Small and medium enterprise networks (SME Networks) are becoming an integral part of the Network Economy. From the 'industrial districts' of the Terza Italia to the entrepreneurial clusters of the Silicon Valley, SME's are a significant driving force of economic growth, job creation, disinflation and productivity enhancement in most industrial countries. After decades of research, these local industrial systems are still poorly understood in terms of their sustained processes of innovation, network interaction and competitive adjustments. While there could be some external economies due to agglomeration, division of labor, specialization and lowered transaction costs, differential innovation, interaction and adjustment capabilities are not fully explained by these mechanisms. A theoretical construct of local industrial system is missing. However, no mechanical or graph theory model of network architecture can substitute for what actually makes people in the network interact in order to become technologically innovative and capable of ongoing adjustment to their competitors. Counting the nodes and edges of graphs would be a poor substitute for understanding SME networks as dynamic ('living') organic systems they are. In this paper we propose a theoretical construct of network production, renewal and adaptation based on autopoiesis (self-production) of living systems.