India’s state governments are announcing AI Cities, AI hubs, AI corridors, technology parks and futuristic urban districts, with acres of land being identified, investment summits organised and data centres promised. The article raises a fundamental question for policymakers: whether they are building the infrastructure AI needs or merely building real estate around the word “AI”. It argues that AI will require enormous computing capacity, data centres, fibre networks, reliable electricity, renewable energy, cooling systems and semiconductor infrastructure, but questions the causality—whether building an AI City creates AI innovation or whether successful AI innovation creates demand for AI infrastructure. The article identifies a conceptual mistake in treating four different things as interchangeable: AI City, AI Data Centre, AI Innovation Ecosystem and AI Economy. An AI data centre is an infrastructure system of GPUs, networking, storage, cooling, electricity and buildings, while an AI innovation ecosystem requires researchers, universities, PhD programmes, algorithms, datasets, patents, startups, venture capital, industry partnerships and customers. An AI economy emerges when technologies become products, companies, productivity and exports. The article notes that the United States did not become an AI leader by constructing AI Cities first; instead, universities developed researchers, government agencies funded research, industry commercialised discoveries, and venture capital financed entrepreneurs, leading to clusters like Silicon Valley, Boston/Cambridge and Seattle.