General-purpose AI is a name that captures raw capabilities: models, compute, and data centres. Industrial AI names that capability once an economy has built the complements around it through redesigned work, new skills, better data and supporting institutions. This working paper explains why productivity gains have been slow to appear, examines Australia's data centre investment debate, and sets out what the Commonwealth, the states and businesses can do to turn AI capabilit

Why does the same technology lift productivity in some places and disappoint in others? The complementarity thesis explains the gap: returns depend on what technology is combined with. Applied to innovation districts, value comes from complements across placemaking, economics, business, and governance, underpinned by infrastructure. Leadership, collaboration, and learning mindsets become practical policy levers, not soft extras.

People regularly ask whether innovation districts can be deliberately created or whether they emerge beyond policy control. This Insight argues that districts arise through a small number of recurring pathways rather than a single replicable model. Drawing on international experience, it shows why policy outcomes depend less on ambition than on correctly diagnosing which pathway is plausible in a given context. Replicating the Silicon Valley or China approaches is fraught.

Innovation policy increasingly targets precincts and districts. But the evidence used to guide investment is often metropolitan in scale. This Insight argues that innovation outcomes emerge from the interaction of metropolitan enabling conditions and district-level mechanisms. It explains why single-scale analysis misleads policy, contrasts Australian governance with North American and European cities.It outlines a nested framework better suited to AI, productivity, and place