A capacity expansion problem is one of choosing the timing, location, and sizing of productive capacity to respond to demand forecasts at minimum cost. Capacity expansion models for realistic problems tend to be too large for present methods that seek exact optima. A problem with myopic structure can be decomposed into parallel subproblems and more easily solved. This paper proposes an efficient heuristic that utilizes successively updated myopic approximations for large, generally non-mypopic, capacity expansion problems. Following termination of this first phase of the algorithm, a capacity exchange heuristic attempts to further improve the approximate solution.