Given that we have direct and personal experience of the process of dealing with e-waste - well, one of us does - we are, we’re afraid, deeply unconvinced here:
The ongoing campaign to build thousands of massive datacenters for AI will create a “tsunami” of discarded electronics, putting the world on a course to triple its annual generation of toxic e-waste in the next 25 years, a new report projects.
That there will be waste, yes, that it will be toxic we’d argue about. But rather more importantly this isn’t a problem. For the waste that will be created is of a type not only easy to recycle into constituents but also profitable to do so. Thus we’re done, just leave natural greed to deal with it.
From the report itself:
We learned that the keystone of true circularity is product design and that first we should assess societal need for products, and then, once necessity is established…
We disagree with that vehemently. That’s a vile idea of society, that folk only be allowd to do what some bureaucracy somewhere thinks they should be allowed to do. The rational reaction to that sort of idea is Anglo-Saxonisms about sex and travel. But they’re also wrong in a technical sense:
The industry lurched from the reckless use of PCBs in the 1970s, to the use of leaded solders, to the design of cathode ray tubes, emitting so much radiation that leaded glass had to be deployed to prevent household harm, while utilizing toxic cadmium phosphors in those same CRTs. And then later, we realized nobody had planned the mass disposal of the same devices as we had to find a means to dispose of them by the millions upon the arrival of flat-screen monitors and TVs. And then, of course, the industry turned around and repeated the same toxic pattern by placing mercury phosphors into the fluorescent backlighting of LCD monitors, which were then inserted into the screens in a manner making them almost impossible to remove without breaking the lamps and releasing the mercury.
It is true that it is input and output devices which are either difficult or unprofitable to recycle. Keyboards and mice and monitors and so on - those are where the difficulties are. The actual computing bit is the easy and profitable - the aluminium of the computer box itself, the circuit boards and wiring internally, everyone knows what to do with those. Further, as above, do so profitably and thus we can leave greed to deal with it.
Well, with one proviso. The major cost in any recycling programme is the collection into one big pile - from the natural widely dispersed state - of the thing to be recycled.
So, data centres. They’re rack upon rack of the circuit boards, each containing their gold, tin and copper, the racks are steel or aluminium and the copper cabling to connect them. They’re also in one big stack all in the same place. This is easy to recycle, profitable to recycle, therefore it is not a problem. And, of course, one of the things data centres do not have is some mountain of input and output devices, mice, keyboards, screens and the rest. The clearance of the building might well cost, but the pile of electronics once cleared has a positive value.
We can even test this. The standard test of whether something is worth recycling - will someone pay you for what is to be recycled? In the case of a data centre, yes. The statement “We’re taking our computers out of this building” will trigger a bidding war among people who will pay you to be allowed to cart it all away. For it’s profitable kit. You offered such a building full of it to us, for free, we’d be there at 7 am tomorrow with the lorries to make sure we got it before you changed your mind.
We do understand that there is an entire industry out there dedicated to mithering about things. But that people will not recycle something which is profitable to recycle is not one of those things worthy of mithering about. Which does mean that there’s a certain part of that mithering industry which is worthy of recycling into something productive…..
Tim Worstall