How modular design is reshaping India’s datacentre landscape

The India opportunity

Amid India’s booming datacentre market, there is significant scope for modular infrastructure given the country’s diverse energy and connectivity challenges, said Biswajeet Mahapatra, principal analyst at Forrester. “Modular infrastructure can support decentralised AI applications, such as edge computing for rural agritech [agricultural technology] solutions and energy-efficient healthcare devices,” he added.

Naresh Singh, senior director-analyst at Gartner, noted that with the growing number of AI workloads, most existing datacentres in India, which are only capable of supporting low- to medium-power density racks, will have to upgrade or build new facilities to accommodate higher-power densities required by AI systems. “They need a plan that meets both traditional needs and AI requirements, and a modular approach can serve this purpose,” he said.

[Existing datacentres in India] need a plan that meets both traditional needs and AI requirements, and a modular approach can serve this purpose

Pankaj Singh, head of datacentre and telecom business solutions at Delta Electronics, said AI workloads typically require high-density racks (20kW to 50kW per rack), and modular datacentres are designed to support such densities through integrated, scalable power and cooling.

Besides AI, other key trends that are fuelling the market opportunity for modular datacentre systems in India include the ongoing 5G roll-out and the push for local data storage under the Data Protection Bill. “Modular systems offer up to 40% faster deployment than traditional builds, standardised quality, reduced on-site labour and ease of expansion,” said Pankaj Singh.

There’s also the growing need for sustainability, driven by costs and compliance pressures. “Modular solutions address AI’s appetite for power by enabling scalable, energy-efficient and localised deployment,” said Mahapatra. “These systems optimise resources by focusing computational power on specific tasks, integrating edge computing and employing lightweight AI models, reducing reliance on energy-intensive centralised infrastructures.”

Growing traction

In India, multinational companies like Schneider Electric, Huawei, Dell and Vertiv are already investing in modular datacentre systems, joining local companies such as VueNow, which is making strides with modular, containerised edge computing setups.

“Nxtra by Airtel has also reportedly implemented modular setups in a few tier-two locations, and Yotta is keenly exploring modular strategies for small datacentres,” said Srihari Srinivasan, director and lead for datacentre services at Savills India. “Digital Connexion is also an early adopter in its Chennai campus, using modular techniques for flexible scaling.”

Real-world applications are growing, too. Agritech firm Cropin uses edge computing and modular AI to optimise agricultural data processing. In healthcare, Niramai leverages modular systems to perform localised computations for its AI-powered cancer screening tools.

In retail, Bassi said companies are using micro datacentres for real-time stock management and point-of-sale analytics, while in manufacturing, modular units are enabling local data processing for production monitoring and warehouse automation. “Smart cities are also beginning to deploy modular nodes to manage traffic and surveillance data,” he added.

Even large campuses will be designed in a modular fashion, Dietrich predicted. “A 500MW datacentre will likely be made up of five 100MW buildings or ten 50MW buildings. The infrastructure can be modularised to support AI training, AI inference and standard compute, as each has different power and heat density profiles,” he said. “Building this way also eases the upgrade path, as you can upgrade one section by ‘plugging in’ new systems while keeping other areas running.”

Despite the growing momentum of modular datacentres, broader industry challenges remain. “In neighbouring regions like the Middle East and Southeast Asia, some service providers have been challenged by over-investments in both datacentres and graphics processing units while waiting for customer demand to materialise,” warned Naresh Singh. “India must avoid such situations by making the right investments to match the real potential of the market.”

Furthermore, modular approaches may not be practical for upgrading legacy datacentre facilities, and the initial capital outlay may be too high for mid-sized companies, said Purvi Shah, a CIO from the construction industry. “In highly regulated sectors, modular units may also raise concerns around compliance if not planned well.”