The Invisible Trillion

The Mining and E-Waste Question, Honestly

The Chapter Where the Problems Are Actually as Bad as You’ve Heard

I owe you this chapter. After two chapters where the answer was “the problem is smaller and more local than you’ve been told,” it might be tempting to assume this whole book is going to be a long defense of the industry. It isn’t, and this chapter is where that becomes clear.

The energy and water stories, honestly assessed, are mostly stories of real but manageable problems, improving technology, and misleading headlines. The mining and e-waste story is different. In places, at the extremes, it is as bad as the worst things you have heard. Some of it is worse than you have heard, because the worst parts don’t get much airtime in the West.

This chapter is going to be uncomfortable. I think that’s appropriate.

Two Categories, Kept Separate

First, a distinction that most coverage smooshes together.

When people talk about “mining for tech,” they are usually talking about two different supply chains with two different problems, and it helps a lot to keep them straight.

The chip supply chain. Silicon (for the chips themselves), copper (for wiring), a small amount of gold and other metals for contacts, some rare earth elements for specific applications. Globally, chip manufacturing is a small consumer of raw materials. The silicon is basically purified sand. The quantities of metals involved per chip are tiny. The environmental footprint of the raw materials going into chips is relatively modest.

The battery and device supply chain. Lithium, cobalt, nickel, rare earth magnets, all the metals in a modern rechargeable battery. Globally, this is enormous and getting bigger — because electric cars, grid storage, and all those phones and laptops have huge battery demand. The chip inside your phone has a small material footprint. The battery next to it has a large one.

Almost everything people worry about — child labor, environmental devastation, human rights violations — is in the second supply chain, not the first. The phone in your pocket is, materially speaking, mostly a battery with a chip stuck to it, and the moral problems are mostly in the battery.

This matters because arguments about “the ethics of AI chips” sometimes borrow moral weight from mining practices that have approximately nothing to do with AI chips. Cobalt mining in the Democratic Republic of the Congo is genuinely appalling. It is also almost entirely about batteries, not compute. A data center full of GPUs involves very little cobalt. A Tesla involves a lot.

There is, I should say up front, a third lens that this book in particular needs, beyond these two supply chains: the hardware churn of the data centers themselves — what the buildout buys, runs hot, and retires. That is the part of the mining-and-waste story that is genuinely about AI rather than about phones, and I’ll come to it near the end of the chapter. But the human horrors are mostly in the device chain, so that is where I’ll start.

I am going to cover all of this, because your phone has both things in it, the buildout has a hardware life of its own, and you deserve the full picture. But please notice the distinctions, because they will come up again.

All 19 chapters of The Invisible Trillion · Library · All courses