Power, water, jobs: Google’s India AI mega‑centre ignites debate over costs, benefits and community impact
- ID
- 34167
- Status
- summarized
- Published
- 11 Oct 2026, 9:00 PM
- Fetched
- 11 Oct 2026, 9:22 PM
- Provider
- Malay Mail Tech
- Category
- malaysia-tech
- Original URL
- https://www.malaymail.com/news/world/2026/10/11/power-water-jobs-googles-india-ai-megacentre-ignites-debate-over-costs-benefits-and-community-impact/238656
- Source URL
- https://www.malaymail.com/feed/rss/tech-gadgets
Summary
- Score
- 4.5
- Created
- 11 Oct 2026, 9:22 PM
- Tags
- Audience
- developersai_ml_learnerssaas_startup_founders
What happened
Google is building an AI and data centre hub at Tarluvada village near Visakhapatnam, India, as part of a $15 billion investment, and the project is drawing local pushback over power and water consumption and how the economic benefits are shared. In response, Google says it ran extensive independent assessments and will use air cooling instead of water cooling to reduce strain on local water supply. The provided excerpt is truncated, so it contains no numbers on jobs created, power capacity, water volumes, or construction timelines.
Why it matters
The one concrete, transferable detail here is the cooling decision: Google is publicly committing to air cooling rather than water cooling to defuse community opposition. If you sell into, build on, or advise on AI infrastructure in Southeast Asia, that reframes cooling architecture as a siting and social-license question, not just a PUE question — expect air-cooled designs to be the easier approval path, with the tradeoff that you may get lower rack density and different workload ceilings. The $15 billion figure is also a reminder of the scale of capital chasing AI capacity in the region, which is context for anyone forecasting compute pricing or colocation availability. Everything else (jobs, power, benefits) is asserted in the headline but not substantiated in the available text.
Discussion angle
Ask whether air-cooled AI data centres will become the default in tropical Southeast Asia, and what that means for the rack density and GPU workloads you can realistically host locally versus renting in a water-cooled region — no one in the article gives hard numbers, so this is a genuine open question.