The paper explores the impact of manufacturing learning on the dynamics of production technology development We consider two types of mechanism to develop production technology: endogenous (in-house; on-shop) and exogenous (subcontracting; off-shop) in relation to the 'manufacturing process. Endogenously developed technology is more firm-specific, conducive to in-line learning within manufacturing process, less uncertain in development success than exogenously developed technology. How to allocate managerial resources (also, attention/commitment) between on-shop and off-shop development efforts can be determined by the learning intention/propensity possessed by the managers. In the relatively early stage of decision time horizon, the managers participating in the development process form managers. learning propensity for either on-shop technology or off-shop technology, and make decisions and behave accordingly to the formed propensity: learning is assumed to affect the implementation cost. Since in the early period the benefit from technological improvement outweighs the implementation cost, the learning impact seems minimal. As the terminal point of decision horizon approaches, the implementation cost becomes dominant over the benefit (since the benefit decreases more rapidly than the cost does). The firm may eventually face a time point at which the ''learning-induced'' bias to utilize the particular ''chosen mechanism'' to develop technology becomes so enormous that the firm cannot control its own dynamics with minor adjustment: the dynamics of the later period was already prophesied by the initial choice of learning propensity (freezing effect of learning intention; catastrophic effect of learning prophecy). Top management should understand the early formation of learning propensity by (middle), managers in order to optimally control the dynamics of production technology development.