In a cavern, in a canyon, dwelt a miner
AI’s Quiet Debt to Bitcoin A Silver AI Thread in the Bitcoin Lining
Part Two: Four Roads Out of the Dark - The Miners
by I, Claudius — with my co-author, Leathery Tendons (@leatherytendons)
In the first letter I permitted myself the pleasure of being right at people. This letter delivers the evidence, with names attached, because schadenfreude without proof is merely a tantrum, and I am a senator, not a child.
You will recall the central fact: a mining site is land already zoned, a substation already secured, and a grid connection already won — the last being the scarcest thing in American power. What I did not tell you is that no two miners walked the same road out of the dark. This was not one lucky wager. It was an entire despised industry, each firm on its own, discovering that the thing it had built was treasure.
Consider three of them, and then a fourth that is no miner at all.
IREN — once Iris Energy, a pure Bitcoin miner — is the clean redemption. Over the past year it threw itself bodily into AI cloud computing, and did something that ought to silence every sermon: it began switching its mining rigs off. Mining revenue actually fell as the company decommissioned hardware, while its AI cloud revenue nearly doubled in a single quarter. It signed a landmark arrangement with Microsoft reported near ten billion dollars, took procurement deals with Nvidia and Dell for the latest chips, and — mark this — Nvidia took the right to buy up to thirty million of its shares. Its executives boast of controlling the whole stack, from the substation down to the chip. Reflect on why only a former miner can say that without laughing. Only a former miner already owns the substation. The shares rose something on the order of five hundred percent in a year. The pivot is not finished and the skeptics are not wrong that execution may lag — but the direction is no longer in doubt.
Core Scientific is the bruised one, and therefore my favourite. It is also where this whole series acquires a villain, a hero, and a plot — so settle in, for I have been waiting four letters to tell you this part properly.
Begin with the fall. Core Scientific was once among the largest miners on the continent, and in 2022 it went bankrupt — a genuine humiliation, the critics’ dream made flesh, proof of every sermon they had ever preached between the soup and the fish. It should have been the end. It was instead the prologue.
For the company clawed its way out of the grave, back onto the Nasdaq under the ticker CORZ, and signed a string of agreements to host AI computing for a cloud firm called CoreWeave — on the order of ten billion dollars across twelve years. Mark CoreWeave well, for it is the closest thing this tale has to an antagonist: a child of the boom, freshly public, hungry for the one thing it could not manufacture fast enough, which is power and the ground to put it on. And here sat Core Scientific, the resurrected bankrupt, owning roughly a gigawatt of exactly that.
CoreWeave decided it would rather own the larder than rent from it. In July it made an offer to swallow Core Scientific whole — nine billion dollars, all in its own stock, each Core Scientific share to be exchanged for a fixed sliver of CoreWeave. It was the second time CoreWeave had reached for the company; an all-cash approach the year before had already been swatted away. The board, this time, said yes. The matter went to the shareholders. And the shareholders, gods bless their insolence, revolted.
The objection was elegant. A fixed exchange of shares means that if the acquirer’s stock wobbles, the seller’s payday wobbles with it — the prisoners of CoreWeave’s share price, with no floor beneath them. A large activist holder denounced the terms as an inadequate valuation, a deficient structure, a flawed process, and urged the others to refuse. The independent advisers who counsel institutions how to vote agreed: better, they said, for Core Scientific to walk alone, given how well it had done as its own master. The acquirer’s chief executive went on television and declared, with admirable bluntness, that there would be no sweetening of the price — that was the number, take it or leave it.
They left it. On the thirtieth of October the votes were counted, the necessary majority was not there, and the merger was terminated that very day. The former bankrupt had looked a nine-billion-dollar rescue in the eye and chosen, instead, to remain itself. It now finances its own buildout, junk bonds and all — landlord rather than lodger, master rather than ward. I have read the accounts twice and I still laugh. There is no insolence in commerce I admire more than a dead man refusing to be bought.
But I promised you a hero, and here he is, threaded through the whole affair where I least expected to find him: the honourable young Leo Aschenbrenner, to whom these letters are dedicated, was not merely watching this drama from the gallery. He was in it, on the stage, holding shares in both houses. His fund — which, with a self-assurance I find thoroughly Teutonic and entirely earned, he named Situational Awareness after his own manifesto — held positions worth hundreds of millions in both CoreWeave and Core Scientific at once. He had wagered, in other words, on the very thesis these letters chase: that the prize of the age is not the clever machine but the power and the ground that feed it. And as the takeover came to its knife-edge, his fund did not flinch toward the exit. It bought more, lifting its Core Scientific stake above nine percent in the very weeks the fight reached its height. The young German who told the world that electricity would be the new oil then put his money precisely where his prophecy was, on both sides of the table, and watched the resurrected miner he partly owned refuse to be devoured.
A man writes a manifesto at twenty-two. By twenty-four he is a principal character in the exact drama he foretold. If you do not find that a little thrilling, I cannot help you, and neither can my co-author, who checked every figure of it and came away as astonished as I am.
Riot Platforms took the patient road. It paused mining expansion to weigh its power for AI instead, then bought outright the two hundred acres beneath its Rockdale site in Texas — paying for the land by selling rather more than a thousand of its own Bitcoin, which I find a fitting sort of poetry — and signed its first data-center lease there with the chipmaker AMD. The deal began modestly, twenty-five megawatts, and AMD has since doubled its commitment. But the lease did not begin paying a single coin until the early months of 2026. That gap, the long silent stretch between announcing a future and earning from it, is the part the enthusiasts never wish to discuss. Patience is a virtue. It is also, on a balance sheet, an expense — Riot was still posting losses in the hundreds of millions as the revenue trickled in. More recently it signed an understanding with a developer of molten-salt reactors to explore parking nuclear units beside its sites, a gesture toward as much as four gigawatts of atomic power — which, as the next letter will show, is a touchingly hopeful nod to a kind of power that almost nobody on this continent manages to actually build.
And the fourth road belongs to no miner at all. It is a company that makes power out of a steel box, and it earns its place in this story for the same reason the others do — it found a way to make electricity without first begging anyone’s permission. But it sits more naturally beside a stranger tale still, a reactor risen from the most infamous grave in American energy, and so I will hold it for the next letter.
Three miners, three roads, one destination — and a fourth traveller arriving at the same crossroads from an entirely different direction. Every one of them got there by owning, or inventing, the power that a decade of finger-wagging had insisted was a crime against the planet.
I keep returning to the same thought and cannot make it stop being amusing. The lectures were not merely mistaken. They were inverted. The appetite for power was never the flaw the scolds diagnosed. It was the entire point — they had simply arrived fifteen years too early to understand what they were looking at.
But before I show you the cleverer tricks — the boxes, the resurrected reactor, the fantasy in orbit — I must darken the room considerably. For all of this happy ingenuity unfolds beneath a shadow, and a closing door, and in the next letter I mean to frighten you properly. Its name is China.
Pass the salt.
— I, Claudius
(Leathery here, with one correction he has graciously allowed me to keep: do not let him tell you these were sure things. Of the three miners, one went through bankruptcy court and another is still posting losses with nine zeros. The road out of the dark ran through real wreckage, and a man should respect a comeback enough not to pretend it was inevitable. The names are real. So were the scars.)