Start with something small enough to hold.

An air fryer is a compact convection oven. The technology is not new. Countertop convection ovens have circulated hot air over food since the middle of the twentieth century. What changed around 2010 was the packaging: a smaller chamber, a faster fan, food closer to the element, and a name that describes a use case rather than a mechanism.

The energy argument for it is real. Heating a two-litre chamber for fifteen minutes uses less electricity than preheating and running a sixty-litre oven for the same meal. If you cook for one person, that difference is not trivial and it repeats every day.

Now ask the second question, the one no label answers. Did the air fryer replace the oven, or did it join it?

For most households, it joined it. The oven is still there. The microwave is still there. The toaster is still there. What the household actually acquired was a fourth device with its own steel, its own aluminium, its own controller board, its own shipping container, and its own end-of-life. The use-phase saving is measured and published. Everything else is not measured at all.

That gap is the subject of this article, and it turns out to be the single most consequential thing in sustainability strategy right now. The efficiency story holds up perfectly well. What it covers is far narrower than the work it gets asked to do.

First, the part that is genuinely working

It would be easy and lazy to write this as a piece about greenwashing. The data does not support that, and starting there would mean skipping the most interesting fact in the whole subject.

EU ecodesign and energy labelling worked. Measurably.

The European Commission estimates these policies reduced consumer energy spending by around €90 billion in 2022, with projections reaching roughly €150 billion a year by 2030. Average annual energy consumption of refrigerators and freezers has fallen by more than 60%. Under the ODYSSEE-MURE monitoring programme, specific consumption of large appliances has declined steadily since 2000, with washing machines improving fastest at around 5% per year and cold appliances, dishwashers and dryers at around 4% per year.

Then there is the fact that most people arguing about this get wrong, in both directions.

EU households own far more appliances than they did in 2000. Specific electricity consumption for small appliances alone rose from around 400 kWh per dwelling in 2000 to 770 kWh in 2019, an increase of about 3.6% per year, driven almost entirely by owning more things. And yet, at EU level, total electricity consumption per dwelling for appliances and lighting combined is now close to where it was in 2000, and has fallen since 2010 in around twenty member states.

Read that carefully. Regulation on individual appliance efficiency was strong enough to more than offset a two-decade increase in how many appliances people own. That is a substantial policy achievement, and anyone claiming efficiency standards accomplish nothing is arguing against a measured result.

The honest version of the critique therefore has to be more specific than “efficiency is a con.”

Now the other ledger

Here is what happened over roughly the same period, in the same economies, to things that no energy label counts.

Global electronic waste rose from 34 million tonnes in 2010 to 62 million tonnes in 2022, an increase of 82%. It is on track for 82 million tonnes by 2030. In 2022, only 22.3% of it was documented as formally collected and recycled in an environmentally sound way. The UN’s own framing is that e-waste is now rising five times faster than documented recycling.

Within that total, the category called small equipment (microwaves, vacuum cleaners, kettles, toys, e-cigarettes, and yes, air fryers) accounted for 20.4 million tonnes, about a third of all global e-waste, with a documented recycling rate of 12%.

Appliances are also failing sooner. A German Environment Agency study conducted with Öko-Institut and the University of Bonn found that the share of large household appliances replaced within five years because of a defect rose from 3.5% to 8.3% of all replacements between 2004 and 2012. The appliance became more efficient per hour of use and less durable per euro spent, at the same time.

The pattern repeats well outside appliances.

Aviation. A flight taken today produces roughly half the CO₂ of the same flight in 1990. Over the same period demand roughly quadrupled, and total aviation emissions doubled, from around 0.5 billion tonnes to around 1 billion tonnes. Every aircraft got better. The sector got worse.

Cars. Engines and drivetrains improved continuously, and the EU built an entire regulatory apparatus around gCO₂/km. Meanwhile vehicles got bigger. SUVs went from under 20% of global car sales a decade ago to 48% in 2023, passing 50% in advanced economies in 2024. There were more than 360 million SUVs on the road in 2023, producing around 1 billion tonnes of combustion CO₂, which would rank the global SUV fleet as roughly the world’s fifth-largest emitter if it were a country. The IEA’s assessment is blunt: the shift toward heavier SUVs has offset both efficiency improvements in smaller cars and the carbon savings from electric vehicles.

Single-use vapes, UK. 1.3 million thrown away per week in 2022. 5 million in 2023. More than 8 million per week by 2024. Even after the June 2025 disposable ban, around 6.3 million vapes and pods a week were still being discarded. Battery fires in the UK waste stream rose 71% to more than 1,200 in 2024.

Data centres. Compute per watt has improved relentlessly for decades. Global data centre electricity consumption was around 415 TWh in 2024 and the IEA projects roughly 945 TWh by 2030, driven by AI. Twelve percent annual growth for five years running, with AI-accelerated servers projected at 30% per year. This is the same story as aviation, arriving faster.

One number appears on both sides of that chart, and it is the most instructive thing in it.

EU packaging waste per person rose about 20% between 2012 and 2022, reaching 186.5 kg. It also fell 6.5% between 2021 and 2023, from a peak of 190.1 kg to 177.8 kg. Both statements are accurate. Choose the first window and packaging policy looks like failure. Choose the second and it looks like a turning point. Neither window is dishonest on its own. Publishing only the one that suits you is where the dishonesty enters, and it is the single most common manoeuvre in corporate sustainability communication.

Why both ledgers are true at once

The usual explanation offered here is the rebound effect, or the Jevons paradox: efficiency makes something cheaper to use, so people use more of it, and the saving evaporates.

That explanation is real but it is not sufficient, and the EU appliance data is the reason. If rebound were the whole story, household appliance electricity would have risen. It did not. Efficiency regulation beat rising ownership in that specific domain.

The better explanation is duller and much more useful to a business.

A metric measures a boundary. The boundary is a choice. Everything outside it is invisible by construction, not by accident.

An energy label measures the electricity consumed by one unit during its use phase. That is what it says it measures, and it measures it well. It does not measure how much material and energy went into making the unit, how far it travelled, how many units were sold, how long it will last, whether it replaced something or was added alongside it, or what happens to it at end of life.

So the outcome we observe is exactly what the system was designed to produce. Use-phase energy improved because use-phase energy was measured, labelled, regulated and sold. Material intensity, unit count, durability and disposal did not improve because nothing scored them.

The closet

There is a simpler way to hold this, and it does not require any economics.

You have a closet with ten bags. It is full. You decide, sensibly, that you own too many bags and that most of them are decorative rather than useful. So you get rid of the bigger ones.

Then you buy twenty smaller bags to fill the space the big ones left.

Then, because twenty bags will not fit, you build a second closet. Which requires wood, and hardware, and space, and a delivery.

At the end of this you own more bags, more storage, and more materials than when you started, and every individual decision along the way was defensible. The small bags really were smaller. The new closet really was needed. Nobody lied at any point.

That is the pattern, and it scales. The question a metric almost never asks is the only one that would have caught it: did the new thing replace the old thing, or join it?

Efficiency claims are only meaningful net of what they retire. An air fryer that replaces an oven is a genuine improvement. An air fryer that joins an oven is an addition wearing the language of a reduction.

E-waste is the physical residue of the pattern

Small equipment is the category that grows when products get smaller, cheaper, more specialised and more numerous. It is also the category with the worst recovery rate, because small mixed-material devices with glued batteries are the hardest and least profitable things to recycle.

The efficiency logic and the waste logic pull in opposite directions here, and only one of them appears on the product.

The business question

None of the above tells a company what to do. That requires being honest about two genuinely different strategies, both of which are legal, and both of which have real costs.

2026 is an unusually good natural experiment

Roughly 80 to 90 percent of the companies that expected to report under CSRD have just been released from the obligation. Nobody is checking any more. Whatever those companies do next is a revealed preference rather than a stated one.

The ones that keep measuring were doing it for business reasons. The ones that stop were doing it for compliance reasons. And by 2028 or 2029 the difference will be legible from the outside, in supplier resilience, in warranty and returns data, in procurement outcomes with large customers, and in who is ready when the next boundary shift arrives.

That is a rare thing in this field: a clean test, running right now, that nobody designed.

Four questions worth putting to your own business

Where this leaves the air fryer

Roughly where it started, but with the argument in the right place.

The air fryer is a reasonable product that does a real thing well, and for a small household it genuinely uses less energy per meal than a full oven. That saving is measured, published and true.

It is also an additional object, made of material that was mined, assembled in a factory, moved across an ocean, likely to fail sooner than the oven it did not replace, and statistically likely to end up in the third of global e-waste that is recovered 12 percent of the time.

Both descriptions are accurate. Only one of them is on the box.

The useful response is precision about what a metric covers, before building a strategy, a product claim or a regulatory position on top of it. Guilt about a kitchen appliance changes nothing. That precision is a commercial discipline, and over a long enough horizon it is the cheaper of the two options.