A report in The Week published on September 27, 2026, filed from Tarluvada in Andhra Pradesh, puts numbers on India's AI data centre build-out and on what it asks of the places that host it. More than 80 data centre facilities are under construction in India, "nearly 10%" of those being built worldwide. The stated rationale is data sovereignty: India produces about 20% of the world's data but hosts only about 2% of it, and would need roughly 16 GW of capacity to keep much more of it at home.
Tarluvada is the site of Google's 1 GW complex, part of a $15 billion hub built with partners including Adani and Airtel's Nextra Data. The site covers 266 acres, and more than 50 Dalit families there were paid Rs 40 lakh per acre for their land. The report's point about jobs is blunt: a 1 GW data centre may create about 1,000 permanent jobs, "possibly fewer". That is a small number against the land, power and water it uses.
Power and cooling are the harder constraints. The Week cites CEEW figures that 57% of Indian districts face high to very high extreme-heat risk, which raises cooling loads. It also reports more than 50 GW of renewable capacity stranded by transmission bottlenecks. Aroma Solar CEO Mayank Garg puts the mismatch simply: solar runs "six to eight hours a day while AI data centres run round the clock".
This is a feature rather than breaking news, but it is the right counterweight to the GPU-count headlines this digest has covered, such as the IndiaAI Mission's subsidised compute and new hyperscaler regions. Compute capacity is being bought quickly. Firm, round-the-clock clean power, transmission and water are slower to build, and they will decide whether India's AI capacity is cheap, reliable and locally accepted.
India's data centre boom is real, at 80+ facilities, but it creates relatively few jobs per gigawatt and depends on round-the-clock power and transmission that the grid does not yet have, which will shape where and how fast AI capacity lands.