TCS's ₹70,000 Crore AI Data Centre Bet: Why HyperVault Could Change the TCS Business Model

TCS is making one of its biggest moves yet into AI infrastructure. Its subsidiary HyperVault has secured 264 acres in Hyderabad to build an AI data centre campus of up to 1 GW, with HyperVault and its partners expected to invest up to ₹70,000 crore at full build-out.

But this is not simply a massive capex announcement. The more important story is that TCS is moving beyond AI services and into the physical infrastructure that powers AI, potentially changing part of its traditionally asset-light business model.

For investors, the opportunity is large, but so are the questions around funding, utilisation and returns.

What Is TCS Building Through HyperVault?

The Hyderabad campus is being designed specifically for AI workloads.

Unlike a conventional enterprise data centre, AI infrastructure needs to support large clusters of GPUs used for training models, running inference and other high-performance computing tasks.

HyperVault plans to build the Hyderabad facility with capacity of up to 1,000 MW, making it one of the largest planned AI infrastructure campuses in India at full scale. The company says it will serve frontier AI companies and hyperscalers and will use technologies such as liquid cooling and high-density computing infrastructure.

The campus will also use green energy and water-neutral design principles. But one detail is more important for investors than the technology itself.

The project will not be built in one shot. HyperVault plans to develop the campus in phases according to customer demand.

That gives TCS some protection against one of the biggest risks in the data centre business, building expensive capacity before enough customers are available to use it.

Is TCS Really Spending ₹70,000 Crore?

No, not on its own. The official filing specifically says HyperVault and its partners are expected to invest up to ₹70,000 crore.

That wording is important because headlines can easily make the announcement look like a ₹70,000 crore direct capex commitment from TCS.

HyperVault already has TPG as a strategic partner. In November 2025, TCS announced that TCS and TPG together would commit up to ₹18,000 crore to HyperVault over the next few years. TPG can invest up to ₹8,820 crore and is expected to eventually hold between 27.5% and 49% of the business. HyperVault is also expected to use debt financing.

TCS itself said that bringing in TPG would help reduce its capital outlay and improve shareholder returns.

So the real funding structure could involve TCS equity, TPG capital, debt and potentially other project partners.

The exact split for the Hyderabad project has not yet been disclosed. That may ultimately be one of the most important numbers for investors to track.

How Big Is ₹70,000 Crore Compared With TCS?

The headline becomes much more meaningful when compared with the size of TCS itself.

TCS reported FY26 revenue of around ₹2.67 lakh crore and net income of roughly ₹52,820 crore.

The potential project investment is roughly 26% of TCS's FY26 revenue and larger than one full year of TCS net profit.

That shows just how large the project is. But it does not mean TCS itself will fund the entire amount.

The project will be phased and partner-backed, which is why comparing ₹70,000 crore directly with TCS's cash balance or annual profit can be misleading. What matters more is TCS's actual share of the capital.

Why HyperVault Fits TCS's AI Strategy

TCS is not entering the data centre business simply because AI infrastructure is attracting investment.

Management is trying to position TCS across a broader part of enterprise AI spending.

TCS describes its strategy as moving from "Infrastructure to Intelligence." CEO K Krithivasan said HyperVault brings AI-ready infrastructure together with TCS's cloud, engineering, enterprise transformation and AI capabilities.

In simple terms, a customer may need more than just an AI application. It may also need compute infrastructure, cloud architecture, data engineering, cybersecurity, integration and ongoing AI management.

Historically, TCS mostly earned from the services layers of this chain. HyperVault gives it a way to participate at the infrastructure layer as well.

That could allow TCS to capture a larger share of each customer's AI spending.

TCS Already Has a Growing AI Business

This strategy is also being built on top of an existing AI services business.

TCS has said it has completed more than 5,500 AI engagements and increased AI revenue to around $2.6 billion. It has also expanded relationships with companies including OpenAI, Google, Anthropic and Mistral.

This creates an interesting potential flywheel. Existing TCS customers may need AI infrastructure.

HyperVault could provide that infrastructure. Those infrastructure customers may then require additional cloud, engineering, consulting and AI implementation services from TCS.

If the model works, HyperVault could become more than a standalone data centre business. It could strengthen TCS's broader AI services platform.

Why the OpenAI Relationship Matters

One of the strongest signs of potential demand is TCS's partnership with OpenAI.

TCS and OpenAI have agreed to a multi-year partnership to develop AI infrastructure in India. The initial phase involves 100 MW of capacity, with an option to scale toward 1 GW.

TCS's annual report also describes OpenAI as an anchor customer for the first 100 MW phase of HyperVault's broader build-out.

This matters because data centre economics depend heavily on utilisation.

A project with committed customers before construction is less risky than one built purely in anticipation of future demand.

But there is an important distinction. TCS has not said that OpenAI will occupy the entire Hyderabad campus.

So the OpenAI agreement should be viewed as evidence of anchor demand for HyperVault, not proof that all 1 GW of Hyderabad capacity is already contracted.

The Biggest Change: TCS Is Adding a More Capital-Intensive Business

This is probably the most important financial change for investors. Traditional IT services requires relatively little physical capital compared with sectors such as telecom, manufacturing or infrastructure.

TCS primarily spends on people, delivery centres, technology and intellectual capabilities.

AI data centres are different. They require substantial investment in land, buildings, power infrastructure, cooling systems, networking equipment and computing capacity.

That changes the economics.

A services company can generally scale revenue without needing to invest huge amounts of physical capital every time it grows.

A data centre must first build capacity. Only then can it earn revenue from customers using that capacity. That makes utilisation extremely important.

Imagine two identical data centres. One operates at 90% utilisation. The other operates at 40%.

Both may have similar infrastructure, financing and depreciation costs. But the first is spreading those costs across much more revenue.

That is why HyperVault's phased development model matters. It gives TCS an opportunity to add capacity as demand becomes visible rather than building the full 1 GW upfront.

What Are the Main Risks?

The opportunity is large, but the ₹70,000 crore headline does not automatically mean value creation.

• Capital intensity: HyperVault requires much more upfront investment than TCS's traditional services business. If TCS eventually has to contribute more capital than investors expect, it could affect free cash flow and capital allocation.
• Utilisation: Expensive infrastructure must remain well occupied. Low utilisation can hurt returns because financing, depreciation and maintenance costs continue even when capacity is underused.
• Technology obsolescence: AI hardware evolves quickly. HyperVault will need to keep infrastructure competitive as GPUs, cooling requirements and computing architectures change.
• Power and execution: A gigawatt-scale campus requires enormous and reliable power availability. Building infrastructure at this scale while maintaining cost discipline will be a major execution challenge.

Ultimately, these risks feed into one question: can HyperVault generate returns high enough to justify the additional capital?

What Should TCS Investors Track Next?

• Funding mix: Investors need clarity on how much of the ₹70,000 crore will ultimately come from TCS versus TPG, debt and other partners.
• Customer commitments and utilisation: More long-term hyperscaler and AI company contracts would reduce the risk of excess capacity.
• Capacity ramp-up: Announced GW capacity matters less than how quickly operational capacity comes online and starts generating revenue.
• Returns from HyperVault: Revenue growth alone is not enough. Investors should eventually track cash generation and return on invested capital.

Author's Take

HyperVault matters because it could change how TCS participates in the AI economy.

Instead of earning mainly from consulting, cloud and software services, TCS is trying to capture part of the infrastructure spending underneath those workloads as well.

That could strengthen customer relationships and allow the company to participate across a larger portion of enterprise AI spending.

But the ₹70,000 crore headline should not be mistaken for guaranteed value creation.

The three numbers that ultimately matter are much simpler: how much capital TCS itself contributes, how much of the capacity customers actually use and what return HyperVault earns on that capital.

If TCS can keep utilisation high while using partners and debt to control its own capital commitment, HyperVault could become a valuable extension of its core AI strategy.

If utilisation disappoints or capital requirements rise sharply, the move toward a more asset-heavy model could put pressure on the economics investors have traditionally associated with TCS.

That is why HyperVault should be viewed not merely as another data centre investment, but as an important test of whether TCS can evolve from an IT services company into a broader AI infrastructure-to-intelligence platform.