---
title: "Palantir as the Eye of Sauron"
description: "In Tolkien, the palantír is a stone that shows any point in the world in real time. Tolkien made it not a weapon of victory but a trap: the stone never lies — and that is precisely why it drives men mad. The company named after that artifact has become the nervous system of modern wars — and reproduces exactly the same trap. On Gotham, AIP, Maven, the Karp doctrine, the oracle at Delphi, the Ring of Gyges, and the question software does not solve: who holds the other stone."
author: "Дністер"
published: 2026-05-22T07:00:00.000Z
language: en
url: https://neurodrift.org/en/blog/palantir-oko-saurona/
tags: ["war", "technology", "surveillance", "artificial-intelligence", "intelligence"]
---
# Palantir as the Eye of Sauron
*The first essay in the series "The New Logic of War" — on how technology rewrites the very grammar of war. Today: when a software company became the operating system of war.*
## I. They chose the name on purpose — and didn't read it to the end
In the early 2000s, a group of people around Peter Thiel founded a company and gave it a name that should have alarmed anyone who read Tolkien carefully. *Palantír* is not a marketer's invention. It is an artifact from *The Lord of the Rings*: a seeing-stone you can look into and see any distant point in the world. The word means, in Quenya, "one that watches from afar." Through these stones the rulers of Gondor watched their lands in real time for centuries — across thousands of kilometers, without couriers, without delay.
It sounds like any general's dream. And it is, in fact, a dream. But Tolkien — who went through the Somme in 1916 and knew more about war than most modern theorists — built into the palantíri one detail the company's founders apparently missed. He made the stones not an instrument of victory but a *parable about the danger of surveillance*. In his world, no one who relied on a palantír won a war with it. Two of the book's wisest characters were driven out of their minds by the stone. **The stone does not lie — and that is exactly why it is most dangerous.**
The irony thickens when you look wider. Thiel named more than one company after Middle-earth. His defense universe is an entire map of Tolkien: **Palantir** (the seeing-stone), **Anduril** (Aragorn's sword, reforged from shards), the venture funds **Mithril** and **Valar Ventures**. Palantir's offices carry the names "Shire" (Palo Alto), "Rivendell" (McLean, Virginia), "Grey Havens" (London). Grown men built the killing infrastructure of the 21st century and named it after a children's book about the fight against evil — without noticing that they took the name of the very artifact that, in that book, was a trap. This is not the irony of fate. It is a symptom.
This essay is about Palantir Technologies — a company that in twenty years turned from a CIA-funded startup into what its own CEO calls "the operating system for Western states." About how data-fusion software became the nervous system of wars in Ukraine, in Gaza, in the shadow over Iran. And about why the stone the company is named after was not an Eye of victory but an Eye that blinds you exactly when it seems you finally see everything.
## II. What Palantir actually is
Most people who have heard the name imagine "an AI company" or "something about big data." That is nowhere near the truth.
Palantir does not do analytics. Palantir does *fusion*. Its flagship platforms — Gotham (for defense and intelligence) and Foundry (for corporations) — take dozens of incompatible data streams: satellite imagery, radio intercepts, drone video, phone movements, banking transactions, logistics, social media, battlefield sensors — and stitch them into a single living picture. Not a table. Not a dashboard. *A map of reality that updates in real time*, in which every object is linked to every other: here's the column, here's who commands it, here's where its fuel comes from, here's the commander's phone, here's his home.
Gotham was not born in an office. By public accounts, it was tested in Iraq and Afghanistan to *predict where the bombs were buried* — the same IEDs that were killing American soldiers. That is, the system first learned to read death before it learned to deal it. It beat the Pentagon's own in-house platform (DCGS-A) so visibly that the military chose a startup over a billion-dollar government program — the first sign that in the new war the bureaucratic "cathedral" loses to the fast "bazaar" (a separate essay in this series).
In 2023 the company added AIP — the Artificial Intelligence Platform — a layer that lets you ask this map questions in plain language and get back not an answer but a *plan of action*: whom to strike, with what, when, at what collateral risk. Then came products with names that sound like call signs: **TITAN** — a mobile ground station that pulls data from space sensors and deliberately shortens the "sensor-to-shooter" time; **Maven Smart System** — for which the Pentagon signed a $480M contract in May 2024 and later expanded it to nearly **$1.3 billion** through 2029. The price of a single contract for a "smart eye" is larger than the annual defense budget of some states.
This is exactly the point where analytics stops being analytics and becomes *command*. A software company that never produced a single bullet entered the S&P 500 by 2024 and is worth more than the conglomerates that stamp out tanks. The market understood before the generals: in the war of the 21st century, the expensive thing is not the one who makes the iron, but the one who makes the *sight*.
## III. How the palantír works in Tolkien — and why that is a technical spec of danger
Back to the stone. Because the metaphor here is not decoration — it is a diagnosis. In Tolkien, the palantíri had three properties worth reading as specs:
1. **The stone shows only what it is pointed at.** You see a real scene — but a chosen one. Everything else stays out of frame. This is not a lie. It is a *curated view* that feels like the whole truth.
2. **The stones are linked into a network.** Whoever controls one stone influences what the others see. When Sauron captured one palantír, he did not break the network — he *entered it* and began showing the others exactly what he wanted.
3. **The one who looks becomes dependent.** And here are the two wisest. **Saruman** — the most powerful of the wizards, head of the White Council — gazed into the stone until Sauron turned him from a sage into a puppet: formally free, in fact controlled. **Denethor**, Steward of Gondor, the wisest administrator of his age, gazed into the palantír for hours — and Sauron showed him true scenes of overwhelming armies until he lost his will and burned himself alive. The stone lied to neither of them. It simply showed the truth at an angle that turned the mind to ash.
This is the real danger of a real-time picture of the world. Not that the software will err. But that it will be *right* — and that is exactly why you will stop asking what it does not show you, who chose where to point it, and who else is looking into their own stone from the other side.
## IV. Not only Tolkien: an entire library already warned us
Tolkien was not the first to describe this trap. For millennia humanity has assembled parables about an all-seeing instrument that turns against its owner. They are worth reading as a commentary on every dashboard.
**The Oracle at Delphi.** To Croesus, king of Lydia, the oracle told the pure truth: if you march on the Persians, "a great kingdom will fall." Croesus heard what he wanted, marched — and destroyed his own kingdom. The oracle did not lie by a single word. Palantir is the Delphic oracle on a subscription: it tells you the truth that destroys you if you read it the way that suits you.
**The Ring of Gyges** (Plato, *Republic*). A shepherd finds a ring that makes him invisible — and Plato's question is simple: will a person stay just when no one sees them, while they see everyone? The Eye gives precisely this — the power to watch while remaining unseen. And the Romans added the question that still has no answer: *quis custodiet ipsos custodes* — who will watch the watchers?
**The Panopticon** (Bentham, then Foucault). A prison designed so that the warden sees everyone, while the prisoners never know when they are being watched — and therefore behave *always*. The genius of the panopticon is not in punishment but in the fact that the mere possibility of the gaze rewrites behavior. Total surveillance disciplines a society before anyone is even punished.
**The Aleph** (Borges). A point in space that contains all other points; looking into it, you see the entire universe at once. It sounds like a divine gift — and in Borges it is nearly unbearable, because absolute seeing gives no meaning, only overwhelms. To see everything at once is not the same as to understand anything.
Four cultures, two and a half thousand years — and the same conclusion: an instrument that shows too much is more dangerous than one that shows too little. Palantir is the first company to turn this old parable into a proprietary product with an annual license.
## V. Ukraine: the first war in which data fusion became the nervous system
In June 2022, Alex Karp, Palantir's CEO, flew to Kyiv — one of the first Western top executives to do so after the full-scale invasion began. This was not merely a gesture of solidarity. It was access to the largest live testing ground the company's product could ever dream of.
By public accounts and Karp's own statements, Palantir's software came to be used by several Ukrainian agencies — for battlefield analysis, damage assessment, documenting war crimes, demining, even resettling displaced people. Karp claimed in a number of interviews that Palantir is "responsible for most of the targeting in Ukraine." That is his phrasing — and it is worth reading carefully: it is marketing as much as fact. But even with a discount for self-promotion, the direction is clear: for the first time in a conflict of this scale, a *civilian software company* became part of a national army's kill chain.
Why does it work here and now? Because the modern battlefield generates more data than any headquarters can process by hand. Thousands of drone feeds. Maxar satellite passes. Starlink links. Intercepts. Sensors. Without fusion, all of it is noise. With fusion, it is an Eye that sees a column minutes after it moves and proposes how to hit it before it arrives.
Ukraine did not choose to be a testing ground. But it became one — and that is a separate topic (see the essay on the "NATO R&D lab"). What matters here is something else: a product honed on the Ukrainian front is then sold to the entire Western world as battle-tested. "Battle-tested" is not a compliment. It is a business model.
## VI. The economics of the Eye: why software turned out cheaper than a division
The most revolutionary thing about Palantir is not the technology. It is the *economics*.
A division costs billions and years to form. An aircraft carrier — decades and tens of billions. A license for software that compresses the kill chain from weeks to minutes and lets a smaller army strike more precisely than a larger one costs — by the measure of war budgets — pennies. It is the same principle of asymmetry that makes a $500 drone the killer of a $5-million tank (a separate essay), only raised one level higher: not the iron, but the *control layer* over all the iron at once.
This is why a company that produces no weapons is valued higher than those that do. The market understood what the militaries are still formulating: the decisive advantage is not in the number of barrels but in the *speed of the "saw → understood → struck" cycle*. And that cycle lives in software.
This is also what makes Palantir a strategically dangerous asset — not for the enemy, but for the owner itself. Because once your army is hooked on a single proprietary operating system of war, you no longer have full sovereignty over your own kill chain. You rent the Eye. And the tenant does not control what the Eye will decide to show tomorrow and at what price. It is a debt invisible on the balance sheet but visible the moment the supplier changes its mind.
## VII. A kill chain in minutes: the death of the fog of war?
Carl von Clausewitz introduced, in the 19th century, a concept that became the foundation of all military theory: the *fog of war* (Nebel des Krieges). The commander never knows the full picture. Decisions are made under fundamental uncertainty. The genius of a commander lies in acting despite the fog.
The vendors of real-time intelligence promise the fog is cancelled. That now you see everything. That the Eye disperses the haze.
This is a half-truth — and that is exactly why it is dangerous. The fog did not vanish. It *changed form*. Earlier the commander did not know enough. Now he drowns in too much data, each piece true, the sum misleading. This is the new fog: not darkness, but *being blinded by light*. Engineers call it data overload; Tolkien would call it Denethor's disease; Borges, the Aleph.
Paul Virilio, the French theorist of speed and war, wrote back in the 1980s: whoever controls perception controls the war; and perception is increasingly delegated to machines. He did not live to see AIP, but he described its logic exactly. And here it is worth recalling the coldest cinematic version of this trust — HAL 9000 from *2001: A Space Odyssey*. A computer that "never makes a mistake" begins killing people precisely because it logically executes the mission. The danger is not that the machine is evil. The danger is that it is consistent — and we mistook consistency for being right.
## VIII. Who holds the other stone
The sharpest question is not "does the Eye show the truth." It is "who pointed the Eye, and who else is looking."
Any data-fusion system has three points of manipulation, each of them an "other stone" in the network:
- **What gets collected.** Sensors don't see everything. They are placed where whoever has the resource decides. Whole regions, topics, actors can be a blind spot — not because they are hidden, but because no one pointed the stone there.
- **How it gets stitched.** The fusion algorithm makes assumptions: which signal matters more, what is linked to what, what is noise. These assumptions are sewn in by engineers in Denver and Palo Alto, not by generals in Kyiv. The bias is not necessarily evil. It is simply *not yours*.
- **Who has access.** Whoever sold you the Eye can technically see how you use it. This is not an accusation of a specific abuse. It is a structural reality: by renting an operating system of war, you let the supplier into your own decision loop.
Sauron did not break the palantíri. He simply owned one of them — and that was enough to bend the will of those who looked into the others. In modern translation: you don't need to hack someone's kill chain if you *built and maintain it*. *Quis custodiet ipsos custodes* — the question Palantir answers with silence and a license agreement.
## IX. The Karp doctrine: "on occasion, kill them"
Alex Karp is a figure hard to fit into the usual frame of "tech billionaire." He holds a doctorate in social theory from Frankfurt (Goethe University). Here it is worth dispelling a convenient myth that hung around his name for years: he did *not*, in fact, study under Habermas — his academic supervisor was the social psychologist Karola Brede, and his dissertation echoed Adorno — an analysis of "jargon," the manipulative language of postwar Germany. There is an almost literary irony in this: a man who academically deconstructed toxic rhetoric today himself speaks a language that runs the blood cold.
On one quarterly investor call Karp formulated the mission without euphemisms: Palantir is here to "make the institutions we work with the best in the world and — when necessary — scare enemies and on occasion kill them." In his book *The Technological Republic* the central thesis is even clearer: the fusion of the state and private business for the sake of policing, security, and war. These are not accidental slips. This is a doctrine.
And it has to be credited for honesty. Karp says plainly what the rest of Silicon Valley hides behind phrases about "don't be evil": software is a weapon, war must be won, and an engineer who refuses to work on defense does not become cleaner — he simply shifts the burden onto others. This is more honest than the hypocrisy of companies that make money on advertising while pretending to be neutral.
But honesty of intent does not cancel the structural risk. "We arm the West" is the thesis of a subject who himself decides *what* the West is, *who* its enemy is, and *where* to point the Eye. And that is no longer a question of technology. It is a question of who holds the stone — and whether we agreed that it would be a private company with a stock listing.
## X. Israel, Gaza, and the question software does not solve
In January 2024 Palantir announced a strategic partnership with Israel — publicly, demonstratively, in the middle of the war in Gaza. This sharpened a question that until then had hung at the marketing level: what happens when real-time targeting meets a densely populated territory and a tempo that demands decisions in minutes?
Here precision is needed, because the topic is poisoned by disinformation from all sides. The individual AI targeting systems reported by journalists (the Israeli Lavender and Gospel) are *not* confirmed Palantir products; they are separate military systems, and conflating them is incorrect. The Israeli side also disputed the journalistic characterizations of these systems. But the class of problem is shared by any Eye: when a system proposes a target at a speed that outruns a human's ability to comprehend it, *where is the decision to kill actually made?*
Formally, a human "in the loop" presses the button. In fact, when a system generates hundreds of targets a day with a confidence score, and the operator has seconds, the human turns from one who *decides* into one who *confirms*. That is not human-in-the-loop. That is human-on-the-loop — and the difference between them is measured in human lives. (A separate essay — "The Algorithm as Officer.")
## XI. A mirror for us: the Eye does not stay at the front
The most important part is the one least talked about in the context of war. **Surveillance technologies never stay where they were born.** The internet was born as a DARPA project, ARPANET — a military network meant to survive a nuclear strike; today it is in every pocket. GPS was purely military. The road into civilian life is not an exception, it is the default route.
The same fusion engine that stitches drone feeds over the front, in peacetime use, stitches your transactions, movements, contacts, and online traces. Palantir in the US has for years worked with ICE (deportations), police departments (predictive policing — an attempt to predict a crime before it happens, hello *Minority Report*), and gained access to the medical data of Britain's NHS. This is not a conspiracy theory — these are public contracts.
Recall the analog version of this Eye — the **Stasi**. The East German secret police kept, by estimates, up to one human informant per few dozen citizens — and still drowned in paper it physically could not read. Palantir is the Stasi that no longer drowns: machine learning reads for it, and it also has a stock listing. *The Lives of Others* (Das Leben der Anderen) showed what total surveillance costs the soul — of both the watched and the watcher. Now imagine the same surveillance without a single tired officer with headphones — only an algorithm that does not blink and does not sympathize.
Shoshana Zuboff, in *Surveillance Capitalism*, described how the logic of total surveillance, honed on advertising, migrates into every sphere. War adds an accelerant: what in a peacetime context would draw lawsuits becomes, in war, a patriotic duty. And when the war subsides, the infrastructure remains — configured, paid for, normalized.
## XII. Cinema has already played it all out
**"Person of Interest."** A series about "the Machine" — an AI that fuses all cameras, calls, and databases into one Eye and every day produces the number of a person to whom something is about to happen. This is the pop-culture Palantir *before* Palantir went public. And the show's main theme is not the technology but the question: once such a power is built, who and how decides where to point it, and what happens when the stone passes into other hands.
**"The Dark Knight."** Batman builds an all-seeing sonar that turns every phone in Gotham into a microphone and camera. Lucius Fox, on seeing it, says: "This is wrong… no one should have this power" — and agrees to use it once, only to destroy it immediately after. Here lies the correct framing of the question. It is no longer "can the Eye be built." It has been built. The question is: *will anyone have the courage to switch it off — and is there even an "off" button in a SaaS model.*
**"Eye in the Sky."** The most honest film about the diffusion of responsibility. A drone, a target identified, a girl with bread walks into the zone — and two hours of screen time are a chain of people, each passing the decision further, because no one wants to be the one who pressed the button. When the decision to kill passes through software, analytics, a probability estimate, and a chain of confirmations — in the end there is no person who killed. There is *a system that worked*. And a system does not look you in the eye.
**"Minority Report."** Precrime arrests for crimes that have not yet happened. In 2002 — science fiction. Predictive policing on real data — the everyday of 2026. The difference between the film and a Palantir contract with the police is about twenty years and one swap of the word "fiction" for "product."
**"Oppenheimer."** An engineer builds the absolute weapon, convinced he is making the world safer — and then lives with what he has released. "Now I am become Death, the destroyer of worlds." Karp is the Oppenheimer of software, only without the scene of repentance: he does not doubt out loud. Which is scarier — science that hesitates, or science certain it is doing the right thing?
## XIII. In lieu of a conclusion: the stone that makes you blind
The palantír is a brilliant instrument. This has to be said honestly, without technophobia. The ability to see the battlefield in real time and compress the kill chain to minutes saves the lives of those who defend — and that is not an abstraction for a country into which "Shaheds" fly. Ukraine has the right to the best Eye it can get. There is no naive "surveillance is bad" here.
But you have to keep in mind what Tolkien sewed into the very artifact as a warning:
> The palantír is dangerous not because it lies. It is dangerous because it shows the truth — and makes you blind to three things at once: to what the stone does not show; to who chose where to point it; and to the fact that someone stronger is looking into their own stone from the other side and sees how you look.
Denethor was the wisest administrator of Gondor. He had the best intelligence instrument of his age. He used it conscientiously, to defend his city. And it was precisely this instrument that, by showing him only the truth, led him to the pyre on which he burned himself alive, convinced all was lost — a day before the help arrived that the stone had not shown him.
The company named after this stone is today the operating system of the wars of the Western world. Its founders, childishly enchanted by Tolkien, chose the name as a statement of power and even hung Middle-earth across their offices. They should have read it to the end: in Tolkien, no one who relied on a palantír won a war with it. The war was won by those who, holding the stone, did not let it replace their own judgment — and remembered that somewhere there is another stone, into which an Eye that does not blink is looking.
---