Even as the AI boom is impacting businesses the world over, it is creating a serendipitous opportunity for India. As the soaring demand for GPU infra, coupled with power shortage in the West, forces companies to look towards India for data centre (DC) capacity, developers are racing to build AI-ready DCs.

L&T Vyoma, the digital infrastructure arm of Larsen & Toubro, is betting big on this window of opportunity and accelerating the expansion of its Chennai campus, which has transformed from a conventional DC into an “AI factory” serving varied customer needs.

After expanding from basic co-location services to private, public, and sovereign cloud options, L&T Vyoma’s “AI factory” now deploys GPU clusters through a landmark Nvidia partnership.

A walk-through of the DC, located near Sriperumbudur on the outskirts of Chennai, offers a glimpse into a future that is already fast hurtling into the present. Entering the campus — after stringent security checks of the airport kind, complete with scanners, frisking and minute inspection of bags and purses — one feels one is stepping into a rarefied zone.

The eerie quietness intensifies this feeling. Unlike other mega factories that throb with the hum and hustle of people and machinery, the surreally quiet DCs are populated with racks of servers processing huge amounts of data. There are rows of servers in the hallways, alongside power electronics, cooling systems and back-ups for seamless operations. Several areas were off-limits to us, owing to client confidentiality requirements.

Once filled with CPU racks, the Chennai DC now hosts the coveted Nvidia H100s, H200s, and Blackwell GPUs, in step with technological shifts.

The nerve centre of a DC is its control room, where we encounter a small team monitoring everything from the temperature of individual racks to grid connectivity, and the parameters of the UPS and standby diesel generators.

There is also a dedicated security control room, where the CCTVs are equipped with AI-powered software to provide real-time alerts. It’s a well-guarded fortress.

“Chennai is our flagship data centre campus, where we currently have a 30 megawatt IT load DC. We have power ready, network ready, permit ready for the second building, which we are kicking off, as we see significant demand for AI data centres,” says Prashant C Jain, Managing Director, L&T Vyoma, and Head of Electronic Products and Systems.

L&T Vyoma has a second DC campus in Panvel, in Maharashtra’s Raigad district. Alongside Phase 2 of the Chennai and Panvel campuses, the company’s expansion pipeline includes the upcoming campuses at Mahape (in Navi Mumbai) and Bengaluru. The live capacity is around 30 MW (Chennai), 2 MW (Panvel); an additional 200 MW is set to go live in the short- to medium-term.

With the L&T group allocating ₹10,000 crore to the DC business, L&T Vyoma plans to invest about ₹9,000 crore across its campuses in FY27 itself, given the high demand for AI-ready infrastructure.

Jain does not rule out external funding in the future, adding that the nature of the funding would depend on the model of the DCs. Clients include neo-cloud players, banking and financial institutions, and commodity exchanges.

There is significant interest from hyperscalers, he says. The business is also constantly innovating and entering into partnerships to support new-age requirements. For instance, L&T Vyoma recently partnered with Fortanix Inc to offer sovereign, secure AI solutions using Nvidia confidential computing technology.

This model specifically targets regulated sectors in India, ensuring data sovereignty and compliance while enabling secure AI adoption. The two companies will jointly develop solutions, co-sell to industries, and engage with AI model builders to strengthen the ecosystem.

Jain says that while it typically takes 15-18 months to get a DC up and running, L&T with its engineering, procurement and construction (EPC) strengths is constantly innovating to make this more efficient.

“L&T Vyoma has a significant advantage. We as a group have a strong EPC experience and competence, which we rely on. We also have good experience in realty and project management,” Jain notes.

What also sets Vyoma’s Chennai DC apart is the fact that facilities designed for conventional cloud computing were re-engineered to handle dense GPU clusters, advanced cooling systems and massive power loads — a feat perhaps only an infra giant like L&T can accomplish.

L&T Vyoma is also exploring improvements in DC design to avoid dependence on power utilities when scaling up AI-ready DCs.

“We are evaluating technologies like battery energy storage system and grid forming inverters to support the grid. Such solutions will be a good hedge to support the grid, rather than expecting the grid to take more load,” says Jain.

This will also help the company increase its share of renewable energy usage, which is about 40 per cent today. “There is a clear roadmap to expand renewable blend; we are evaluating battery energy storage systems or PPAs (power purchase agreements) that include storage,” he says.

He sees accelerated demand not only from domestic sources but also global clients, as the West Asia crisis makes pockets of the world unsafe for business continuity.

Among the challenges faced, Jain says, one needs to stay constantly updated, given AI’s pace of change. “I might start building assuming a certain base load and cooling requirements but, by the time the DC is up months later, the technology has likely made two shifts,” he points out.

Given the growing competition, what is the differentiator that L&T Vyoma brings to the DC race? “For [building] large AI-ready capacities going forward, there will be a need for strong balance sheets and strong execution capability, and who better than L&T with this advantage,” he stresses.

Published on July 20, 2026