AirTrunk India pipeline — 600 MW across five sitesAirTrunk, via Lumina CloudInfra · Mumbai, Chennai and Hyderabad · 601 MW across five location cards · all stage unverified

The 600 MW AirTrunk acquired with Lumina CloudInfra in April 2026 breaks down into five location cards — BOM1 at 62 MW, BOM2 at 120, BOM3 at 48, MAA1 at 128 and HYD1 at 243 — which sum to 601 MW. That is the cleanest multi-site reconciliation in this registry: six hundred stated, six hundred and one in components, a difference of one megawatt. And it produces the most uncomfortable result in this file. The cards are additive; the aggregate above them is not. So AirTrunk enters this registry as its largest counted operator at 601 MW — 24.8% of the entire 2,420.4 MW counted market — on the strength of five buildings whose operating stage no card states. Every one carries an at least qualifier. Every one is stage unverified. If this registry counted only verified operational capacity, AirTrunk would contribute zero and the national total would fall to 1,819.4 MW.

1At a glance

2601 against 600

Multi-site portfolios in this registry almost never reconcile. Palava’s three cards sum to 171 MW inside a 400 MW design capacity and cannot be resolved. STT Mumbai publishes 133 MW against six non-additive buildings and nothing closes. This one closes to one megawatt.

One megawatt in six hundred is rounding, and it is a good sign: the five cards and the portfolio total were produced from the same underlying schedule. That is not true of most of the multi-site aggregates in this file.

The substation column is the second useful thing here. Every site except Chandivali publishes a high-voltage connection — 440/220 kV in Hyderabad, 230 kV in Chennai, 220 kV at both Mumbai sites. Chandivali’s 48 MW runs off a 33 kV feed, which is an order of magnitude smaller and tells you it is a dense urban infill site rather than a campus.

3The largest counted operator in the registry

Now the part that matters more than the reconciliation, and it is a problem with this registry as much as with the company.

Read the last two rows together. A quarter of this registry’s counted capacity comes from five cards that say nothing about whether the buildings exist. The logic is internally consistent — children of a non-additive aggregate count individually, which is the same rule that makes NTT’s 181 MW countable — but the effect is that the registry’s total is far less defensible than its NTT component.

The honest statement, and the one this page puts in public: 2,420.4 MW is what this registry counts. 1,819.4 MW is what it can point at. The 601 MW gap is AirTrunk’s pipeline, and it may be entirely real — blackstone-backed operators do not usually announce sites they have not bought.

What is known independently: Lumina broke ground on a 60 MW project in Airoli in May 2023, with the first phase due live in 2025, and no launch announcement has been made. It reportedly bought a Chandivali plot for a 60 MW facility, and acquired a 16-acre parcel in Ambattur. So at least three of the five have land and one has construction history.

4$5 billion for 600 MW

The pipeline carries a cost figure of its own, and it does not match the programme’s rate.

Eight point three million dollars a megawatt against six is a large gap, and the likely explanation is scope. Five billion dollars is described as future development potential, which is a valuation of the land, permits and pipeline rather than a construction budget. The programme’s six million per megawatt is what it costs to build.

For pricing purposes the registry uses the programme rate, because it is a construction cost. But the five-billion valuation is the one that matters to anyone valuing the pipeline itself: it puts US$8.3 million on every megawatt of acquired pipeline, before a building exists.

Compare that to the ₹51.4 crore (US$5.8 million) per megawatt that ICRA reports for NTT’s Airoli phase one — land, construction and finance. AirTrunk’s pipeline is valued 43% above NTT’s construction cost, which is the premium a buyer pays for land and permits in a market where both are the binding constraint.

5Three markets or five

The portfolio’s geographic scope is published inconsistently, and the difference matters for anyone using it to size regional demand.

The location cards settle it for now: five sites in three cities. Pune and Delhi NCR appear in the corporate narrative and do not appear as cards.

That is worth recording because a market report that allocates 600 MW across five cities will overstate Pune and Delhi NCR by whatever share it invents. The published allocation is 230 MW in Mumbai, 128 in Chennai, 243 in Hyderabad.

And the Mumbai concentration is the commercial point. Three of the five sites, 230 of the 601 MW, in one metropolitan region — in Navi Mumbai, Mahape and Chandivali, which are the three submarkets where NTT, CtrlS, PDG and Lumina are already stacked against each other.

6Procurement

7Limits and open questions

• All five cards are stage unverified. No card states whether the facility is operating, under construction or planned.
• All five carry an at-least qualifier, which means each figure is a floor.
• The cards are facility capacity, not IT load. They cannot be compared with NTT’s IT-load figures without a ratio, and no PUE-adjusted IT figure is published per site.
• The 601 MW counts in this registry and the 1,819.4 MW alternative total is stated here for the first time because the difference is material.
• The five-billion-dollar development potential is a valuation, not a construction cost, and is not comparable with programme rates.
• The market count is published as three, four and five; the location cards show three.
• What would close it. One line per site stating commissioning status. Five sentences would resolve the largest single block of uncertainty in this registry.

8Sources

Sources: AirTrunk’s announcement of the Lumina CloudInfra acquisition of 20 April 2026, AirTrunk location pages for BOM1, BOM2, BOM3, MAA1 and HYD1, Data Center Dynamics coverage of the acquisition and Lumina’s Airoli groundbreaking, Moneycontrol, Fortune India and Bloomberg on the transaction, The Next Web on Lumina’s original 600 MW plan, and Datacentermap listing for the BOM1 Navi Mumbai campus.