Research on dynamic capabilities emphasizes the importance and role of organizational routines in explaining interfirm differences in performance. While performance differences are well documented, few empirical analyses explore the processes inside organizations that lead to dynamic capabilities or attempt to define and measure their performance effects. This paper examines one type of dynamic capability – the development and introduction of new process technologies in semiconductor manufacturing. This dynamic capability is an important source of competitiveness in the semiconductor industry, given the short product lifecycles, rapid price declines, and rapid technological advances that define the industry. Because much of the knowledge that underpins semiconductor manufacturing is idiosyncratic, firm-level R&D organization and information technology practices that facilitate problem solving and learning-based improvement provide important and enduring advantages. We derive models of the rate of improvement in manufacturing yield (i.e. the quality of production) and cycle time (i.e. the speed of production) following the development and introduction of new process technologies in manufacturing facilities, and test the empirical specifications of these models. The ways in which semiconductor manufacturers accumulate experience and articulate and codify knowledge within the manufacturing environment build new process development and introduction dynamic capabilities that improve performance.