Land might well hold the key to building a business in data centres. While other factors such as connectivity, access to power are of course critical factors, everything starts with land. This means that landowners have emerged (even if they have not realized it) as key players in designing viable business models in the business of data centres. This is true for governments as well as they are among the biggest landowners. The current trend of using land for residential/commercial construction might shift towards data centres.
Whatever the sweep and scale of AI in the future, one thing is certain: data centres are a major business opportunity, which also means investments in land. Those who are holding historically-owned land parcels are sitting on a gold mine. The Times of India, October 31, 2025, page 15, reports of TCS’s $6.5 billion proposed investment in AI data centre signalling ‘a push to build capabilities and prevent technological colonization. It also lists some recent large deals.
The McDonald’s Model
There is this famous slogan that is associated with McDonald’s – location, location, location as the three factors most important in their business. In data centres, it is going to be size, size, size as the three most vital factors. Given their need for premium locations, McDonald’s embraced the franchise cum JV model, the landowner becoming a venture partner.
That is the most likely model even in the burgeoning data centres business. Land cost is rising in India and would gobble up a substantial part of the total investments, thus making it a no-brainer to usher in the landowner as a venture partner. ‘Land prices surge across India’s top micro-markets’ was the headline in an ET story identifying urbanization and infrastructure as the leading factors causing the rise in land cost, especially in the satellite towns.
The Economic Times quoted a Deloitte report which said that ‘India may need additional 50 mn sqft real estate for data centres by 2030’ identifying in its report “six pillars for building a world-class AI ecosystem: real estate, power and utilities, connectivity, compute infrastructure, talent, and policy framework”.
From construction to data centres
Most even reasonably keen observers would have noticed the entry of several non-construction companies into the residential-commercial construction sectors. Among the lates to enter is Sundaram Housing, of the Sundaram group in Tamil Nadu. Raymond, Bombay Dyeing are now established names in construction.
I will not be surprised if Indian companies, especially those who have been around for decades and who own vast amounts of land, enter into some model of joint venture with companies who aim to build a business in data centres. This approach will avoid the cumbersome and politically-influenced process of acquiring land.
This list will include several state and central public sector undertakings too, offering a credible alternative to partial or total disinvestment to raise funds. Imagine what the Indian Railways can do as the biggest landowner in the country or the Waqf Boards who rank next to the Indian Railways.
The Public Accounts Committee of the Government of India has recently published a report on this titled “Development of Railways Land for Commercial use by Rail Land Development Authority”. According to a Press Release, Indian Railways hold about 4.90 lakh hectares of land as of March 31, 2024, furnishing details of holding by several Zonal Railways. The Honourable Minister for Railways, Ashwini Vaishnav, told ET Infra that “1% of railway land under commercial use, 0.22% under encroachment”. The potential to turn idle land into capital is just staring at the Railways.
Land for data centres
According to TechTarget, “On average, a full-scale data center occupies 40 acres of land. According to Statista, as of March 2025, more than 10,000 data centers exist globally. If the 10,000 existing data centers occupied 40 acres each, the estimated total land use would be 400,000 acres, or about 528,000 football fields.
As of 2024, the average data center lot size has changed. According to Cushman & Wakefield, the average data center land transaction is 224 acres, a 144% increase since 2022. AI data centers typically require a minimum of 200 acres to accommodate the physical footprint needed to support AI workloads. The total land requirements for modern hyperscalers often exceed 200 to 500 acres. Some companies purchase 1,000 acres or more for future growth and expansion of their data center campuses.
Investment in power plants
Linked to the demand for data centres is the demand for power plants. Given the size of power-hungry data centres, renewables may not be the first choice. Consider this fact: “Renewable energy use can decrease carbon emissions, but methods like solar cannot sustainably power large data centers. According to Simple Thread, it would take 1,446 acres of solar panels to power a 100 MW hyperscale data center”. Clearly, while a boon from an environmental standpoint, it is financially inefficient use of land.
Let us look at how much land is required for setting up a thermal power plant.
According to a study conducted by the Central Electricity Authority, New Delhi, of the Ministry of Power on land requirements for thermal power plants, “The land requirement for the main plant is found to vary from 29.7 acres for 2×500 MW plant to 108.8 acres for 6×660 MW. The main plant area requirement is a function of the capacity of the plant, unit size and also the quality of coal used in the station. In the first four cases it is presumed that coal with 40% ash will be used and hence the main plant area varies proportionately with the plant capacity”.
The role of politics
Land everywhere is an intensely political question. Perhaps, more so in India where politics has dominated the economic sphere right since Independence. The prospect of substantial gains to be made will lure politicians of every hue and companies must be on guard.
Should governments eye the business of data centres, banking on land parcels owned by them, containing political interference will be key. Although it can never be wished away, it is possible to contain it through designing legal/organizational structures that will help focus on the business. For governments to set up businesses with limited interference from politicians, historically, statutory corporations have been the preferred model. Perhaps, it time to think beyond, for the simple purpose of keeping politicians at arm’s length. That may well be the toughest task.
