Atmospheric Water Generators: The Physics, the Yield, and the Real Cost Per Litre

If you want the short verdict, it is this: atmospheric water generators can absolutely make drinking water, but they do it by spending electricity to fight thermodynamics. The hard floor is not marketing copy, it is the latent heat of vapourisation. For every litre of water you condense, you must remove about 2,260 kJ of heat, which is roughly 0.63 kWh. That is only the thermal floor, not the electricity bill. Once you include refrigeration losses, airflow, defrost cycles, and the fact that dry air contains very little recoverable water, the real cost per litre moves into a range that is hard to justify anywhere a tap, well, or rain catchment exists.

Water yield from air: the number marketing skips

Here is the part buyers should look at first: how much water is actually in the air?

The table below uses standard psychrometric saturation data — the same kind of values found in a psychrometric chart or steam-table reference — and then multiplies by relative humidity. The numbers are approximate, but they are good enough to show the size of the problem.

Read that slowly. At 25°C and 50% RH, one cubic metre of air contains only about 11.5 grams of water vapour. That means you need to process roughly 87 cubic metres of air to collect one litre of water, before losses. At 10°C and 30% RH, it is only 2.8 grams per cubic metre, so you need more than 350 cubic metres of air for one litre. The machine is not “making water from nothing”; it is concentrating a thin gas into liquid.

This is why the same appliance can look respectable in a humid coastal summer and disappointing in a cool dry climate. It is also why the headline figure of “30 litres per day” is a best-case number: it usually assumes hot, humid air close to the machine’s ideal operating window, not average weather across the year.

First-principles thermodynamics: the floor is real

The basic physics is simple. To turn water vapour into liquid water, you must remove the latent heat released during condensation.

For water:

• latent heat of vapourisation at room temperature: ~2,260 kJ/kg
• 1 kg of water = 1 litre of water
• 1 kWh = 3,600 kJ

So the thermal energy that must be removed per litre is:

2,260 kJ/kg ÷ 3,600 kJ/kWh = 0.628 kWh/kg

That gives a theoretical thermal floor of about 0.63 kWh per litre.

That number is easy to misunderstand, so I want to be precise about what it means:

• it is not the electricity a real machine will use
• it is the heat that must be pulled out of the water vapour stream
• it ignores all losses, all fans, all compressors, all controls, and all heat that leaks back in

A refrigeration-cycle AWG must do more than just condense vapour. It also has to cool incoming air down toward the dew point, keep the coil cold enough to maintain condensation, and move enough air through the system to gather water at a useful rate. That means the electrical input is materially higher than the condensation heat alone suggests.

A clean way to think about it is this:

Electricity per litre = thermal heat removed per litre ÷ system COP + auxiliary loads

Where COP is the coefficient of performance of the refrigeration system. In the real world, a small air-to-water unit is not an ideal heat pump. Once you include coil temperature lift, partial icing, cycling, fans, controls, and standby overhead, a practical operating range is usually much worse than the clean textbook number.

A realistic kWh-per-litre range

Let me work the arithmetic in a way that stays honest.

Best-case operating envelope

In hot, humid air, a decent refrigeration-cycle AWG might end up in the rough neighborhood of 0.5 to 1.5 kWh per litre of electricity input. Why that range?

• the thermal floor is 0.63 kWh/L
• the machine is not moving heat perfectly; real systems have compressor and fan losses
• the air itself must be cooled and dehumidified, not just the water vapour isolated in a vacuum
• if the unit is sized to make meaningful daily volume, it often spends time outside the most efficient part of its curve

One reasonable way to see the lower end is to imagine a machine averaging 600 W while producing 30 L/day in ideal conditions:

0.6 kW × 24 h = 14.4 kWh/day
14.4 kWh/day ÷ 30 L/day = 0.48 kWh/L

That is a genuinely good result, but it is also a best-case result in hot, humid air. It is not the number I would assume for planning.

A more conservative planning range is 1 to 2.5 kWh/L for household-scale machines in real use, with worse effective numbers once ambient humidity drops. At 1.5 kWh/L, the machine is using about 2.4× the thermal floor. At 2.5 kWh/L, it is about 4× the thermal floor. That is entirely believable once losses and partial-load behavior are included.

Why dry or cool air hurts twice

Dry or cool air hurts in two separate ways:

1. There is less water available per cubic metre, so the machine must move more air for the same output.
2. The coil has to work harder relative to the water recovered, so the electricity per litre rises.

If the humidity drops, the yield falls roughly in proportion. But the electrical overhead does not fall in the same neat way. The fans still spin, the compressor still runs, and the machine still has to manage heat transfer. That is why the effective cost per litre can jump very fast as conditions worsen.

What the cost per litre actually looks like

Now let’s translate the physics into money.

I’ll use three electricity tariffs that many readers will recognise:

• $0.10/kWh: cheap power
• $0.20/kWh: common residential power in many places
• $0.35/kWh: expensive power

AWG electricity cost per litre

Even before filter replacements, maintenance, or depreciation, the arithmetic is plain: AWG water is expensive water.

Compare that with other sources

Here is the rough comparison people usually care about:

The gap versus tap water is enormous. If tap water costs around $0.002/L, then an AWG running at $0.20/kWh and 1.0 kWh/L costs $0.20/L just in electricity. That is 100 times more expensive than tap water on a direct per-litre basis. Even if your municipal bill is unusually high, tap water is still usually orders of magnitude cheaper.

Compared with bottled water, AWG can look less outrageous, especially if you are buying convenience-store bottles. Compared with trucked water, the answer depends on local logistics, but AWG is still often dearer than a delivered bulk source once you account for efficient delivery volumes.

Why the 30 litres/day headline is not the whole story

A machine that claims 30 L/day is not lying by definition. It is usually reporting output under a favourable combination of heat and humidity. The issue is that buyers often infer a year-round average from a peak-condition figure.

The psychrometric table already showed the trap. At 30°C and 80% RH, the air carries about 24.2 g/m³ of water vapour. At 20°C and 50% RH, it is only 8.6 g/m³. That is a drop of roughly 65% in available water per cubic metre.

So if a machine is tuned around hot, humid conditions, the same compressor and fan package can suddenly be trying to harvest much less water from the same amount of work. The output falls, the specific energy rises, and the cost per litre follows.

That is the central reason AWGs are so climate-sensitive. They are not universal water sources. They are humidity harvesters.

Where atmospheric water generators make sense

I am not anti-AWG. I am anti-fuzzy arithmetic.

There are narrow cases where an AWG is a reasonable tool:

• No water source exists at all.
• The climate is warm and humid enough that the output is stable.
• Cost sensitivity is low, and water security matters more than unit cost.
• Emergency, military, remote industrial, or expedition use where logistics are harder than electricity.
• Supplemental water production, not primary household supply.

In those settings, the question is not “Is this cheaper than tap water?” because tap water is not available. The question is whether the machine beats the alternative of trucking water long distances, hauling containers, or not having reliable potable water at all. In that framing, AWG can be useful.

When this is the wrong choice

For most buyers, this is the section that matters.

An atmospheric water generator is usually the wrong choice if you have any of the following:

• a functioning municipal tap
• a reliable well
• meaningful rainfall with practical catchment potential
• access to delivered water at normal market rates

If you already have one of those, the math generally ends the discussion. Tap water is dramatically cheaper. Well water is often cheaper still after drilling and pump amortisation. Rainwater harvesting is usually a better capital project if the climate and roof area cooperate. Even delivered water, while expensive, is commonly easier to justify than running a compressor-based machine to mine moisture from air.

The one exception is special context: emergency preparedness, resilience, or locations where the plumbing itself is unreliable. But that is a resilience decision, not a cost-saving one.

My recommendation

My recommendation is straightforward:

• Buy an AWG only if you need water independence and your climate is humid enough to support it.
• Do not buy one expecting cheap water.
• Do not compare it with bottled water and stop there. Compare it with tap, well, rain, and trucked delivery first.

If the goal is lowest-cost potable water, an atmospheric water generator loses badly almost everywhere infrastructure exists. If the goal is portable, self-contained water production in a humid place with no dependable source, it can make sense.

That is the value of the arithmetic: it tells you where the machine belongs, and where it does not.

FAQ

How much electricity does an atmospheric water generator use per litre?

A realistic planning range is about 0.5 to 2.5 kWh per litre for many refrigeration-cycle systems, with best-case humid-climate operation sometimes near the low end and poor conditions pushing higher. The latent heat floor alone is 0.63 kWh/L of heat removed, but electricity input depends on system efficiency and operating conditions.

Is atmospheric water from air safe to drink?

It can be, if the machine is properly designed, maintained, filtered, and sanitized. The bigger issue is usually not whether water can be condensed, but whether the storage tank, filters, and internal surfaces are kept clean.

Why does an AWG work better in humid climates?

Because humid air contains more water vapour per cubic metre. At 30°C and 80% RH, air holds about 24.2 g/m³. At 20°C and 50% RH, it is only 8.6 g/m³. Less vapour means less water recovered per unit of air processed.

How much does atmospheric water cost per litre?

If electricity costs $0.20/kWh and the machine uses 1.0 kWh/L, then the electricity alone is $0.20 per litre. At 1.5 kWh/L, it is $0.30/L. That is far above municipal tap water.

Is an atmospheric water generator cheaper than bottled water?

Sometimes, but only against expensive retail bottled water. It is still usually far more expensive than tap water, and often more expensive than trucked bulk delivery.

Does an atmospheric water generator make sense off-grid?

Sometimes, yes. If you have solar or generator power, no dependable water source, and warm humid air, it can be part of a resilience plan. It is not a general replacement for wells, cisterns, rain capture, or municipal supply.

What is the main limitation of atmospheric water generators?

The main limitation is not the machine alone; it is the combination of thermodynamics and climate. Water in air is thinly distributed, and removing it costs electricity. In dry or cool weather, the yield drops sharply.

Read next

• VPS vs Shared Hosting: Which Should You Choose in 2026?
• What Off-Grid Power Actually Costs Per Month (2026 Breakdown)
• How to Get Water Without a Well: What Each Option Actually Costs (2026)