India's data centre build-out has accelerated sharply. Operational stock reached roughly 1.8 GW of IT capacity in the first half of 2026, with about 258 MW added in that half alone — a 59% jump year-on-year — and industry forecasts point to more than 7 GW by 2030. Chennai has become the country's second-largest hub after Mumbai, taking about 26% of absorption in H1 2026 against Mumbai's 38%. Its advantages are structural rather than incidental: it is the principal submarine-cable gateway on India's east coast, giving low-latency routes to Singapore and Southeast Asia, while Mumbai's landing stations face west towards Europe, Africa and West Asia. Tamil Nadu has also used electricity duty exemptions and land support to court digital infrastructure investment. For an AI Factory, where power availability is the real bottleneck, a site already engineered for gigawatt-scale draw matters more than proximity to customers.

Three things make this more than a large construction contract. First, the direction of the money: a US AI company is buying Indian compute capacity rather than an Indian company renting foreign capacity, which is the pattern policy has been trying to reverse. Second, the executor: L&T is an engineering and construction house, and its move into owning and operating AI compute marks the convergence of India's infrastructure industry with its digital economy — the constraint on AI capacity in India is grid connection, land and cooling engineering, which is precisely what an EPC major supplies. Third, the scale: 10,000 top-end GPUs in a single cluster is comparable to the entire GPU commitment of the IndiaAI Mission's compute pillar, in one private facility. For an economy-section answer, this is a clean illustration of how data localisation rules, cheap land and power incentives, and AI demand combine to pull physical infrastructure investment into a country.