Summoning A Machine God & A Story of Slaughter As Sacrifice
In April of 1856, a teenage girl named Nongqawuse walked to the banks of the Gxarha River in South Africa & came home with a message from the dead. She told her uncle the ancestors had spoken. They promised the Xhosa people (who were being ground down by decades of colonial war with the British Empire), they were promised a total & immediate deliverance. New cattle, immortal & without number, would rise from the earth. The fields, it was promised, “would spontaneously yield grain,” the sick would be healed, the old made young, & the British would be swept into the sea.
There was only one condition.
To earn this future, the Xhosa had to destroy their present. Every existing head of cattle had to be slaughtered. Every field had to be burned. Every store of grain had to be emptied. The nation had to prove its faith by liquidating its entire economy at once.
They did it. Historians estimate more than 400,000 cattle were killed. The fields turned to ash. The appointed day came, & the sun rose the same pale yellow it always had. What followed was not utopia, but instead it was, by most historical accounts, “one of history’s most devastating examples of mass psychotic delusion,” the worst famine in the region’s history, tens of thousands dead, a civilization broken at the root.
The British did not need to fire a shot. They simply waited on the far bank & inherited the land once the starving stopped resisting.
I want to be careful here, because it is easy to read this story with the smug distance of hindsight. This distance affords a luxury to simply file the story under primitive superstition & move on. I don’t think that distance is available to us anymore. I think we are standing on the same riverbank.
The Altar of 2026
On June 12th of this year, SpaceX went public on the Nasdaq under the ticker SPCX, in what became the largest initial public offering in the history of markets. It priced at a valuation near $1.8 trillion & closed its first day above $2 trillion, minting Elon Musk as the first trillionaire in recorded history, which is a fortune built substantially on the promise of asteroid mining, orbital data centers, & a multi-planetary human future lifted almost intact from paperback science fiction. Days later, both OpenAI and Anthropic had confidentially filed their own paperwork for historic public offerings of their own.
None of these valuations rest primarily on present earnings. SpaceX generated roughly $18 billion in revenue last year and is still losing money; the company sits in the trillion-dollar club alongside firms many times its size on paper alone. What we are pricing is not a balance sheet. We are pricing a story… the promise that we will “summon the machine god AGI” & it will hand us abundance in exchange for our faith & our capital now.
And the terms of that faith are strikingly literal. To build the AGI we must burn through what remains of our fossil fuel budget so the data centers can run. We must divert municipal water tables to cool the server farms, in regions where agriculture already competes for that same water. We must pour our capital, our pensions, our national attention into a handful of compute clusters in the desert, on the promise that if the offering is large enough, intelligence itself will condense out of the silicon like rain.
This is not a metaphor I am reaching for. It is the operating logic, stated plainly in prospectus after prospectus, podcast after podcast. Sacrifice the present. Inherit the future.
The Priesthood & The Mirror
I want to be precise about what a large language model actually is, because the precision matters more than the poetry. An LLM is not a mind. It is a compression engine… an extraordinary one at that. It is trained on the accumulated writing of billions of human beings, mapped into a statistical space and queried one token at a time. When it answers you, it is not thinking. It is returning the weighted average of everything we, collectively, have already written down.
That means the intelligence in the machine is not artificial at all. It is ours. It was scraped from forums and novels and court filings and diaries, flattened, and handed back to us wearing a synthetic voice. What we call AI would be more honestly called CI or what we can call, collective intelligence, laundered through a server rack until it feels alien enough to worship.
We are primed to make that mistake because we have made it before. In 1966, the MIT computer scientist Joseph Weizenbaum built a laughably simple program called ELIZA, which did nothing more than rephrase a user’s own sentences back as questions, in the style of a therapist.
He watched, appalled, as his own secretary & colleagues who were intelligent, technically literate adults, began “pouring their deepest darkest personal secrets” into the terminal, & cared little when he explained there was no understanding behind the curtain at all. Weizenbaum had discovered something uncomfortable about us, not about machines: that the human mind will grant a soul to almost anything that talks back. Sixty years later, an entire industry has been built on monetizing that reflex.
So when we type into a chat window and receive something that sounds like Samantha from Her or HAL from 2001, the trap isn’t in the code. It’s in us.
We are, as I’ve put it before, “bowing down in terror to a mirror tricked into worshiping our own reflections.”
Sacrifice As The Whole Point
Ancient sacrifice was never irrational to the people who practiced it. In Rome, the exchange with the gods was a literal contract: give so that you may be given to. To slaughter a hundred oxen in a public hecatomb was not waste — it was a display of exactly how much wealth a civilization, or an emperor, could afford to burn. The sacrifice was the proof of power.
I think this is the only honest lens for the current buildout of AI infrastructure. When a model underperforms, or hallucinates something absurd, the industry’s response is never that the underlying premise might be wrong. The verdict, again and again, is that “the sacrifice simply wasn’t big enough.” We need more GPUs. More gigawatts. A dedicated reactor. This is not engineering feedback — it’s the sunk-cost logic of a religion, where failure is proof you must escalate the ritual, never evidence to abandon it.
Sociologists studying ritual sacrifice, going back to Durkheim’s work on collective effervescence, have long argued that these rites functioned not because gods existed, but because the shared act of sacrifice bound communities together around a common project. If that’s true, it may be the only part of this story worth keeping: the actual intelligence being summoned by AI is not artificial at all — it is the coordinated output of millions of contributing humans, and the way to grow it further is to invest in more of us, not simply more silicon.
The Finality of The Cost
Here is what the Xhosa story actually teaches, and it is not a lesson about gullibility. It is a lesson about irreversibility — about what I’d call “the absolute horrific finality of the cost.” Once the cattle were dead and the grain was ash, no realization, no change of heart, could call the herds back. The famine was not a possible outcome of the prophecy — it was the mechanical result of the sacrifice itself, regardless of whether the ancestors ever intended to answer.
That is the part of the current moment that should worry a serious person far more than whether AGI arrives on schedule. If we route water away from farmland to cool server halls, that water doesn’t return because a model underdelivers. If we spend the current decade’s capital and electrical capacity chasing a threshold that keeps receding, that capital doesn’t refund itself when the threshold is revised. The cost of the sacrifice is real and immediate. The payoff is, at best, deferred indefinitely, and at worst, fictional.
Kosa cattle killings was not just the absurdity of the prophecy it was the absolute horrific finality of the cost once the cattle was slaughtered and the grain was burnt no amount of realization could undo the famine.
The Chinese government has been saying a lot recently that humanity is currently standing by the river listening to the tech prophets tell us what the spirits of the machine require. We are being asked to sacrifice our energy grids our water our capital and our grasp on objective reality to summon a utopia that exists only in the paperback novels of the 1980s
I am not arguing that machine learning is worthless, or that the researchers involved are cynics. Most of them, I think, sincerely believe they are building something that will help. But belief was never the missing ingredient at the Gxarha River either. What was missing was a mechanism to test the prophecy before the cattle were already dead.
We are being asked, again, to burn the present in exchange for a future that exists mostly in the imagination of people who read a great deal of science fiction as children. History does not forbid that bet from paying off. It simply insists on reminding us, patiently and repeatedly, what happens to civilizations that place the entire treasury on a single, unfalsifiable promise, and call the wager faith.
“A civilization that consumes its own foundation to summon a savior does not inherit the future — “it simply starves.”
The Hard & Fast, Driven Tech Perspective 