NTT Global Data Centers is the largest operating data centre operator in this registry at 273.2 MW across 16 named buildings. It is also the company whose own disclosures span a wider range than any other in the file. In December 2025 NTT stated about 400 MW of live IT load in India with more than 200 MW under construction across 22 live data centres. In June 2026 a third-party tracker recorded 237 MW live across 16 data centres with 105 MW under construction in five. This registry counts 273.2 MW operating across 16 records and 169.4 MW in pipeline. Three sources, the same estate, a 163 MW spread on live capacity — 41% of the company’s own figure. None of the three is dishonest, and that is exactly why the difference is worth publishing: it is what “400 MW” costs you when someone asks which 400.
1At a glance
2Three numbers for one estate
| Source | Live capacity | Facilities | Under construction | Method |
|---|---|---|---|---|
| NTT statement, Dec 2025 | ~400 MW | 22 | 200+ MW | Company aggregate, live IT load |
| This registry | 273.2 MW | 16 | 169.4 MW | Sum of named operating building records |
| Third-party tracker, Jun 2026 | 237 MW | 16 | 105 MW | Reported live and earning capacity |
| Spread | 163 MW | 6 | 95 MW | 41% of the company claim |
The six-facility difference is the first clue and the easiest to explain. NTT says 22 live data centres in India; this registry holds 16 operating records. The six that are missing are not hidden — they are facilities that publish no capacity at all, or facilities counted as part of a campus.
A data centre that publishes no megawatt figure cannot be counted, even when it is visibly running. That is the honest limit of a registry built from public disclosure, and NTT is not required to fill it.
The capacity difference is the harder problem. 273.2 MW across 16 buildings is 17.1 MW per building. 400 MW across 22 is 18.2 MW per building. Those averages are close enough that the gap is more likely to be six missing facilities than sixteen overstated ones — but that is an inference, not a reconciliation, and this registry does not publish inferences as facts.
3Why 400 and 273.2 cannot both be right
Both could describe reality if they measured different things. Three explanations are possible, and only NTT can say which applies.
| Explanation | Arithmetic check | Assessment |
|---|---|---|
| Six live facilities with no published capacity | 126.8 ÷ 6 = 21.1 MW each | Plausible. Several NTT buildings in the file are small, but a Delhi or Chennai campus building could carry 21 MW |
| “Live IT load” means contracted or energised rather than occupied | 126.8 ÷ 400 = 31.7% | Plausible and consistent with the whole Indian market, where contracted capacity routinely runs ahead of occupied racks |
| The 400 MW is a design or built figure described as live | Built across 22 facilities at 400 MW averages 18.2 MW | Possible. Sify uses “built” and “operational revenue-generating” as separate figures 54 MW apart |
| The registry is undercounting its own campus children | 16 NTT records, 6 facilities unaccounted | Partially true by construction: a building inside a campus with no separate record contributes nothing |
The second explanation is the one a buyer should price. If “about 400 MW live” means energised or contracted, then the gap between 400 and 273.2 is not an error in either number — it is the difference between capacity that exists and capacity that is paying. In a market where operators routinely pre-lease 60% to 80% of a new building before commissioning, a 32% gap between energised and occupied is entirely normal.
But nobody publishes which one they mean, and that is the defect. Every operator in this file uses “operational”, “live”, “built” and “available” as if they were synonyms. They are not, and the difference across NTT’s own range is 163 MW.
If India has 778.69 MW of operating capacity and NTT alone claims 400 MW of live IT load, then either NTT is 55% of the Indian market, or the national figure is understated by a similar mechanism.
This registry’s answer is that NTT is 273.2 MW of a 778.69 MW operating total — 35.1% — and that its 16 counted buildings are individually sourced.
The two statements cannot be reconciled from public data, and any market report that presents one number without the other is guessing. Price the range, not the point.
4The 16 buildings behind 273.2 MW
| Record | Facility | MW | Basis |
|---|---|---|---|
| PRJ-154 | Mumbai 9 | 62.00 | Facility figure, basis unresolved |
| PRJ-153 | Mumbai 8 | 36.00 | Facility figure, basis unresolved |
| PRJ-159 | NAV1A Navi Mumbai | 30.00 | Operator-stated IT |
| PRJ-152 | Mumbai 7 | 25.60 | Operator-stated IT |
| PRJ-164 | Noida 2 Building A | 22.40 | Operator-stated IT |
| PRJ-173 | Bengaluru 4A | 22.40 | Operator-stated IT |
| PRJ-151 | Mumbai 6 | 19.20 | Operator-stated IT |
| PRJ-168 | Chennai 2A Building A | 17.40 | Operator-stated IT |
| PRJ-150 | Mumbai 5 | 12.50 | Operator-stated IT |
| PRJ-171 | Bengaluru 3 | 12.00 | Operator-stated IT |
| PRJ-170 | Bengaluru 2 | 4.50 | Operator-stated IT |
| PRJ-157 | Mumbai Vikhroli IV | 3.00 | Operator-stated IT |
| PRJ-156 | Mumbai Vikhroli | 2.40 | Operator-stated IT |
| PRJ-166 | Chennai 1 | 2.00 | Operator-stated IT |
| PRJ-163 | Noida 1 (DEL1) | 1.50 | Operator-stated IT |
| PRJ-155 | Mumbai BSE | 0.30 | Operator-stated IT |
| Total, 16 records | — | 273.20 | — |
Two of the sixteen carry 98 MW and neither has a resolved basis. Mumbai 9 at 62 MW and Mumbai 8 at 36 MW are the largest single facilities in the national operating total, and the registry cannot say whether 62 MW is IT load or facility power. If both are facility power at a PUE of 1.5, the true IT contribution is 65.3 MW, not 98 MW — a 32.7 MW overstatement inside the most important number on this page.
That is the single largest open item in the whole registry: 1,456.1 MW of the reconciled national total is facility power or an unresolved basis, and NTT Mumbai 8 and 9 are the biggest names in that bucket.
| Basis | MW | Records | Confidence |
|---|---|---|---|
| Operator-stated IT | 175.20 | 14 | High |
| Facility figure, basis unresolved | 98.00 | 2 | Low — could be 65 MW on IT basis |
| Total | 273.20 | 16 | — |
The defensible NTT operating figure is therefore a range: 240.5 to 273.2 MW. Both ends come from the same 16 records; the difference is whether the two Mumbai facilities publish IT load or facility power. Either way NTT is the largest operating operator in India, and either way the 400 MW claim is above it.
5The under-construction claim
| Source | MW | Facilities | Note |
|---|---|---|---|
| NTT statement, Dec 2025 | 200+ | Not stated | Company aggregate |
| This registry | 169.4 | 7 | Named building records only |
| Third-party tracker, Jun 2026 | 105 | 5 | Reported under construction |
| Spread | 95 MW | — | 47.5% of the company claim |
| Record | Facility | MW |
|---|---|---|
| PRJ-161 | NAV1C and NAV1D, combined planned phase | 40.00 |
| PRJ-160 | NAV1B Navi Mumbai | 30.40 |
| PRJ-165 | Noida 2 Building B | 30.40 |
| PRJ-174 | Bengaluru 4B | 22.40 |
| PRJ-175 | Bengaluru 4C | 22.40 |
| PRJ-169 | Chennai 2B Building B | 17.40 |
| PRJ-177 | Kolkata 1A first building | 6.40 |
| Total | — | 169.40 |
The tracker’s 105 MW against this registry’s 169.4 MW is a 64 MW difference on construction status — and it points the other way from the live-capacity gap. Here the public building records are higher than the tracker, because several of these buildings are announced with dates and footprints but have not reached whatever threshold the tracker applies.
Bengaluru 4B and 4C are the clearest case. Both are identical 22.4 MW buildings on a site with a live 220 kV substation and a commissioned first building. Are they under construction? Physically almost certainly. In this registry, they are pipeline, because no source states a construction start. That is the conservative answer, and it costs the registry 44.8 MW of building-tier capacity it could probably claim.
A dataset that claims 1,277.69 MW of operating-plus-building capacity and can name every building is worth more than one that claims 1,400 MW and cannot.
The conservative choice costs 44.8 MW here and buys the ability to defend the other 1,184.89 MW. That is the trade this registry makes on every ambiguous record.
6What this means for the national total
| Measure | NTT | National | Share |
|---|---|---|---|
| Operating capacity counted | 273.2 MW | 778.69 MW | 35.1% |
| Operating and building counted | 273.2 MW | 1,277.69 MW | 21.4% |
| All counted records | 442.6 MW | 3,513.69 MW | 12.6% |
| Against NTT's own live claim | 400 MW | — | NTT would be 51.4% of the operating total |
Two consequences follow, and only one of them is comfortable.
If 273.2 MW is right, then NTT is 35.1% of India’s named operating capacity, and the market is more concentrated than the eight-operator table on the reconciliation page suggests. One operator, one country, more than a third of the evidenced estate.
If 400 MW is right, then India’s operating capacity is understated by roughly the same 32% across every operator, because every operator uses a similarly loose definition, and the true national figure is nearer 1,140 MW than 778.69 MW — 778.69 divided by the same 68.3% ratio. Both conclusions are publishable. Only one of them is currently supported by building-level records, and it is the lower one.
For an acquirer, the asymmetry is the point. A dataset that says 778.69 MW and shows its work is a diligence asset. A dataset that says “about 400 MW” and does not is a press release.
7The one question to ask NTT
Everything on this page resolves if one question is answered, and it is a fair question rather than a hostile one.
“Of the approximately 400 MW of live IT load you report in India, how much is occupied by a paying customer, how much is contracted but not yet occupied, and how much is energised and available?”
Follow with: “Which of the 22 live data centres are included, and what is the IT load of each?” Six of them are not in this registry because no capacity figure exists for them.
Then: “Are Mumbai 8 and Mumbai 9 published on an IT-load basis or a facility-power basis?” That single answer moves the national operating figure by 32.7 MW.
None of this is an accusation. It is the standard set of questions a lender asks before terming out a facility, and an operator that answers them cleanly is easier to finance and easier to sell.
Gridvik’s position is that the answer is worth money to both sides. NTT will publish an offering circular or a bond prospectus eventually, and the buyer of a capacity dataset wants to already know how the numbers will look when it does.
8How this record is handled
| Field | Value |
|---|---|
| Record type | Portfolio aggregate |
| Capacity | None — the claim of about 400 MW live and 200+ MW under construction is recorded in the notes, not in a field |
| Additive flag | No |
| Counted | 0 MW |
| Why | A portfolio aggregate is a ceiling. Summing it with the 16 facility rows beneath it would double-count NTT’s entire Indian estate |
| What it is for | Holding the operator’s own claim beside the building-level evidence, so the two can be compared in one place |
| Related records | PRJ-150 to PRJ-177 and PRJ-233, the 23 NTT facility, building and campus records that carry capacity |
This is the record type that causes the most damage when mishandled. In the previous count, 45 container records were mixed into a single national total alongside 0.3 MW exchange buildings. The reconciliation of 26 October 2026 separated them, and the effect on NTT was to move it from a diffuse presence to the clearly identified largest operator in the file at 273.2 MW operating.
The 400 MW claim stays published here, dated, attributed and excluded from every total. That is not a rebuke of NTT. It is the only way to hold two true statements about the same company without pretending they are one number.
9Sources
Sources: Moneycontrol, 3 December 2025, reporting NTT’s statement of about 400 MW of live IT load in India, more than 200 MW under construction, 22 live data centres and a ₹2,400 crore Devanahalli commitment; NTT Global Data Centers Bengaluru 4 specification page and launch release, 3 December 2025; Economic Times, 3 December 2025 and June 2026, on the Bengaluru investment and the 22.4 MW first phase; Data Center Dynamics, April 2026, on the Prestige Tech Cloud partnership and the US$288 million investment; Business Standard and PTI reproductions of the Bengaluru 4 launch; a third-party capacity tracker recording 237 MW live across 16 data centres and 105 MW under construction in five, as of 22 June 2026; Gridvik records PRJ-150 to PRJ-177 and PRJ-233, the 23 NTT records carrying capacity. Record verified 26 October 2026.
