Data Transport, AI, and Growth

Artificial intelligence services require model-building and computation (model/compute) together with data transport, the technical and legal work that makes usable data available to computing, storage, monitoring, and learning systems. We develop an endogenous-growth model in which firms invest in both. Because data are nonrival, data-transport barriers create a double wedge: they raise current delivery costs and reduce data available for future innovation. Model/compute innovation alone generally cannot sustain growth when data-delivery costs remain incompressible; data-transport innovation restores balanced growth. Regulations that fragment data use across agencies or jurisdictions reduce growth, while trust-enhancing safeguards can raise it.