Why India’s $30 Billion Data Centre Boom Demands a New Kind of Leadership
Topic: Technology
Format: Article
Published Date: October 2026
AI for Leaders: Strategy to Action Programme
India’s data-centre market is entering a period of extraordinary expansion, with capacity projected to rise fivefold to 8GW by 2030 and capital expenditure expected to exceed $30 billion. Global technology companies are committing billions to build AI and cloud infrastructure in the country, while Indian players are making large bets of their own. But the scale of investment is only part of the story. Data centres are long-lived assets being built for a technology that is evolving rapidly, while their economics increasingly depend on power, connectivity, cooling, regulation and local infrastructure. For leaders, the challenge is to make large, long-term commitments while preserving the flexibility to respond as technology, demand and the operating environment change.
Data centres, once largely treated as back-end technology infrastructure, are becoming some of the country’s largest and most consequential capital projects. Jefferies estimates that India’s data-centre capacity could rise fivefold to 8GW by 2030, requiring around $30 billion in capital expenditure.
The capital is already following the opportunity. Amazon announced in June that it would invest a further $13 billion in India by 2030 for data centres, taking its total investment in the country to $48 billion between 2026 and 2030. Microsoft has committed about $20.5 billion to expanding its operations in India and has opened its largest data-centre hub in Hyderabad. Google has committed $15 billion over five years to an AI hub in Visakhapatnam, Andhra Pradesh. Yotta plans to invest more than $2 billion in an AI supercluster using more than 20,000 Nvidia Blackwell Ultra GPUs.
But while the capital exists, the leadership challenge lies in how to make long-term infrastructure decisions for a technology that is evolving rapidly.
India is committing long-lived capital to infrastructure that has to support a technology cycle moving much faster than the infrastructure itself. Data-centre facilities can last for decades, while the technology they house is changing rapidly; their economics also depend on electricity, connectivity, cooling, land and regulation. Leaders therefore have to make decisions that remain viable even when the assumptions behind them change.
The infrastructure has to outlast the technology
AI is changing the economics of the data centre itself. Jefferies estimates that AI servers consume five to six times as much power as traditional servers and require advanced liquid-cooling systems. As computing density rises, the physical design of a facility becomes increasingly important to its ability to support future workloads. Yotta’s planned supercluster in Greater Noida illustrates the scale of the shift: its 20,736 liquid-cooled GPUs are being deployed initially at a 60MW facility designed to scale to 250MW.
The challenge is that the technology inside the facility will evolve long before the building reaches the end of its useful life. A facility designed around today’s computing requirements may need different power densities, cooling systems or network architecture as successive generations of AI hardware arrive. Building too narrowly around current technology can leave expensive infrastructure poorly suited to future workloads, while building too much capacity too early creates a different risk if demand develops more slowly or in a different direction.
The quality of the initial decision is therefore unusually important. Leaders need to distinguish what they know from what they are betting on, and how much flexibility they can preserve when making a difficult-to-reverse commitment.
The bigger constraint may sit outside the company
Connectivity, customers, land and cost remain important, but access to reliable power is increasingly part of the equation. India’s peak power demand reached 270.8GW in May 2026, a record high, and the government expects demand to approach 300GW next year. Manohar Lal, Minister of Power in India, has linked the increase to the expansion of data centres, AI and electric vehicles, while calling for greater investment in grid infrastructure and energy storage. For data-centre operators, the issue is therefore not simply the price of electricity, but whether the surrounding power system can support the scale and reliability that AI workloads require.
Water presents another local consideration, although the implications vary with the cooling technology and location. Google’s $15 billion project in Visakhapatnam has faced concerns from environmental groups and local activists over pressure on water resources and the nearby Kambalakonda Wildlife Sanctuary. Google says it will use advanced air-cooling technology to minimise the project’s environmental impact, while the Andhra Pradesh government has defended the development. The broader business implication is that resource availability and local conditions can become material variables in a major investment decision—even when they sit outside the company’s direct control.
India is building an AI infrastructure ecosystem
The government’s policy choices suggest that it sees data centres as part of a larger economic opportunity. The 2026 Budget proposed a tax holiday until 2047 for eligible foreign companies that provide cloud services globally while procuring data-centre services in India. The measure is intended to attract global cloud operations and give investors greater certainty around the economics of building digital infrastructure in the country.
If global cloud providers use India as a base from which to serve customers elsewhere, data centres could become part of the country’s digital-services infrastructure and export economy. At the same time, companies investing in the sector remain closely tied to global technology supply chains. Yotta’s Blackwell deployment, for instance, combines Indian infrastructure and capital with Nvidia’s advanced processors and technology.
The leadership question is therefore not whether companies can control every part of the value chain, but which capabilities they should control, which can be sourced from partners and where dependence creates a vulnerability worth reducing.
The leadership problem is balancing competing time horizons
The difficulty lies in the interaction between these decisions. A choice about AI hardware affects power and cooling requirements; a decision about location affects access to energy, connectivity and regulation; a commitment to capacity determines how much capital remains available for the next generation of technology. The time horizons are also different: data centres can operate for decades, AI hardware evolves rapidly, customer demand can change within a few years and regulation can shift during the life of a project. A five- or ten-year forecast may therefore be an incomplete guide to a decision that could shape a business for much longer.
India’s data-centre boom is likely to create substantial value as cloud adoption and AI accelerate demand for computing. The companies best placed to capture that value will need to make large, long-term commitments while retaining the flexibility to adapt as technology, demand and the operating environment change. The leadership challenge is knowing where to commit for the long term and where to preserve room to change course.