TCS Plans a 1GW AI Data Centre in Hyderabad: Compute Is Becoming Public Infrastructure

When an AI data centre is measured in gigawatts, it no longer belongs only in a technology conversation. It becomes an energy, land, water, employment, transport, resilience and public-policy project.

Tata Consultancy Services and its HyperVault AI Data Centre subsidiary have announced plans for a large campus in Hyderabad. According to reporting published on 5 September 2026, the project is proposed across 264 acres, could scale to one gigawatt and may involve investment of up to ₹70,000 crore. The build is phased and remains subject to demand, technology choices and execution.

What one gigawatt changes

A one-gigawatt campus is not one large server room. It is an industrial system: substations, backup power, networking, cooling, water management, security, construction logistics, chip supply and continuous operations. High-density GPU clusters for training and inference create thermal and power demands very different from ordinary office computing.

The proposal includes liquid cooling, larger power blocks and design principles connected to green energy and water neutrality. Those intentions matter, but they must eventually be measured through engineering plans, power sourcing, water accounting and transparent operating data.

From infrastructure to intelligence

TCS has described an “Infrastructure-to-Intelligence” approach. The phrase captures a genuine strategic shift. Businesses once treated computing capacity as a background utility. In the AI economy, the ability to secure accelerators, memory, energy and data-centre space can decide how quickly a product launches and how much each inference costs.

Telangana’s chief minister described compute as fast becoming public infrastructure. That does not mean every data centre should be publicly owned. It means compute availability increasingly affects education, health, government services, industrial competitiveness and national resilience—the same way other foundational networks do.

The opportunity for Hyderabad

• Construction and specialist operations: a large campus can support direct and indirect employment across engineering, maintenance, security and services.
• AI ecosystem density: nearby universities, start-ups and established firms may benefit from skills, partnerships and procurement.
• Regional infrastructure upgrades: power and network investment can strengthen a wider technology corridor when planned responsibly.
• Research capability: access to advanced compute can support Indian-language models, scientific work and industrial AI.
• Strategic capacity: domestic infrastructure reduces reliance on distant facilities and creates more choices for regulated workloads.

The public-interest questions

The scale also creates obligations. How much firm power will the site require at maturity? Which generation sources will support it hour by hour? How will cooling water be measured during drought? What happens to hardware waste? Which jobs are permanent, and what training will local people need to reach them?

Communities should not be asked to accept vague promises in exchange for precise resource commitments. My earlier article on communities resisting the AI data-centre boom explains why early consultation matters. Responsible infrastructure earns legitimacy through evidence and accountability.

Why the economics favour scale

AI workloads benefit from concentrated power, fast networking, shared cooling and high utilisation. A large campus can serve multiple customers and move between training and inference demand. Yet concentration also creates a resilience question: a fault, cyber incident or grid constraint can affect an enormous amount of capacity.

The correct design is therefore not merely “bigger”. It is modular, observable and recoverable. Power islands, fire separation, network redundancy, supply-chain diversity and tested human command all matter. The same principle guides the Scaler Queen Technology Villages: infrastructure should connect capability with continuity.

The wider business lesson

Small businesses will not build gigawatt campuses, but they will buy services shaped by them. Leaders should understand where their models run, how pricing changes with compute demand, what happens during outages and whether data can move to another provider.

AI strategy is becoming infrastructure strategy because intelligence at scale must live somewhere, draw power from somewhere and answer to someone.

Sources and date note

This article reflects information available on 5 September 2026. Project details and quotations are drawn from Financial Express reporting. Proposed capacity and investment are not the same as completed infrastructure.