How much power is 100 MW?
The number sounds large. But it is hard to picture. A city comparison makes it real.
In the model used for this article, a data center has 100 MW of IT capacity, an average load factor of 85%, and a PUE of 1.30.
That gives an average facility load of 110.5 MW and annual electricity use of about 0.97 TWh.
One large AI data center can use electricity on the scale of an entire city. In this model, it uses about half as much electricity as Sheffield and nearly 60% as much as Liverpool or Bristol.
Start with One Important Distinction
A megawatt measures power at a moment in time. A megawatt-hour measures energy used over time.
A steady 100 MW load running for a full year would use:
100 MW × 8,760 hours = 876,000 MWh = 0.876 TWh
But our 100 MW figure describes maximum IT capacity. The computing equipment averages 85 MW. Cooling and electrical systems then raise total facility demand.
100 MW × 0.85 × 1.30 = 110.5 MW
IT capacity × load factor × PUE = average facility load
Over one year:
110.5 MW × 8,760 hours = 967,980 MWh ≈ 0.97 TWh
The earlier 100 MW data center power model guide explains every step in this equation.
Put 0.97 TWh on a Map
We compared the model with the total annual electricity used inside selected city boundaries.
This is wider than a household comparison. Citywide electricity includes homes, offices, shops, public buildings, and most industrial and commercial users recorded in the local dataset.
The map shows location and scale. The percentages answer one question:
What share of this city’s annual electricity would equal the data center’s 0.97 TWh?
One Data Center, Seven City Comparisons
| City | Citywide electricity | Data center share | Data year |
|---|---|---|---|
| Bristol | about 1.63 TWh | 59% | 2024 |
| Liverpool | about 1.65 TWh | 59% | 2024 |
| Sheffield | about 1.90 TWh | 51% | 2024 |
| Manchester | about 2.40 TWh | 40% | 2024 |
| Leeds | about 2.86 TWh | 34% | 2024 |
| Birmingham | about 3.74 TWh | 26% | 2024 |
| Boston | about 6.50 TWh | 15% | 2016 |
The comparison is strongest in Sheffield. One data center would use just over half as much electricity as the whole city.
In Liverpool and Bristol, the figure is close to 60%. In Manchester, it is about 40%. Even against Boston, a city with large commercial, medical, and university loads, the data center is equal to about 15% of annual citywide electricity.
Why the Same Data Center Produces Different Percentages
Population is only part of the answer.
A city with dense offices, hospitals, universities, factories, transport, and digital infrastructure may use much more electricity than another city with a similar population.
That is why household counts can mislead. Boston and Liverpool may look closer when we count homes. Their total electricity systems are very different.
Citywide data captures more of that difference.
A Data Center Is Not Spread Across a City
There is another difference that annual totals do not show.
A city’s electricity demand is spread across thousands of streets, buildings, substations, and customer connections. The data center concentrates more than 100 MW at one site.
The challenge is not only producing 0.97 TWh in one year. It is delivering roughly 100 MW or more to one location, at every hour when the computing load needs it.
That may require a new substation, high-voltage connections, large transformers, backup systems, and upgrades beyond the site boundary.
This is why a country can have enough electricity in total while a specific data center project still waits years for grid access.
How to Read These Comparisons Carefully
The numbers are designed to build intuition, not to rank cities.
- City boundaries differ. A city proper is not the same as a metropolitan area.
- Boston uses 2016 official data. The UK examples use 2024 data.
- The UK values include domestic and non-domestic metered electricity.
- The 0.97 TWh figure depends on the assumed load factor and PUE.
- Some large users may connect directly to higher-voltage systems and may be treated differently across datasets.
So the correct language is “about,” “roughly,” and “under this model.”
Change the Assumptions and the City Comparison Moves
Suppose PUE falls from 1.30 to 1.20. Annual electricity use drops. Every percentage on the chart also falls.
Suppose the average load factor rises above 85%. Annual electricity use grows. The data center becomes a larger share of each city.
The AI Data Center Power Cost Calculator in Python lets readers change these assumptions directly.
The Number to Remember
Under an 85% load factor and a PUE of 1.30, a 100 MW IT data center uses about 0.97 TWh per year. That is about half of Sheffield’s annual electricity, 40% of Manchester’s, or 15% of Boston’s.
A 100 MW data center is not just a large building. It is a city-scale electrical load concentrated at one location.
The next question follows naturally: how many loads of this size could the world need by 2030?
Continue to: How Many 100 MW AI Data Centers Will the World Need by 2030?
Key Vocabulary
Citywide electricity consumption
Electricity used within a city boundary by residential and non-residential customers.
City-scale load
A single electrical demand large enough to be compared with a meaningful share of an entire city’s electricity use.
Concentrated load
Large electricity demand located at one site rather than spread across many users.
Administrative boundary
The legal area used to define a city for statistics and government.
Related Articles
- Why Power, Not Chips, May Limit the AI Data Center Boom
- How to Understand a 100 MW Data Center Power Model
- AI Data Center Power Cost Calculator in Python
Sources and Method
- UK Department for Energy Security and Net Zero, Electricity consumption by Local Authority, 2005–2024 .
- City of Boston, Carbon Free Boston: Energy Technical Report .
- Model calculation: 100 MW IT capacity × 0.85 load factor × 1.30 PUE × 8,760 hours.
Published: July 2026 · City totals are rounded. Comparisons are illustrative and use the source years shown.