On 1 June 2026 the Government of Maharashtra issued AirTrunk a non-binding letter of intent for land at the Raigad Pen Growth Centre in Pen Taluka, for a data centre campus the Chief Minister described as 3 GW at completion, at an announced investment of about ₹2 lakh crore — roughly US$21.05 billion. Three gigawatts is 1.24 times everything this registry counts across all 242 records in every Indian state. It is 70% of AirTrunk’s entire US$30 billion, 5 GW India programme. It is also about three times India’s total operational data centre capacity as measured in 2024. What exists today is a letter. Not a land allotment confirmation, not a power approval, not an environmental clearance, not a phased schedule, and not a construction start. The page records it as the largest claim in the Indian market and treats it as exactly that.
1At a glance
23 GW against everything
The scale is best understood by putting it beside the numbers that can be checked.
| Comparison | Figure | Multiple |
|---|---|---|
| Raigad Pen LoI | 3,000 MW | — |
| This registry, counted | 2,420.4 MW | 1.24x |
| AirTrunk’s whole India programme | 5,000 MW | 70% |
| India’s operational capacity, 2024 | about 950 MW | about 3.2x |
| Maharashtra, counted here | 786.4 MW | 3.8x |
| NTT India, counted | 181 MW | 16.6x |
| Telangana, counted | 341.3 MW | 8.8x |
The fourth row is the one that decides whether this is credible. Three gigawatts is about three times the operational capacity of the entire Indian data centre market as measured in 2024, by a market source outside this registry.
That does not make it impossible. India has announced well over 8 GW of data centre capacity across operators, and a market in which demand is doubling every two to three years can absorb a 3 GW campus over a decade. But it does mean the campus would, on its own, be the Indian market.
For anyone using this record commercially, the useful sentence is: 3 GW is 1.24 times the counted national total, and it is a letter of intent. Both halves of that sentence are load-bearing.
3The ₹20,000 crore error in the source file
This record contains a data error in its own source file, and the registry flags it rather than propagating it. Publishing the correction is more useful than hiding it.
| Source | Figure | Correct? |
|---|---|---|
| Registry master file, this record | ₹20,000 crore, about US$21 billion | no — internally inconsistent |
| Reuters, 1 June 2026 | US$21.05 billion | primary, attributed to the Chief Minister |
| Data Center Dynamics, 4 June 2026 | 200 billion rupees — US$21 billion | equivalent to ₹2 lakh crore |
| CNBC TV18, ANI, UNI | ₹2 lakh crore | consistent with the dollar figure |
| Forbes | about US$21 billion | consistent |
| Correct figure | about ₹2 lakh crore / US$21.05 billion | — |
The error is a factor of ten. ₹20,000 crore is about US$2.26 billion, not US$21 billion. The correct figure is ₹2 lakh crore, which is what the Government of Maharashtra, Reuters and the company all published.
It matters because the per-megawatt arithmetic changes completely. Three gigawatts at ₹20,000 crore would be ₹6.7 crore per megawatt — a fifth of the cost of any other Indian build in this file, and obviously wrong. At the correct ₹2 lakh crore it is ₹66.7 crore per megawatt, which sits sensibly above the ₹53 crore programme rate because this site needs new transmission at 3 GW scale.
A data product that silently fixes its own errors gives a buyer no way to judge its reliability.
A data product that shows the error, names the sources that resolve it, and states the corrected arithmetic gives a buyer exactly that. This page is doing the second.
4A letter of intent is not a project
What the LoI does and does not do is the whole commercial content of this record, and it is worth being mechanical about it.
| Requirement | Status | Evidence |
|---|---|---|
| Political commitment | yes | Chief Minister’s announcement, 1 June 2026 |
| Letter of intent for land allotment | yes | exchanged, non-binding |
| Land allotment confirmed | no | LoI precedes allotment |
| Power approval or allocation | no | not disclosed at any level |
| Environmental clearance | no | not disclosed |
| Phasing or capacity milestones | no | not disclosed |
| Construction start | no | none |
| Commissioned capacity | no | none |
| Binding investment commitment | no | the LoI is explicitly non-binding |
Eight of nine rows are empty. The one that is filled is the one that generates headlines.
For a 3 GW campus the binding constraint is not land and not capital. It is power. No Indian state has 3 GW of spare transmission capacity sitting at a greenfield site in Raigad district waiting for a tenant, and delivering it would require a dedicated 400 kV or 765 kV corridor, generation contracts and years of regulatory process. None of that is published.
The registry records the claim, dates it, attributes it to the Chief Minister’s office and Reuters, and counts zero. It will stay at zero until a substation is energised.
5Pen, and the third Mumbai
The site is inside a state plan rather than a private parcel, which is why the political commitment is real even though the project is not.
| Item | Detail |
|---|---|
| Location | Pen Taluka, Raigad district — the Konkan mainland south-east of Mumbai |
| Size | 1,217 acres of planned smart city, branded Orange City |
| State framing | Described by the Chief Minister as the first city of a third Mumbai vision, modelled on Bandra Kurla Complex |
| Institutional step | MMRDA signed a shareholder agreement for the project on 18 April 2026 |
| Who was present at the LoI | Chief Minister Devendra Fadnavis; Australia’s Consul-General Paul Murphy |
| AirTrunk’s share of the site | Not published. 3 GW on 1,217 acres would be 2.5 MW per acre across the whole city |
Two and a half megawatts per acre across the entire smart city is the lowest land efficiency of any project in this registry — against 12.9 at Chandivali, 9.0 at Mahape and 7.4 at Chandanvelly. That is not a criticism of the plan; it is a description of a greenfield 1,217-acre site with room for roads, power infrastructure and setbacks.
The commercial angle for anyone following Indian data centres is that Raigad is the state’s answer to Navi Mumbai’s land constraint. Navi Mumbai is described by one market source as holding roughly 44% of India’s data centre capacity, and this registry counts 786.4 MW across all of Maharashtra. Either way, the Mumbai region is full, expensive and constrained by 220 kV capacity, and the state is opening a greenfield alternative 60 kilometres south.
Whether AirTrunk builds 3 GW there is a different question from whether Pen becomes a data centre district. The second is already government policy.
6Procurement
7Limits and open questions
- Non-binding letter of intent. Nine of nine project requirements are unmet apart from political commitment.
- The source file carried a tenfold error — ₹20,000 crore instead of ₹2 lakh crore — corrected and explained on this page.
- 3 GW is the Chief Minister’s statement at the LoI signing, not an engineering design or a phased plan.
- No power allocation, environmental clearance or land allotment is published.
- The site area of 1,217 acres is the whole smart city, not AirTrunk’s parcel, which is not published.
- The ₹66.7 crore per megawatt figure derives from a total-investment claim and is not a construction cost.
- Zero megawatts counted. Additive flag unknown.
- What would close it. A power allocation. For a 3 GW campus that is the only number that matters, and India has never allocated 3 GW to a single facility.
8Sources
Sources: Reuters report of 1 June 2026 attributing the figure to the Maharashtra Chief Minister’s post; Data Center Dynamics coverage of 4 June 2026; CNBC TV18, ANI and UNI reports of the ₹2 lakh crore letter of intent; Forbes on the $21 billion project; TechCrunch on the wider $30 billion programme; and coverage of the MMRDA shareholder agreement of 18 April 2026.