cafeconomics

For any observer of emerging markets, India’s macroeconomic scoreboard presents a staggering picture of resilience. In a global environment bogged down by geopolitical friction and energy shocks, India’s headline real GDP grew at an enviable 7.7% for the financial year 2025–26. On paper, it is a performance that commands national pride and global celebration. Yet, beneath this golden veneer lies an agonizing macroeconomic puzzle that has split the economic community down the middle. If the headline growth is rocketing past 7%, why are the captains of Indian industry refusing to build new factories? This paradox—recently brought into sharp relief by former Reserve Bank of India (RBI) Governor Dr. Raghuram Rajan—points to a profound structural divergence between Gross Fixed Capital Formation (GFCF) and headline growth. For a nation looking to cement its status as a global manufacturing superpower, understanding this mismatch is not merely an academic exercise; it is an urgent economic di...

Part 1: The Foundations — The Data Centre Life Cycle The critical challenge of managing data centre infrastructure is navigating a massive economic mismatch: the heavy physical shell (the building, generators, and concrete) is designed to last 15 to 25 years, while the logical architecture inside it (the servers, switches, and GPUs) faces economic and technological obsolescence every 3 to 5 years. To manage this contradiction, operators visualize the facility through a cyclical loop. The following comprehensive infographic illustrates the primary phases of a data centre’s life, from ground-breaking to physical destruction. Phase 1: Strategy, Site Selection & Feasibility This is the multi-year planning phase where the biggest long-term cost decisions are locked in. The focus is securing the key inputs: cheap power, available land, and ultra-fast network connectivity. Power and Grid Access : Securing hundreds of megawatts of guaranteed grid capacity, preferably near renew...

​The economics of artificial intelligence are fundamentally distinct from previous technological shifts. Traditional rules governing industrial production, and even early digital products, break down when applied to large language models. This exploration tracks a series of interconnected structural realities: how diminishing returns apply to the foundational inputs of machine learning, whether base model training can ever be truly finalized, the trajectory of synthetic data, and the political economy of user data compensation. These insights represent a record of thinking in progress for an unmapped economic landscape. ​I. Diminishing Marginal Utility in Machine Learning ​Data: A Classic Case of Diminishing Returns ​When evaluating data and compute, data represents a classic textbook resource governed by diminishing marginal utility. The first thousand training examples provide massive learning gains, whereas subsequent millions yield progressively less. Error rates in neural netw...