TCS Plans a 1GW AI Data Center Campus in Hyderabad

That is the significance of TCS subsidiary HyperVault’s new Hyderabad plan. According to TCS, HyperVault has secured 264 acres for an AI data center campus that could reach 1GW of capacity. The company says the site will be developed in phases for frontier AI companies and hyperscalers, with high density GPU infrastructure, liquid cooling, green energy, and water neutral design principles. TCS and its partners expect investment of up to ₹70,000 crore.

The headline number is impressive. The harder question is what it takes to turn a one gigawatt plan into usable AI capacity.

One gigawatt changes the planning problem

At smaller scale, operators can sometimes think about racks, cooling, network capacity, and power as related but separable workstreams. At gigawatt scale, those dependencies become impossible to ignore.

A campus may have land available while grid capacity is constrained. Electrical capacity may be secured while cooling equipment becomes the long lead item. A building may be ready while the GPU deployment changes rack density assumptions. Even when total megawatts look sufficient on paper, usable capacity can be limited by distribution, redundancy, thermal conditions, or the layout of individual halls.

That is why AI era data center capacity planning has to focus on usable capacity rather than headline capacity. A planned 1GW campus is a statement of potential scale. The meaningful operational milestones will be the capacity actually energized, commissioned, cooled, and occupied.

The project is also an ecosystem test

TCS says the Hyderabad campus is intended to support AI training, inference, and other advanced computing workloads. It also expects the development to stimulate activity across power, cooling, networking, engineering, construction, and operations.

That matters because large AI campuses depend on far more than accelerator supply. They create demand for substations, transformers, switchgear, cooling systems, pumps, water treatment, fiber, monitoring, maintenance, and skilled operations teams.

The physical supply chain becomes part of the compute supply chain.

For planners, this makes early cost estimates more difficult. Data center construction cost is shaped by site work, electrical systems, cooling, racks, cabling, controls, commissioning, and contingency. Higher density AI infrastructure can increase the importance of several of those categories at the same time.

Water neutrality will need measurable evidence

HyperVault says the campus will use water neutral design principles. That is an important commitment, especially as water use becomes a more visible issue around large data center developments.

The phrase alone does not explain how the target will be achieved. Operators and local stakeholders will eventually need practical information about cooling architecture, water sources, reuse, wastewater treatment, seasonal conditions, and how neutrality is measured.

That is the broader pattern across AI infrastructure. Sustainability claims are moving from corporate language into engineering requirements that can affect permitting, operating cost, and community acceptance.

What to watch next

The most useful updates will be operational rather than ceremonial.

How quickly does the first phase move from land to commissioned capacity? How much electrical capacity is secured for each phase? What rack densities and cooling architectures are deployed? How is water neutrality measured? And how much of the planned 1GW ultimately becomes active customer capacity?

The Hyderabad announcement shows how quickly the scale of AI infrastructure is changing. A gigawatt campus sounds like a single project. In practice, it is a chain of interdependent systems, and the weakest link can determine how much compute actually reaches production.

*Originally published on the Sensaka blog.*