In short
- MVA is the grid connection; facility MW is the site load; IT MW is what servers get and tenants lease.
- MVA × power factor − headroom = facility MW; facility MW ÷ peak PUE = IT MW.
- A 60 MVA connection typically yields about 38 MW of IT, 63 % of the nameplate.
- Installed MW includes redundancy: N+1 (4+1) is 1.25 × load, DR 1.33 ×, 2N 2.00 ×.
The power chain, from grid to rack
Power arrives from the utility at high voltage, is stepped down by transformers, distributed by switchgear, protected by UPS and delivered to racks by busbar or cable. At each step some capacity is reserved or lost, so the number shrinks from the grid to the rack. The calculator on the home page walks the same chain.
Six numbers, six meanings
| Term | Meaning | Used for |
|---|---|---|
| MVA | Apparent power the utility will supply | The ceiling for everything below |
| Facility MW | Everything the site draws at peak: IT, cooling, losses, offices | Sizing generators and the substation |
| IT MW (critical load) | Power available to servers, storage and network | What tenants lease and investors value |
| kW per rack | Design power per cabinet | Cooling type, floor loading, busbar sizing |
| MWh per year | Energy actually consumed | Bills, PUE reporting, carbon |
| Installed MW | UPS and generator nameplate including redundancy | Equipment cost; always above the load |
MVA vs MW: power factor and headroom
The grid contract is in MVA, apparent power. Real power in MW is MVA × power factor; modern IT with active power-factor correction runs close to 0.95 or better. Operators also keep headroom, often around 10 %, for start-up currents, growth and the utility's limits. 60 MVA × 0.95 = 57 MW; minus 10 % headroom = 51.3 MW of facility power.
Facility MW vs IT MW: the peak PUE
Facility power includes cooling and electrical losses. Dividing by the peak design PUE, not the annual average, gives the IT capacity that can be guaranteed on the hottest day: 51.3 MW ÷ 1.35 = 38.0 MW IT. The annual PUE, usually lower, is used for energy. More on PUE at allinPUE.
Installed MW and redundancy
| Topology | Meaning | Installed ÷ load |
|---|---|---|
| N | Exactly the capacity needed, no spare | 1.00 |
| N+1 (4+1) | One spare module per block of four | 1.25 |
| DR (4 make 3) | Distributed redundant: four systems carry three loads | 1.33 |
| 2N | Two complete, independent paths | 2.00 |
UPS capacity is sized on the IT load; generators on the facility load. For 38 MW IT with N+1: about 47.5 MW of UPS and 64 MW of generators.
Five questions to ask about any "MW"
- Is this MVA, facility MW or IT MW?
- Is it contracted grid capacity or the site's design load?
- Which PUE was used, peak or annual?
- Is it built, under construction or planned?
- Does it include redundant plant?
Questions about megawatts
What does "a 100 MW data center" usually mean?
Most often 100 MW of IT, the leasable critical load. Utilities and planners often mean grid capacity instead, which is larger. Always ask.
Why is IT MW so much lower than MVA?
Power factor, headroom and the cooling and losses counted in PUE each take a share. Together they often leave 60–65 % of the nameplate for IT.
What is critical load?
The IT load that must stay up through any failure: servers, storage and network. It is protected by UPS and generators.
Does 2N double the cost?
It roughly doubles the UPS and generator plant, not the whole facility. Shell, cooling and fit-out grow far less.
Sources
- Definitions follow common industry usage; redundancy factors as shown in the home page terms section.
- Worked figures use the calculator defaults: 60 MVA, PF 0.95, 10 % headroom, peak PUE 1.35.