India, Feb. 7 -- The most recent budget offers a twenty-year tax holiday to foreign companies using data centres in India. At first glance, it sounds like a straightforward growth play. Bring global technology infrastructure into India. Create jobs. Signal ambition. Become indispensable to the world's digital future. But this logic begins to unravel once data in the AI world is understood not as "software", but infrastructure. It is not merely a warehouse of computers, but an industrial facility designed to convert electricity and water into intelligence. Unlike traditional data centres, which mainly store and retrieve information, AI data centres continuously train and run models by performing trillions of calculations every second. This process generates enormous heat. Preventing failure depends on relentless cooling. That physical reality changes everything. AI-optimised data centres consume several times more electricity than conventional ones, and far more water for cooling-especially in hot, coastal climates like India's. Water is not used because it is cheap, but because at scale it is the fastest way to extract heat from machines that cannot be shut down. This is where the risk of compute-colonialism begins to take shape. This is a situation where a country hosts the physical machinery of AI-the land, the power, the water, and the data centres-but does not own the intelligence those machines produce. The costs are local and immediate. But the benefits remain elsewhere. Supporters of the tax holiday argue that these facilities will be powered by renewable energy through captive solar or wind plants. That may be technically true. But electricity and water are not abstract credits. They are finite, local resources. When a hyperscale data centre locks up the best solar land or receives priority grid access, that power is no longer available to nearby factories, housing projects, or small businesses. When it draws groundwater or surface water for cooling, that water is no longer available for agriculture or urban use. The resource does not disappear; it is displaced. This displacement produces a second-order effect that rarely enters policy conversations. Large global firms are able to clean up their carbon balance sheets using Indian sun and wind, while local industry is pushed onto a more expensive and often dirtier grid. The global ledger turns green. The local economy pays a green premium. If foreign data-centre operators receive a twenty-year tax holiday while Indian firms do not, domestic players cannot compete on equal terms. This is not market competition; it is policy-engineered asymmetry. As Ajai Chowdhry, author and co-founder of HCL points out, "the same deal must be offered to domestic data centre companies as well as they won't be able to compete" . Then comes the question the policy does not adequately answer: what problem is the tax holiday actually solving? Industry voices who have been closely watching argue that taxes are not the binding constraint. Power availability is. Grid-connection timelines are. Water permissions are. Countries currently attracting large AI infrastructure investments such as Malaysia and parts of the Gulf, for example are not winning because they are cheaper. They are winning because they are faster. Speed-to-power matters more than generosity-to-capital. Viewed through this lens, India risks offering a discount on a product it cannot yet reliably deliver. If grid access takes years, a tax holiday becomes less a strategic lever and more a signal of desperation. The state gives up future revenue upfront, while delays, local shortages, and environmental backlash are absorbed domestically. This is compute-colonialism in its most practical form. A facility that can be remotely restricted or repurposed is not national capability; it is rented infrastructure. While this is one way to look at it, the Bengaluru-based Debajit Ghosh, who is now neck deep into studying AI, environment and social governance makes the case that "everything comes to trade-offs and not binaries." His larger point is that data centres can build capacity, jobs, and strategic capability. But clear guardrails are needed as well so that India does not subsidise the carbon and resource footprint of global Big Tech. This is why the tax holiday must come with conditions that reflect the scale of public subsidy involved: parity for domestic firms, access to computation for Indian research and public services, and a credible path to local ownership of the intelligence trained on Indian resources....