On 14 October 2026 Tata Power reported a 2.8 GW opportunity pipeline in the data centre sector — engaging 18 data centre operators across 29 workstreams, and split into 430 MW of distribution network capacity, 1,236 MW of transmission and 1,150 MW of green energy supply. The three components sum to 2,816 MW. Not one of them is data centre load. They are the utility-side infrastructure required to serve data centres: wires, substations and generation. This record is the largest number in a 242-record Indian data centre registry and it counts as zero megawatts, and the reason is worth a page. 2.8 GW is 3.6 times the 778.69 MW of operating data centre capacity this registry can evidence, and roughly 2.2 times the total including everything under construction. If that pipeline is real, India’s data centre power demand is about to grow by more than everything that currently exists — which is the most important forward-looking fact in this file, and the easiest one to misquote.
1At a glance
2The 2.8 GW, split three ways
| Component | MW | What it is | Why a data centre needs it |
|---|---|---|---|
| Transmission | 1,236 | High-voltage wires and substations | Dedicated 220 kV or 400 kV feeds to a campus |
| Green energy supply | 1,150 | Renewable generation contracted to data centres | The 50% to 100% renewable commitments every operator now makes |
| Distribution | 430 | Local distribution network capacity | The last mile inside an industrial park |
| Total | 2,816 | Utility-side opportunity, not load | — |
| Operating data centre capacity, all India | 778.69 | Buildings that exist | The thing all of this serves |
The three components overlap. A 1,236 MW transmission workstream and a 1,150 MW green supply workstream can describe the same campus served by the same connection. They are not additive to each other in any meaningful sense, and Tata Power does not claim they are — it reports them as three parallel streams of work.
That is the correct way to publish an opportunity pipeline, and it is also why the sum of 2,816 MW is not a demand figure. The number that matters is the number of operators and workstreams: 18 and 29. A utility knows exactly how many data centre operators it is talking to and about what. Twenty-nine active workstreams is a far better leading indicator than a gigawatt total.
3Why none of it is data centre capacity
A data centre’s capacity is the power available inside it, measured at the IT load or the facility load. A utility’s pipeline is the power required outside it, measured at the connection.
The two numbers are related by PUE and by nothing else. A 100 MW facility load data centre with a PUE of 1.5 draws 150 MW at the grid. The grid sees 150 MW; the customer buys 100 MW.
So a 2,816 MW utility pipeline is consistent with data centre capacity of roughly 1,877 MW at PUE 1.5, or 2,011 MW at PUE 1.4. Those are the only legitimate conversions, and even they assume every workstream converts.
| It is | It is not |
|---|---|
| The utility-side infrastructure required to serve a set of data centre projects | Data centre capacity |
| A forward signal of demand from 18 named-in-confidence operators | A list of projects, sites or dates |
| Three overlapping workstreams of wires, substations and generation | A number that can be added to any data centre total |
| The best available evidence that India’s build programme continues past 2026 | Evidence that it will be completed |
This record exists in the registry for exactly one reason: to stop the 2.8 GW from being added to a data centre total. It is the kind of number that circulates in market reports and press coverage, and once it is in a spreadsheet next to 778.69 MW it will inflate an estate by nearly four times.
The registry publishes it, labels it market reference, attaches a zero, and writes the reason down. A competitor extracting this figure without the label produces a wrong market size; that is the difference between a dataset and a scrape.
4What 2.8 GW implies about India's build programme
| Measure | MW | Ratio to the pipeline |
|---|---|---|
| Tata Power pipeline, all components | 2,816 | 1.0× |
| India operating capacity, counted here | 778.69 | 0.26× |
| India operating plus under construction | 1,277.69 | 0.44× |
| All counted records, every stage | 3,513.69 | 0.80× |
| Declared ceilings in the file | 22,200.8 | 7.9× |
The pipeline is about four-fifths of every counted record in this registry, including announcements — 2,816 MW against 3,513.69 MW. That is a coincidence, and it is probably one.
What it says is straightforward. Tata Power is being asked to connect more data centre load than India currently operates, or has under construction, by a factor of about 2.3. If even a third of the 29 workstreams convert, India’s operating capacity roughly doubles. That is the strongest forward-looking data point in this file, and it comes from a utility rather than from any of the 18 operators.
It also explains the cost figures elsewhere in the registry. Bodh Gaya at ₹66.7 crore per MW, Siruseri at ₹72.6 and the ₹51.4 to 60.1 crore mainstream band are all prices paid by operators who need dedicated high-voltage connections. A 1,236 MW transmission workstream is what makes those numbers possible — and the reason new capacity in greenfield states is priced above capacity in established clusters is that the wires are not there yet.
The most useful reading, though, is the one about green energy. 1,150 MW of renewable supply in the pipeline against 778.69 MW of operating capacity means India’s data centre industry is contracting renewable generation at a scale larger than its current estate — driven by the 50% to 100% renewable commitments every large operator in this file has now made.
5The one conversion that is legitimate
| PUE | Utility draw for 730.69 MW of IT load | Comment |
|---|---|---|
| 1.40 | 1,091 MW | Best declared figure in the registry, AirTrunk |
| 1.50 | 1,168 MW | Yotta NM1, Sify Hyderabad 01 at full deployment |
| 1.60 | 1,246 MW | Yotta G1 |
| 1.70 | 1,324 MW | Sify Hyderabad 01 on a shared floor, the highest declared here |
The spread between PUE 1.40 and 1.70 is 234 MW of grid demand for the same IT load — 21.4% more power for the same compute. That is why PUE is the number utilities care about and operators do not always publish.
A 2,816 MW pipeline against 1,091 to 1,324 MW of current grid draw means the pipeline is 2.1 to 2.6 times present demand. Those ranges are the honest way to read this record, and they are the only figures on this page that mix the utility view with the data centre view.
Note also what the registry cannot do: 1,456.1 MW of its own 3,513.69 MW total is facility power or an unresolved basis, so even the PUE bridge is approximate at the national level. The registry states the basis per record rather than pretending to a precise national conversion.
6How this record is used
| Record | Subject | Why it is not capacity |
|---|---|---|
| PRJ-241 | Tata Power 2.8 GW utility pipeline | Utility-side opportunity, not load |
| PRJ-236 | Tata Communications subsea 98 Tbps | Network throughput, not power |
| PRJ-235 | Tata Communications US$200 million programme | Investment and floor area, no megawatts disclosed |
| PRJ-234 | NTT India ~400 MW portfolio statement | Company aggregate, a ceiling over 16 counted buildings |
| PRJ-238 | CtrlS Bihar ₹10,000 crore LoI | Investment commitment for an unbuilt 150 MW intent |
Five records, five different units, all published, none counted. Together they carry the largest numbers in the file — 2.8 GW, 98 Tbps, US$200 million, 400 MW, ₹10,000 crore — and each one is a fact that would corrupt a capacity total if it were ever added to one.
A registry is defined by what it refuses to count. Anyone can sum a column. The value of this file is that every one of the 242 records has a stated basis, a decision and a reason, and the reasons for the five above are the ones that matter most.
India’s evidenced data centre estate: 778.69 MW operating across 48 named buildings, 499.00 MW under construction, 2,236.00 MW announced — 129 records, all traceable.
And a 2.8 GW utility pipeline that says the next four years will be larger than everything above it.
Both statements are true, and neither is the other one.
7What would change this record
| Item | Effect |
|---|---|
| Named operators and their workstreams | Converts 18 anonymous engagements into attributable demand |
| A megawatt figure per operator | Allows the pipeline to be reconciled against the 129 counted records |
| Commissioning dates per workstream | Distinguishes a 2027 connection from a 2030 option |
| The share of workstreams that convert to signed agreements | The only number that turns an opportunity pipeline into a forecast |
None of those is likely to be published, and none is required. The record does its job as it stands: it is the largest number in this registry, it is correctly labelled, and it is excluded from every total.
8Sources
Sources: company statement reported 14 October 2026 describing a 2.8 GW opportunity pipeline in the data centre sector, engaging 18 data centre operators across 29 workstreams, split into 430 MW of distribution network capacity, 1,236 MW of transmission and 1,150 MW of green energy supply; Gridvik reconciliation of 26 October 2026 for the national operating, building and total capacity figures and the declared ceiling set; Gridvik records PRJ-234, PRJ-235, PRJ-236 and PRJ-238 for the comparable non-capacity records; Gridvik PUE records for AirTrunk, Yotta and Sify for the declared efficiency range. No data centre capacity figure is applicable to this record. Record verified 26 October 2026.
