G1 is the only data centre in this registry inside a financial special economic zone. 21,000 square feet, 350 racks, 1 MW of scalable IT load, design PUE about 1.6, district cooling with DX redundancy, and a carrier-neutral fabric at under one millisecond. It is built for IFSC compliance — data embassy arrangements and localisation obligations for foreign financial institutions operating in India’s offshore jurisdiction. Two extremes in one small building. At 350 racks over 1 MW it runs 2.86 kW per rack, the lowest density in this registry. And its declared PUE of 1.6 is the highest of the four efficiency figures published anywhere in these 212 records.
1At a glance
2The highest declared PUE, on the smallest building
| PUE | Overhead on a 1 MW IT load | Relative to 1.4 |
|---|---|---|
| 1.40 — AirTrunk design target | 400 kW | Baseline |
| 1.45 — Colt Navi Mumbai | 450 kW | +12.5% |
| 1.50 — Yotta NM1 design | 500 kW | +25% |
| 1.60 — Yotta G1 | 600 kW | +50% |
A 1.6 PUE on 1 MW of IT load means 600 kW of cooling, losses and distribution to deliver it. Against a best-in-class design at 1.4 that is 200 kW of avoidable overhead, and on Indian industrial tariffs that is real money every month for a decade.
The explanation is not bad engineering. District cooling is a shared utility, not an on-site plant, and at 1 MW in a 21,000 sq ft building the fixed losses dominate. Small facilities almost always carry higher PUE because there is no scale to amortise pumps, transformers and losses across.
So the honest reading is: G1 is the least efficient facility in the registry because it is the smallest, and 1.6 is what a 1 MW build looks like. Which is exactly why the number is worth publishing — a tenant comparing a 1 MW colocation quote against a 30 MW one is comparing energy overheads that differ by half.
Only four records in this registry declare an efficiency figure: AirTrunk below 1.4, Colt at 1.45, Yotta NM1 below 1.5 and Yotta G1 at 1.6. Everything else is silent.
Between the best and worst declared figure is a 14% difference in total power draw for identical IT load. At ₹8 a unit and 1 MW continuous, that is roughly ₹98 lakh a year on the difference.
Any tenant negotiating a colocation contract in India is doing it without this number. That is the single most fixable information gap in the market.
3Data embassy, and why a megawatt matters here
GIFT City is India’s International Financial Services Centre — an offshore jurisdiction inside India, with its own regulator and its own data rules. G1 is the facility that exists to serve it.
| Element | Detail | Why it matters |
|---|---|---|
| Jurisdiction | IFSC zone, Gandhinagar | Offshore financial rules inside Indian territory |
| Compliance scope | Data Embassy and localisation | Foreign regulators accept it as a home-equivalent location |
| Latency fabric | Under 1 ms | Co-located with the exchange and the banks |
| Tenants | Foreign banks, funds, insurers in GIFT | A captive demand pool |
| Capacity | 1 MW | Deliberately small |
| Competition | Effectively none in this registry | No other IFSC-zone facility recorded |
One megawatt is a rounding error in a 2,420.4 MW national total. It is also the only facility in the registry that a foreign financial institution can use to satisfy both its home regulator and Indian localisation law, and that makes it a monopoly on a very small market.
Gujarat counts 5.0 MW in this registry across four records. G1 is 20% of the state — which says as much about Gujarat’s thin data centre presence as about Yotta.
That is the tension worth noting for anyone tracking Indian data centre geography. Gujarat is one of India’s largest industrial states and one of its smallest data centre markets in this file, and the Dholera announcements — L&T’s ₹25,000 crore campus among them — are bets that this changes.
4Procurement
5Limits and open questions
- 1 MW is a scalable IT load, not necessarily current commissioned capacity.
- PUE about 1.6 is a design figure from the operator page, not a measured annual value.
- District cooling efficiency depends on the utility, so the PUE is partly outside the operator’s control.
- “Data Embassy” is the operator’s term and no regulatory instrument is cited on the page.
- No site area is published, so no land-efficiency figure exists.
- 1 MW is counted. It is the smallest counted record in the file and it is counted anyway.
- What would close it. A measured PUE and a regulator reference for the data embassy claim.
6Sources
Sources: the Yotta data centre portfolio page and its GIFT City listing; Yotta colocation pages; and this registry’s records for the declared PUE comparison at PRJ-204, PRJ-207 and PRJ-209.