For about a hundred years, the back of almost anything that ran on a battery came off. A torch, a mantel clock, a transistor radio, a child's walkie-talkie, the early mobile phones: each had a little door you slid aside or lifted with a thumbnail, and behind it a cell you could pull out and replace in the time it took to find the right size. Nobody sold this as a feature. It was simply the condition a powered object arrived in, the way a book arrives with pages that turn.
Then, in the space of a decade, the door went away. The iPhone came sealed in 2007; by the middle of the 2010s the smooth, unbroken back was ordinary; by the decade's end it was very nearly everywhere. There was no vote. The door had not failed, and it had not gone out of fashion. It was designed out.
A fair reader will object here, and rightly. Part of that trade was real. We did choose the sealed phone. Gluing the cell in let the body go thin, helped it survive an afternoon in a puddle, and let the battery take the odd shape that filled the shell. People wanted those things and paid for them. That was a bargain struck honestly at the till, with the buyer walking away holding something in return.
But the same sealed body drifted downward, onto things where no such bargain was ever on the table. The £5 light-up toy, the cheap electric toothbrush, the vape built for a fortnight and the bin. These are welded shut too, and they hand the person holding them nothing. No thinness worth the name, no waterproofing anyone uses. Only a few pennies shaved off the cost of making the thing, and a small fire waiting somewhere downstream.
So what is this move — sealing the one valuable part inside the part you throw away? It looks new. It is, in fact, the youngest member of a very old family.
Obsolescence comes in three kinds
On the twenty-third of December 1924, three days before Christmas, a small group of men sat down together in Geneva. They ran the great lightbulb companies of the world: Germany's Osram, the Dutch firm Philips, France's Compagnie des Lampes, the American giant General Electric. The thing they set out to arrange has an almost fairytale simplicity to it. They agreed that a lightbulb should not last as long as a lightbulb could.10
At the time an ordinary household bulb burned for fifteen hundred to two thousand hours, and the better ones a good deal longer. Within a few months of that meeting the cartel (it took the name Phoebus, after the Greek god of light) had written the number down: one thousand hours, and no more. A factory whose bulbs stubbornly outlived their allotted span was made to pay a fine the longer their bulbs lasted. The arrangement was policed with real diligence until the war scattered it in 1940. This is the purest example we have of a whole industry building a thing to break on schedule, and it has a name that has outlived the cartel: the obsolescence of failure.10
A generation later a second kind of obsolescence was given its name, and this time in the open, from a stage. The American industrial designer Brooks Stevens spent the mid-1950s cheerfully preaching what he called planned obsolescence, which he defined as "instilling in the buyer the desire to own something a little newer, a little better, a little sooner than necessary."11 Where Phoebus made the old thing wear out, Stevens made you tired of it while it still worked. One steals the object's lifespan; the other steals your patience. Between them they cover most of what people mean when they say things aren't built to last.
Notice, though, how the intent thins out as we move from the first to the second. Phoebus was a conspiracy in the schoolbook sense: men in a room, a signed agreement, a fine for disobedience. Stevens was softer — no cartel, no punishment, just a philosophy of desire that the whole consumer economy found congenial and adopted without needing to be told. And the third member of this family, the one sitting in your kitchen drawer right now, is softer still. There was no meeting. No designer stood up and named it. Nobody decided that the cell should be welded to the board so that neither could ever be got back out. It simply became, everywhere at once, the cheapest way to build the thing. Phoebus stole the hours on purpose; this steals the metal by accident of arithmetic.
Give it the name it never got, and the family is complete: the obsolescence of failure, the obsolescence of desire, and the obsolescence of recovery — sealing the one valuable part so tightly inside the disposable part that when the object dies, the value dies with it. The question is whether that third thing is real in the way the first two were real, or just a tidy phrase that completes a set. For that we need a number.
Forty-six per cent is lost before the recycler
In May 2026 the scientists at the European Commission's Joint Research Centre (the Commission's in-house science service, the people whose job is to measure things the rest of the argument merely asserts) published a figure that ought to be better known than it is. Of all the strategic and critical raw materials packed into small electrical goods, they reported, forty-six per cent is lost in collection.1
The precision worth holding onto is in that last word. Not lost at the recycler, through some failure of the furnaces or the chemistry. Lost in collection — at the stage before a recycler ever sees the object, while the thing is still whole, still lying in a drawer or a general-waste bin or the back of a van. Nearly half of the cobalt and lithium and the rest, the materials that Europe is anxiously trying to secure from the rest of the world, is gone before recovery even begins.1
What the JRC measured is a scale and a place: how much, and at what stage. It did not, and could not, measure why. That question belongs to argument, not to any laboratory reading, and it is the argument this whole account has to make. The number is not yet an accusation. It is a location. It tells us the loss happens early, at the point where the object is still intact and could in principle still be saved. So the real question is why so much of it is beyond saving by then: why the loss should be decided not at the bin, where you might blame the collector, but far upstream, at the drawing board, months before the gadget was ever sold.
Why the cell can't be saved later
Here is the fact that makes recovery different from the way we usually imagine it. A critical mineral is not fuel. When a gadget dies, its lithium is not burned up or used up; it is still lithium, sitting there in full, chemically exactly as valuable as the day it was mined. Nothing has been consumed. What has changed is only whether you can get at it. And in the language of the people who study this, an inaccessible metal is not gone but dissipated, scattered so finely through the wrong stream that re-concentrating it would cost more than digging up fresh ore.13 The loss is not chemical. It is a loss of access. And access is decided at assembly, by one detail: whether the cell can be lifted out, or whether it is fixed in place.
Follow a welded gadget onto the floor of a recycling plant and you can watch the access close. A lithium cell is supposed to be taken out of a device before the device is crushed: dedicated battery recyclers go further and shred their cells under nitrogen, with the oxygen held very low, precisely so that a cell punctured by the blades cannot catch.12 But a cell welded inside a sealed body cannot be hand-separated at the speed a line runs. So the whole object is sent, instead, down the general route: the open-air shredder built for ordinary scrap. There the crush ruptures the thin separator inside the cell, the two sides touch, the cell shorts and heats and runs away with itself, and the fire that follows is not bad luck. It is the routing, working as the design made it work. The welded cell was always going to take the flammable path, because the design left it no other.12
This is why the loss is locked in early, and it is worth being exact about the strength of the claim. A welded cell is not certainly lost. If the entire device somehow reaches a dedicated lithium-battery recycler, the cell can still be recovered. The loss is driven by routing, not by some absolute law of physics. But most binned gadgets never reach that recycler; they take the ordinary path, the one their sealed bodies force them onto. So the honest way to put it is conditional and unforgiving at once: design sets the ceiling on how much any collection system could ever recover, and collection decides whether that ceiling is reached. The weld does not guarantee the loss. It guarantees that nothing downstream can prevent it. And that raises the obvious defence, the one every sealed gadget seems to offer up in its own favour: isn't the seal simply necessary? Doesn't waterproofing force the weld?
Waterproofing doesn't require a weld
It does not, and the reason is that a single word, sealed, is quietly doing the work of two entirely separate decisions.
The first decision is about the outside of the object: whether the seams of the case are closed against water. That is what an IP rating measures, and it is a statement purely about the joints of the shell: the gaskets, the glued seams, the closed ports. It says nothing whatever about how the battery inside is attached. The second decision is about the inside: whether the cell is held by a clip and a connector you could undo, or fused to the board with a spot-weld you cannot. These two things feel like one thing because a waterproof gadget is usually also a welded one, but nothing about keeping water out of a case requires the cell within it to be permanent. You can gasket the seams and still clip the cell. The two were only ever bundled by habit.
The clearest proof that they are separable is that the law itself separates them. The European Union's battery regulation, in the provision that takes effect across the bloc in early 2027, states the default plainly: a portable battery must be "readily removable and replaceable by the end-user at any time during the lifetime of the product," using "commercially available tools" and no proprietary ones.2 And then it does something telling with the hardest cases. For appliances built to work in water (the ones "intended to be washable or rinseable") and for professional medical imaging equipment, it grants a carve-out. But read what the carve-out actually says. Those devices may be designed so the battery is "removable and replaceable only by independent professionals."2 The exemption does not free the waterproof device from removability. It merely lowers the bar, from any owner to a trained technician. Even the genuinely immersion-proof device, in other words, must keep a cell that comes out.
That is the whole fence, and it is worth pausing on, because it defeats the sealed gadget's best excuse on the excuse's own ground. Where waterproofing is real, the law still refuses to let it justify a permanent weld. So the weld is not the price of keeping water out. It is a separate choice, made for a separate reason. And once that is clear, the only question left is why the choice runs, almost every time, the same way.
The weld is the cheaper joint
Set the two designs side by side on a factory costing sheet and the answer is not mysterious. A removable cell needs somewhere to live: a compartment, which means extra moulded parts (a door, a holder, a set of contacts) and it needs an assembly step to put them together, a moment where a clip is snapped or a fastener driven or a gasket seated. An engineering guide to battery enclosures describes exactly this trade, and notes drily that cost "can be reduced" by moulding parts together "to eliminate both materials and assembly costs."8 A weld needs none of it. It is a single fast joint, laid down in a fraction of a second on an automated line.9 One deleted part, one deleted step, against a joint so fast it is almost free.
At the price of a phone, the difference is a rounding error and the pressures run the other way: toward the thinness and the sealing people will pay a premium for. But at the price of a light-up toy or a supermarket vape, where the whole selling price is a handful of coins and the margin is thinner than the plastic, a deleted part and a deleted step are not a rounding error. They are the margin. And so, at the disposable end of the market, permanence becomes the cost-optimal design — not because anyone chose to trap the metal, but because the cheapest way to join a cell to a board happens also to be the way that can never be undone. Nobody's plan; everybody's incentive. The structure rewards the weld, and no single hand needs to reach for it.
What keeps that incentive from ever correcting itself is that the person who saves the pennies is not the person who pays for them. The maker banks the deleted step in full. The costs land somewhere else entirely: on the lost mineral that has to be mined again from somewhere in the world, and on the fires. Those fires are now frequent enough that several bodies have tried to count them, and it is worth keeping their counts separate, because they measure different things. Material Focus, an industry-funded not-for-profit, reported in May 2024 that more than 1,200 battery fires had broken out in bin lorries and at waste sites across the country in a year, up 71 per cent from 700 in 2022.15 Separately, research by the consultancy Eunomia with the Environmental Services Association (a waste-industry body) estimates that lithium batteries cause nearly half of all waste fires and cost the economy at least £158 million a year.14 Both come from interested parties, and neither is an independent national tally; the exact headline figures are each organisation's own. What is corroborated from the outside is only the direction: one London waste authority reports fires up by more than half, and insurers have begun repricing the risk. The precise number is contested. The trend is not. And every one of those fires began as a cell someone couldn't get out.
This investigation continues below.
Want the next one in your inbox?
The same toothbrush is built both ways
If permanence were truly forced by the engineering, you would expect a given kind of product to be sealed the same way wherever you found it. It isn't, and the people who take these things apart for a living can show you the seam where the choice was made.
Shahram Mokhtari, the lead teardown engineer at the repair company iFixit, has opened up electric toothbrushes from across the market: the same product, the same wet bathroom shelf, the same water constraint every one of them has to meet. In some, iFixit found the battery could be replaced; in others it was "soldered to the circuit board" and effectively unreachable.5 Same category, opposite decision, which is precisely why Mokhtari calls the industry's habit of declaring these devices "not user serviceable" what he calls it: "blatantly untrue."5 The waterproofing was never the reason. Both designs are waterproof. Only one of them threw away the cell for good. iFixit has found the same disappointment even in a brush sold on its green credentials: a motor, a charger, and a battery its owner cannot replace.7
The starkest version of the move lives at the very bottom of the market, in the disposable vape. Crack one open, as iFixit has, and you find not a spent scrap but "a nice little Li-po battery," a genuinely good rechargeable cell: good enough that a cottage industry of harvesters now pulls them out of the bin by the thousand and rebuilds them into packs for power tools and e-bikes. The teardown's own verdict on the design was blunt: "quite a waste to market this product as disposable with a rechargeable battery inside."6 Here the asymmetry the phone complicated is perfectly clean: the buyer got no thinness, no waterproofing they use, nothing at all in exchange for the sealed body — only a rechargeable cell they were forbidden to recharge. The single-use disposable vape was banned in Great Britain from 1 June 2025, which is worth saying plainly;16 but the move survives the ban, because the "reusable" successor that replaced it can carry a cell just as welded and just as unreachable as the one before. The problem was never the nicotine. It was the weld.
Keep the vape in its place, though, as one vivid instance, not the whole argument. The reason to test the pattern on the mid-market toothbrush rather than the vape is that the toothbrush is where the defence is strongest and still loses: a real wet-environment device, genuinely waterproof, built one way by one maker and the other way by the next. The vape shows you a buyer who got nothing. The toothbrush shows you that even when the buyer got something, the weld still wasn't the thing that delivered it. So the honest build exists, in the same aisle, meeting the same constraint. Which raises the last hopeful question: if the alternative is right there, is the system quietly correcting itself?
Two fixes come, about six months apart
It is, in part, and the fairest reading has to say so. Two corrections are genuinely on the way, both real, both dated. From mid-August 2026 (the collection targets run from 15 August) vape producers in Great Britain are made to finance the collection of the waste stream their products create.3,4 That is a real obligation with real money behind it, and it attacks a real problem. But look carefully at which problem. It prices the cleaning up of the mess. It does nothing to the design that made the mess un-makeable in the first place. A producer can pay the collection levy in full, down to the last pound, and go on welding every cell exactly as before. The bill is for the sweeping, not for the thing that keeps needing to be swept.
The only instrument that reaches the design itself is the European Union's removability rule (the one that insists the cell must come out), and it applies from 18 February 2027, about six months after the collection money starts to flow. It reaches Northern Ireland directly, through the Windsor Framework arrangements that keep Northern Ireland aligned with the EU's goods rules, so a shopper in Belfast will be covered. Great Britain, so far, has written no equivalent duty of its own: a consultation is anticipated, but as things stand there is no domestic law requiring the cell in a cheap gadget to be removable. So the map is uneven in time and in place. Collection is priced now; design is priced from 2027; and the part of the design rule that would bite in Great Britain is, for the moment, unwritten. The fires happen in the gap.
None of which is beyond argument, and it is worth naming what would change this reading. If the sealed cells at the disposable end turned out to serve a genuine engineering purpose the buyer actually uses (some real waterproofing or safety benefit), then the charge of pure extraction would soften into the same honest trade the phone made; but no such purpose has surfaced in the disposable class, where the sealed vape gives its owner nothing. If the new collection scheme were already capturing this stream effectively, the 46-per-cent loss would be shrinking; but that loss is the most recent measurement there is, and the collection duty is brand new and its enforcement wholly unproven. And if Great Britain had already legislated a removable-cell duty of its own, the design gap would close; but it has not. Each of these is a door the argument leaves open — a way it could be shown wrong. As of now, none of them holds. Which leaves the reader more or less where they started: at the shelf, holding the gadget, with the whole pattern now visible in it and one practical thing to do about it.
What the sealed body tells you
There is a single test, and it is free, and it needs no chemistry and no teardown. Turn the thing over and ask whether you could get the battery out. A serviceable seam (a little door, a couple of visible screws, a line in the small print promising a replaceable cell) is the sign of an object whose valuable part can still, one day, be recovered. A smooth, seamless body you cannot open without destroying it is the tell: the good part is welded to the throwaway part, and when this dies, it dies whole.
That test does not fix the world, and it should not be sold as if it does. A removable cell is not a saved cell; it is only a savable one. Whether it is actually recovered still depends on the collection system doing its job, which is exactly the thing that has not yet been proven to work. So the promise is modest and precise: choosing the openable design keeps recovery possible, where the welded design forecloses it. That is worth doing, and it is not everything.
Where the sealing buys you nothing (a cheap toy, a novelty gadget, a "disposable" anything) the openable version is the one to reach for, and it exists more often than the market lets on. Some gadgets still run on standard removable cells you can buy in any shop and swap in seconds. There are electric toothbrushes, as iFixit found, that ship with a battery designed to lift out. And some companies build whole devices (laptops, phones) around the promise that any part, the battery included, can be replaced with an ordinary screwdriver. None of them is more exotic than the choice to leave a seam.
But the weight of this should not come down on the shopper, and it is important to be clear about where it belongs. The person turning the gadget over in the aisle did not weld the cell in, and no amount of diligence at the bin will get it out again once it is. The failure is upstream, in the joint, decided months earlier by an arithmetic that rewards the weld and charges the cost to strangers. The test in your hand is not a chore you have to succeed at to be absolved. It is a way of reading, at a glance, a decision that was made for you and against you. What ought to exist is not a more conscientious binner. It is a cheap gadget whose battery comes out — and there is no law of engineering, only a law of cost, standing in the way of building it.
The Magic Wand