20 Million Litres a Day? Fact-Checking the Water Claim at RBI’s New Odisha Data Centre

What was alleged—and what the number means

On 23 October 2026, before the inauguration, Odisha Congress leader Panchanan Kanungo said the RBI facility would require about 2 crore litres of fresh water each day and asserted that the supply would come from groundwater in Khordha district. PTI reported the remarks from a press conference in Bhubaneswar. The report also quotes Kanungo saying that cooling machinery at a single data centre could require 20 million litres a day. ^[4]

The PTI report does not cite a project water assessment, RBI filing, engineering design, meter reading or named technical study behind the figure. It does not define whether “requirement” means a design maximum, typical daily withdrawal, water consumed through evaporation, or a campus-wide figure. The groundwater source is also part of Kanungo’s allegation; it is not confirmed by the RBI or the state in the official material reviewed here.

The scope is unclear as well. A PMO pre-event note described four data centres at the Bhubaneswar complex, but did not publish their combined IT load or explain how the political figure maps to one unit, the whole campus, or the associated training institute. ^[2] The allegation should therefore be reported as an attributed claim—not converted into an established RBI design figure.

What official records confirm

The RBI’s 25 October press release says Prime Minister Narendra Modi inaugurated its “newly established” Data Centre and Cybersecurity Training Institute in Bhubaneswar. It records the Finance Minister’s statement that the new data centre will house mission-critical RBI systems including RTGS and NEFT. It does not report a water figure or a water source. ^[1]

The RBI’s March 2023 bulletin records that the foundation stone was laid on 22 March 2023 for a Greenfield Data Centre and Enterprise Computing and Cybersecurity Training Institute in Bhubaneswar. It gives a combined area of 18.55 acres. ^[3] Public reporting places the campus at Info Valley-II in Jatni, Khordha district; the RBI and PMO announcements reviewed describe it as being in Bhubaneswar, Odisha. ^[5]

That is enough to establish the project’s public milestone and stated purpose. It is not enough to establish actual water use. The official release says systems “will” be housed there; the sources reviewed do not give a date for workload migration or a post-opening meter reading. Inauguration is not the same as proof that every system is already operating at full load.

Cooling water is not fire-suppression water

Data-centre water use cannot be inferred from the label “data centre” alone. Cooling architecture, IT load and utilisation, operating conditions and local weather all matter. A 2024 Lawrence Berkeley National Laboratory (LBNL) report models multiple cooling configurations and describes an energy–water trade-off: air-cooled chillers can use no on-site cooling water in the modeled scenarios but may use more energy, while evaporative water-cooled systems can consume more water. A 2025 peer-reviewed LBNL study likewise identifies cooling technology, climate, server efficiency, workload and electricity as important determinants. ^{[6] [7]}

Those studies explain why the design matters; they are not estimates for the RBI campus. Applying a U.S. fleet average or a generic litres-per-megawatt assumption to Jatni would require project-specific IT capacity, cooling design, operating profile and a clearly defined water metric. None of those inputs was found in the official sources reviewed.

Fire protection answers a different question. A 25 October Bhaskar English report says the facility’s fire-detection and suppression systems use gas rather than water. That is secondary reporting about fire suppression—not an RBI-published cooling specification. A gas-based fire system does not establish that cooling is waterless, and it neither proves nor disproves the 20-million-litre claim. ^[8]

For a useful comparison, a disclosure should also distinguish withdrawal (water entering the site), consumption (water not returned to the same water system, often through evaporation) and effluent returned or reused. A single undifferentiated “water use” number can obscure those differences.

What remains unlocated in public sources reviewed

As of 1 November 2026, the official inauguration material and project background reviewed do not provide a public, site-specific answer to the following:

The campus also includes a Cybersecurity Training Institute, so a complete figure should say whether it covers the data-centre units alone or the whole campus. The official RBI bulletin gives the combined land area, but not a water split between the two functions. ^[3]

Because groundwater is one of the allegations, the strongest evidence would be a project-linked approval or allocation naming the source and permitted quantity, paired with metered abstraction and monitoring data. If the source is municipal, surface or recycled water, the supplying agency and allocation should be stated instead. We did not obtain a direct response from RBI or the Odisha government for this article; this is a review of public sources, not an official disclosure or legal determination.

The ₹7,200-crore figure is a hypothetical valuation—not a water bill

The PTI report says Kanungo valued the alleged daily volume at ₹10 per litre and arrived at ₹7,200 crore a year. ^[4] That figure depends entirely on two unverified inputs: the alleged 20 million litres per day and the assumed ₹10-per-litre price. The report does not identify an RBI tariff, supply contract or bill at that rate.

This arithmetic check does not validate the water-volume allegation. It shows that ₹7,200 crore is a scenario calculation using a 360-day year, not a reported project cost or confirmed annual water charge. A real bill would depend on the source, tariff, metered withdrawal, discharge and any reuse arrangements.

What would settle the question?

A short, project-specific disclosure from RBI or the relevant Odisha agencies could replace speculation with a testable water account. It should state:

1. Scope and demand: design and expected operating demand in KLD/MLD, peak versus average, seasonal profile, and whether the number covers each data-centre unit or the whole campus.
2. Source and authority: groundwater, municipal, surface or treated/recycled supply; the issuing or supplying authority; and the applicable allocation or permission reference and daily cap, where relevant.
3. Cooling and efficiency: cooling/heat-rejection type, IT load and a defined site-WUE metric with its reporting period.
4. Water balance: intake, reused water, evaporative losses, cooling-tower blowdown, sewage/effluent treatment and discharge or reuse destination.
5. Actual operations: metered monthly withdrawals and consumption once the relevant systems are in service, with a clear distinction between design estimates and observed use.

If the 20 MLD allegation is accurate, a source, method and water balance should make it possible to test. If it is not, the same disclosures would allow the claim to be corrected. Until then, the fairest conclusion is not “the RBI centre uses 20 million litres a day” or “it uses no water”; it is that the figure and source remain unverified in the public material reviewed.