Economic globalization and the modularization of value chains increasingly challenge long-held conceptual models explaining the spatial evolution of industries. This paper seeks to re-interpret early industry life cycle dynamics by disintegrating an industry's value chain into upstream, core and downstream parts and characterizing each part according to its underlying global innovation system (GIS) configuration. We distinguish between firms in parts of the value chain that depend on formalized, science-based innovation and cater for globally standardized mass markets ('footloose' GIS) and firms in parts of the value chain that rely on spatially more stable GIS structures, in which either the innovation activities or the valuation dynamics (or both) depend on spatial embedding in given territorial contexts. Our hypothesis is that firms which occupy parts of the value chain with footloose GIS characteristics will have shorter survival times than firms which operate in spatially more stable GIS types. Demand-side policies will accordingly produce stronger competitive advantages for firms operating in GIS with spatially stable valuation structures. The empirical context of our study is the solar photovoltaics (PV) industry. We analyze market entry and exit of 129 German and 127 Japanese PV firms from 1960 to 2016 using a Cox Proportional Hazards model. The results support the hypotheses that firm survival and policy effects depend on a value chain part's underlying GIS configuration.