Architectural Agency in the Global Artificial Intelligence Infrastructure: The Case of India

Abstract

This article proposes an analytical model for assessing the architectural subjectivity of countries within the global artificial intelligence (AI) infrastructure. The focus is on institutional mechanisms through which countries with limited technological resources can construct a position within this distributed infrastructure. Rather than approaching AI development through the lens of technological sovereignty, the study conceptualizes AI as a multilayered infrastructure shaped by institutional regimes of coupling and reproduction, rather than by direct control over individual components. Building on an architectural approach, the model distinguishes six infrastructural layers (hardware, compute, data, models and algorithms, competencies and human capital, normative-diplomatic) and applies three criteria of subjectivity (type of access, reproducibility, type of institutional operator). The research combines spatial and institutional analysis with a comparative review of policy strategies and infrastructural practices. The case of India demonstrates how a country lacking sovereign control over key technological nodes can nonetheless develop institutional forms of integration into the global AI infrastructure. The proposed model offers a tool for comparative analysis of national trajectories of AI development.