NITI Aayog's Frontier Tech Hub released a Data Centre Evaluation Framework on 14 October 2026, announced the following day. On the face of it this is a document about digital infrastructure. Read the five pillars it is built on and a different subject appears: electricity, water, land and waste. Two of the five pillars are about resources that have nothing to do with computing and everything to do with whether a district can support one.
That reframing is the whole value of the document, and it is the examinable idea.
The numbers behind the urgency
India's installed data-centre capacity has grown from about 375 MW in 2020 to roughly 1.5 GW now — a fourfold expansion in five years. The Minister of State for Electronics and Information Technology gave the figure of over 1,500 MW in 2025, against 375 MW in 2020, in a parliamentary reply in March 2026; a later official factsheet put installed capacity at about 1.57 GW. Projections for 2030 vary by source, with official estimates near 8 GW and some industry research closer to 6.5 GW.
Two cautions belong with those numbers, and stating them is the mark of a careful candidate. The 2020 baseline came from a commercial estimate of colocation capacity, while the later figures are government estimates of installed capacity — so the comparison is indicative rather than exact. And capacity is measured in megawatts, not square feet or server counts, which is itself the clue: the industry measures itself by the power it draws.
Capacity is also heavily concentrated. Mumbai and Navi Mumbai account for roughly 790 MW, with Chennai near 305 MW and Bengaluru near 182 MW. Coastal cities dominate because that is where submarine cables land. Nothing in that pattern tracks where power is cheapest or water is abundant.
The policy push is real and recent. The Union Budget for 2026-27 proposed a tax holiday until 2047 for foreign companies supplying cloud services to customers outside India using Indian data-centre infrastructure — with conditions, including that sales to Indian customers route through an Indian reseller entity and be taxed domestically, and a safe harbour where the Indian data-centre provider is a related entity. The intent is to make India a base for global AI and cloud workloads.
So demand is rising, central incentives are generous, and States are competing. The Framework's premise is that this is exactly when a bad siting decision becomes expensive.
The five pillars
The Framework is organised around five pillars: Power and Energy; Water and Cooling; Siting and Environmental Sustainability; Economic and Strategic Value; and Policy, Governance and Implementation.
Power and Energy. The recommendations ask States to assess grid readiness and net grid draw, and to promote renewable energy with storage-backed power. The term that repays attention is net grid draw — not the sanctioned load on paper, but what the facility will actually pull from the grid after captive and renewable supply is counted. A data centre is a flat, round-the-clock load. It does not peak in the evening and fall away at night like a residential feeder, which makes it both attractive to a distribution utility and demanding of firm supply. Solar alone cannot serve it, which is why storage appears in the same sentence.
Water and Cooling. The Framework asks for assessment of basin-level water stress and the impact on public supply, and recommends rainwater harvesting and on-site water treatment and storage. Cooling is where a data centre's water goes, and the quantities are not trivial. Asking the question at the level of the river basin rather than the municipal connection is the methodological point: a facility can have a lawful water connection and still be drawing from a system that is already over-committed.
Siting and Environmental Sustainability. Alongside siting, this covers lifecycle plans for e-waste, batteries and other hazardous waste — a reminder that uninterruptible power supply at this scale means a large and periodically replaced battery estate.
Economic and Strategic Value. Here the Framework makes its sharpest recommendation: distinguish colocation from active compute services, identify suitable data-centre zones, and link State incentives to measurable outcomes. Colocation is, in essence, a landlord business — floor space, power and cooling rented to clients who bring their own equipment. Active compute is the provision of computing capability itself. A State that gives land, power tariffs and capital subsidies without making that distinction may be subsidising a warehouse while believing it has bought an AI industry. Attaching incentives to outcomes, and monitoring project commitments, is the proposed discipline.
Policy, Governance and Implementation. Time-bound approvals, clear accountability and monitoring — the familiar administrative layer, and the one most dependent on State capacity.
What a framework can and cannot do
NITI Aayog's CEO described the purpose as moving from investment attraction to investment readiness, and was explicit that the Framework is meant to work without creating an additional layer of regulation. That phrasing is not modesty. It is constitutional arithmetic.
Land, water supply and electricity distribution are predominantly State responsibilities. State industrial policies decide incentives; State transmission and distribution utilities decide connectivity; State pollution control boards and local bodies decide consent and siting. NITI Aayog is an advisory body and not a regulator. It cannot mandate an evaluation method; it can supply one that is good enough to be adopted.
This is the same constraint that shapes most of India's recent infrastructure governance. The PM GatiShakti National Master Plan worked by putting thousands of data layers on one map so that ministries and States could see each other's plans — persuasion through visibility rather than direction. The Integrated Transport and Logistics Authority took the same route for large projects. A State is far more likely to adopt a framework that improves its own decisions than to accept one that limits them.
There is also a genuine economic argument for the States, and the Framework makes it: a project sited where power is not firm or water is not available ends up delayed, renegotiated or stranded. Early identification of constraints is cheaper for the State than a stalled project on which incentives have already been committed.
The part that will be argued about
Prof. Abhay Karandikar, Member of NITI Aayog, pressed a point at the release that goes beyond siting. He argued for strengthening domestic R&D, industry-academia collaboration and Indian participation across the data-centre technology value chain, naming advanced cooling, energy efficiency and resource-efficient computing as areas where India could develop its own solutions.
That is the harder ambition. Hosting compute and building the technology that compute runs on are different industrial outcomes, and the first does not automatically produce the second. A data centre is a large construction and electrical project; the chips inside it, and increasingly the cooling systems around them, are high-value imports. The Framework's "Make in India across IT and non-IT infrastructure" recommendation points at the non-IT portion — power equipment, cooling, construction systems — where domestic capability is realistic in the near term.
The honest framing for an answer is that India is currently buying a location advantage — cables, power, land and a tax holiday — and that converting it into a technology advantage requires something the Framework can encourage but not deliver.
Why energy makes this a strategic question
One further link is worth drawing. A fourfold rise in data-centre capacity, with projections of several more times that by 2030, is a new and inflexible demand on a power system that is simultaneously decarbonising. Firm round-the-clock supply from renewable sources means storage, and storage means batteries, critical minerals and a supply chain India does not control. The country's energy dialogue with major oil producers is about one half of that security question; where the electrons for compute come from is the other.
That is the sentence an examiner will reward: data centres turn a technology policy question into an energy and water policy question, and the Framework is an admission that States were deciding the first without examining the second.
🔑 Revision block
- What: the Data Centre Evaluation Framework, released by NITI Aayog's Frontier Tech Hub on 14 October 2026, developed with inputs from MeitY, the Ministry of Power and DPIIT. It is a decision-support tool for States, not a regulatory requirement.
- Five pillars: Power and Energy; Water and Cooling; Siting and Environmental Sustainability; Economic and Strategic Value; Policy, Governance and Implementation.
- Capacity: about 375 MW in 2020 to roughly 1.5 GW now (over 1,500 MW in 2025 per a March 2026 parliamentary reply; about 1.57 GW in a later factsheet). Projections for 2030 range from about 6.5 GW to 8 GW.
- Concentration: Mumbai and Navi Mumbai about 790 MW, Chennai about 305 MW, Bengaluru about 182 MW — driven by submarine-cable landings, not by power or water availability.
- Key recommendations: assess grid readiness and net grid draw; storage-backed renewable power; basin-level water stress and public-supply impact; rainwater harvesting and on-site water treatment; lifecycle plans for e-waste and batteries; distinguish colocation from active compute; link incentives to measurable outcomes; time-bound approvals.
- The fiscal backdrop: the Union Budget 2026-27 proposed a tax holiday until 2047 for foreign cloud providers serving overseas customers from Indian data centres, with domestic sales routed through an Indian reseller and taxed in India.
- Why advisory: land, water and electricity distribution are State subjects; NITI Aayog advises rather than regulates.
🎯 Practice MCQs
Q1. The Data Centre Evaluation Framework released in October 2026 was issued by: (a) NITI Aayog's Frontier Tech Hub (b) the Central Electricity Authority (c) the Telecom Regulatory Authority of India (d) the Ministry of Environment, Forest and Climate Change
→ (a) It was released by NITI Aayog's Frontier Tech Hub, with inputs from MeitY, the Ministry of Power and DPIIT.
Q2. Which of the following is NOT one of the Framework's five pillars? (a) Power and Energy (b) Water and Cooling (c) Taxation and Customs (d) Economic and Strategic Value
→ (c) The five pillars are Power and Energy; Water and Cooling; Siting and Environmental Sustainability; Economic and Strategic Value; and Policy, Governance and Implementation.
Q3. India's data-centre capacity is reported to have grown from about 375 MW in 2020 to approximately: (a) 700 MW (b) 1.5 GW (c) 4 GW (d) 8 GW
→ (b) Roughly 1.5 GW, a fourfold increase. The 8 GW figure is a projection for 2030, not current capacity.
Q4. The term "net grid draw" in the Framework refers to: (a) the power a facility actually draws from the grid after captive and renewable supply is accounted for (b) the total sanctioned load approved by the distribution utility (c) the transmission losses attributable to a data centre (d) the peak demand of the district in which a data centre is located
→ (a) It is the real demand placed on the grid, not the sanctioned load on paper.
Q5. The Framework recommends assessing water availability at the level of: (a) the municipal water connection (b) the industrial estate (c) the river basin (d) the individual borewell
→ (c) Basin-level water stress is the recommended unit of assessment, along with the impact on public supply.
Q6. Why does the Framework ask States to distinguish colocation from active compute services? (a) Because colocation facilities are exempt from environmental clearance (b) Because active compute facilities draw no grid power (c) Because colocation is regulated by the Centre and compute by the States (d) Because incentives may otherwise subsidise rented floor space rather than computing capability
→ (d) Colocation rents space, power and cooling; active compute supplies computing capability. Incentives that do not distinguish them may buy the former while intending the latter.
Q7. Data-centre capacity in India is conventionally measured in: (a) megawatts of power capacity (b) square metres of floor area (c) number of installed servers (d) terabytes of storage
→ (a) Capacity is measured by power, which is itself a clue to the resource intensity of the sector.
Q8. NITI Aayog's role in relation to the Framework is best described as: (a) a statutory regulator issuing binding siting norms (b) a licensing authority for data-centre projects (c) an advisory body supplying a decision-support tool to States (d) a financing agency funding State data-centre parks
→ (c) NITI Aayog is advisory. The Framework is explicitly designed not to create an additional layer of regulation.
Q9. The Union Budget 2026-27 proposal concerning data centres offered: (a) a production-linked incentive for server manufacturing (b) a tax holiday until 2047 for foreign cloud providers serving overseas customers from Indian data centres (c) customs duty exemption on all imported cooling equipment (d) a central subsidy on electricity tariffs for data centres
→ (b) The tax holiday is conditional, with sales to Indian customers routed through an Indian reseller and taxed domestically.
Q10. Which pair of subjects explains why NITI Aayog can only advise States on data-centre siting? (a) Defence and foreign affairs (b) Banking and insurance (c) Posts and telegraphs (d) Land and electricity distribution
→ (d) Land, water supply and electricity distribution fall predominantly to the States, so siting decisions and incentives are State decisions.
📋 How this gets asked (PYQ pattern)
This item sits at the junction of three CDS staples — economy, environment and federalism — which is why it is better value than its technology label suggests.
The first form is the institution question. NITI Aayog appears regularly, and the reliable discriminator is its character: it is an advisory body, successor to the Planning Commission, with no statutory or regulatory power. A statement that describes NITI Aayog issuing binding norms is false, and the paper writes exactly that statement.
The second is the number-with-unit question. Capacity measured in megawatts, growth from 375 MW to about 1.5 GW, projections to 2030. Option sets here usually mix the current figure with the projection, so hold both and keep them apart.
The third is the concept-definition question, which this Framework supplies in unusual abundance: net grid draw, basin-level water stress, colocation against active compute. Definitional items of this kind are cheap to set and hard to guess, which is why they get set.
A fourth, slower-burning form is the federalism application. Land, water and electricity distribution are State responsibilities, which is why a central body produces a framework rather than a rule. Expect this as a reasoning question — why is this advisory rather than mandatory — rather than as a direct constitutional item.
For a written answer, the argument worth making is that a data centre is an electricity and water project carrying a technology label, and that India's current capacity is concentrated where submarine cables land rather than where power is firm or water is abundant. Add the distinction between buying a location advantage and building a technology advantage, and the answer is assessing a policy rather than summarising a press release.
Preparing for CDS/OTA? Economy questions increasingly arrive dressed as technology stories — read them for the resource and the institution. Build the base with our CDS/OTA study material and the economy section, follow the daily CDS current affairs, and prepare with our faculty in the upcoming Cavalier courses in Delhi.
✍️ Written by Hitendra Deswal — Polity & Economy faculty at The Cavalier. Reviewed by the Cavalier Faculty Desk.