Data Centre vs Factory Power Use: Compare

When a community is asked to host a data centre, the first question is usually about electricity: how does its power use compare with other large electricity users, such as factories, hospitals, or airports? In direct terms, a typical hyperscale data centre campus consumes 100–500 MW, which is comparable to a large steel mill or aluminium smelter, and far more than a hospital (5–20 MW) or a regional airport (10–30 MW). This article explains the key comparisons, the factors that make data centres different from other industrial users, and how a community can evaluate a proposed facility.

Background: Why Data Centres Are Different

Data centres are not factories. They do not make a physical product, and they do not consume raw materials. Their electricity goes to two places: powering the servers, storage, and networking equipment (about 70–80% of the total), and running the cooling, lighting, and power conditioning systems (the rest). The defining feature is that the load is constant—24/7/365—with no seasonal shutdown, no shift pattern, and very little tolerance for interruption. A factory may stop at night or on weekends; a data centre cannot without losing customer data or service.

This constant, high-density load is what makes a data centre look like a heavy industrial user, even though it is classified as an information technology facility. In terms of peak demand, a single large data centre can equal a mid-sized town. For example, a 100 MW facility uses about as much electricity as 80,000 average U.S. homes (at roughly 1.2 kW per home). That comparison is often the first thing a community hears, and it is accurate, but it is only part of the picture.

How Data Centre Power Use Is Measured

The standard metric is IT load (the power consumed by the actual computing equipment), usually measured in megawatts (MW). The facility’s total power draw is higher because of overhead. The efficiency ratio is called Power Usage Effectiveness (PUE): total facility power divided by IT power. A PUE of 1.2 means that for every 1 kW of IT load, the building uses 1.2 kW total. Modern hyperscale facilities run at PUE 1.1–1.3; older or poorly designed ones can be 1.5–2.0.

When comparing to other users, use the total facility draw, not the IT load. A 50 MW IT data centre with a PUE of 1.2 will draw about 60 MW from the grid. The same distinction applies to factories: a plant’s nameplate capacity (the sum of all motor ratings) is higher than its actual average draw. Utilities care about both peak demand (for grid capacity planning) and annual energy (for revenue), and both numbers matter to a community.

Comparing to Other Large Users

Here are typical total power draws for large facilities, based on public utility data and industry reports:

• Aluminium smelter: 300–500 MW (one of the largest single-site loads anywhere).
• Steel mini-mill (electric arc furnace): 100–200 MW.
• Large data centre campus: 100–500 MW (hyperscale, multiple buildings).
• Petrochemical refinery: 50–150 MW.
• Automotive assembly plant: 20–50 MW.
• Regional hospital: 5–20 MW.
• Large shopping mall: 10–20 MW.
• International airport (major hub): 30–80 MW (including terminal, runway lighting, and ancillary).

So a data centre is in the same league as heavy industry, but it is not the largest. A single aluminium smelter can dwarf a data centre. The difference is that data centres are often built in suburban or rural areas where the grid was not designed for that load, whereas smelters are usually sited next to hydroelectric dams or coal plants specifically for the power. That siting difference is what makes data centres a hot topic for communities.

Key Differences Beyond the Megawatt Number

The raw power draw is only one factor. Communities should also consider:

• Load factor: Data centres run at near-100% load factor (average power divided by peak power). A factory might have a load factor of 60–70% because of shifts and seasonal demand. That means a data centre buys more energy per MW of capacity, and it places a steadier burden on the grid.
• Power quality requirements: Data centres need very stable voltage and frequency. They often require dedicated substations and redundant feeds. This can mean new transmission lines, which are visible and controversial.
• Backup generation: Data centres install large diesel or gas generators for emergency backup. They are rarely run, but they are tested regularly and can emit noise and particulates. A factory’s backup generators are usually smaller relative to its load.
• Job creation: A 100 MW data centre might employ only 50–100 permanent staff. A factory of the same load employs thousands. This is often the biggest community concern: high electricity use with relatively few jobs.
• Water use: Older or poorly designed data centres use evaporative cooling, consuming millions of litres of water per day. Modern facilities use closed-loop or air-cooled systems, but water is still a factor in arid regions.
• Tax revenue: Data centres are capital-intensive (buildings, IT equipment, generators) and often qualify for property tax abatements. The net fiscal benefit varies widely by state and local policy.

These differences mean that a data centre is not simply a factory with computers. It is a high-availability, high-reliability load that behaves more like a small city’s critical infrastructure than a typical industrial plant.

How to Evaluate a Proposal

A community comparing a data centre to other large users should ask the developer for specific numbers, not just marketing claims. The following steps form a practical checklist:

1. Request the total connected load (MW) and the expected average power draw. The developer should provide a load profile over a year.
2. Ask about the PUE and how it will be measured. A PUE of 1.2 is good; a PUE above 1.5 is poor for a new facility.
3. Check the water usage: cooling tower vs. closed-loop vs. air-cooled. Ask for the expected gallons (or litres) per minute at peak.
4. Ask about backup generator hours: how often they are tested, and whether they will be used for grid demand response (some data centres sell power back to the grid during peak events).
5. Review the grid connection plan: new substation? New transmission line? Who pays for it? This is often the biggest hidden cost to ratepayers.
6. Estimate the employment and tax impact: permanent jobs, construction jobs, and the net property tax after any abatement.
7. Consider the end of life: what happens in 20 years when the facility is obsolete? Is there a decommissioning plan?

This list is not exhaustive, but it covers the main areas where data centres differ from other large users.

Common Problems and Misconceptions

A frequent misconception is that data centres are inherently wasteful. In fact, modern hyperscale data centres are among the most energy-efficient buildings ever constructed, with PUEs approaching 1.1. The problem is not efficiency; it is scale. A 200 MW facility is still a 200 MW facility, even if it is efficient.

Another misconception is that data centres create many jobs. They do not. A large data centre employs about 50–100 people directly, plus some contractors for security and maintenance. That is far fewer than a factory of the same power draw. However, data centres can attract other tech businesses to an area, which is a secondary benefit that is harder to quantify.

A third issue is the so-called “stranded asset” risk. If a data centre leaves or downsizes, the community is left with a purpose-built building that is hard to repurpose, and the utility may have built expensive infrastructure that is now underused. This is less common with factories, which can often be converted to other uses.

Costs and Benefits to the Community

The main cost to a community is the grid upgrade. A 100 MW data centre may require a new substation costing tens of millions of dollars. If the utility spreads that cost across all ratepayers, residential bills rise. Some states require the data centre to pay for the upgrade, but many offer incentives. The benefit is a large, stable tax base (if no abatement is granted) and potentially lower rates for other customers if the data centre’s demand helps the utility achieve economies of scale. The net effect depends heavily on local policy.

There is also the opportunity cost: the same power could serve a factory with more jobs, or thousands of homes. Communities with limited grid capacity must choose. That choice is a political one, and the data centre’s low job count is often the deciding factor.

Alternatives and Efficiency Measures

For communities that want the tax revenue without the full power load, there are alternatives:

• Edge data centres: smaller facilities (1–10 MW) located near users. They have a much smaller footprint and can use existing grid capacity.
• Co-location: multiple tenants share one building, which can be more efficient than a single-tenant hyperscale site.
• On-site generation: solar, wind, or fuel cells can offset some of the grid draw, but they do not eliminate the need for reliable backup.
• Waste heat reuse: some data centres in Europe heat nearby buildings or greenhouses with the warm air from cooling systems. This can turn a waste product into a community benefit.
• Demand response: data centres can reduce their load briefly during grid emergencies, providing a service to the utility. This is increasingly common, but it requires the data centre to have enough redundancy to shed load safely.

These options do not change the fundamental comparison, but they can make a data centre more palatable to a community that is worried about power.

Frequently Asked Questions

How much electricity does a typical data centre use per year?

A 100 MW facility running at 80% average load uses about 700,000 MWh per year. That is roughly the annual consumption of 60,000–80,000 U.S. homes. The exact number depends on the load factor and PUE.

Is a data centre bigger than a hospital?

Yes, usually by a factor of 5 to 20. A large hospital might draw 10–20 MW, while a small data centre starts at 5 MW and a hyperscale campus can be 100 MW or more. The comparison is not close for large facilities.

Do data centres pay for their own power lines?

Often not directly. The utility builds the substation and lines, and the cost is recovered through rates. In some cases, the data centre pays an upfront contribution, but it is common for the cost to be socialized across all ratepayers. Communities should ask specifically about this.

Why do data centres need so much backup power?

Because they must maintain uptime of 99.99% or higher. A power outage of even a minute can cause data loss and financial penalties. Backup generators provide power during the seconds-to-minutes it takes for the grid to stabilize or for the facility to switch to an alternative feed.

Can a data centre run on renewable energy?

Yes, many do. They either build on-site solar/wind, buy renewable energy credits, or sign power purchase agreements with wind and solar farms. However, the intermittent nature of renewables means the data centre still needs a firm grid connection for when the sun is down or the wind is calm.

What is the difference between a data centre and a server room?

A server room is a small room in an office building, typically a few kilowatts to tens of kilowatts. A data centre is a purpose-built facility with redundant power and cooling, often measured in megawatts. The comparison to factories only applies to data centres, not server rooms.

Related

• Survived Winter Storm on Battery Power? What to Know
• Bizarre Things You Can Power With Milwaukee Batteries
• Hercules Power Tools at Harbor Freight: Worth It?