.The next result asserts that the optimal level of automation declines as the planning horizon shortens.Proposition 3. For all a, s, and n, p (s,a) > p .(s,a). In summary, the inventory carrying cost function is modeled as follows: where s represents the interest-technology pair, and c(0) represents the "base-line" cost per period of producing with no automation. The multiplier T(p-a) represents the diseconomies (economies) associated with purchasing flexible automation incrementally over time rather than all at once. To capture this phenomenon, we assume that T(*) is increasing and concave (convex), T(0) = 0, and T(l) = 1. To better understand the influence of uncertainty on the acquisition process.