September 7, 2026 · Infrastructure
Tata Consultancy Services said a unit will invest up to $7.4 billion in an AI data-center campus, according to Reuters. Follow-on coverage described a HyperVault project aimed at a roughly 1-gigawatt, liquid-cooled campus on hundreds of acres in Hyderabad, built for frontier AI workloads rather than ordinary enterprise hosting.
The number matters because AI infrastructure is no longer a U.S.-only arms race. India wants to be a training and inference geography, not only a services exporter. A gigawatt-class site is the scale at which power contracts, water, and grid politics become the product, not the footnote.
It also lands in a week when Meta has been reported to be planning a multi-billion-dollar first AI data center in Canada, and when local opposition has already delayed large U.S. projects. Capital is chasing electrons wherever permitting is faster. Hyderabad is bidding to be one of those places.
For model builders, more non-U.S. capacity is a hedge against export rules, power shortages, and community vetoes. For India, it is a bid to capture more of the value stack than body-shop software services. The $7.4 billion is the headline; the 1 GW figure is the real claim.
Key takeaway. TCS is putting up to $7.4 billion behind a gigawatt-scale AI campus in Hyderabad. Compute geography is spreading, and power—not just GPUs—is the scarce input.
Photo: Unsplash (photo-1558494949-ef010cbdcc31). Sources: Reuters; AI Weekly.
