Part IThe Story
Karola Brede told him he could write it in English. It was a kindness, and a reasonable one. The candidate was an American, a Stanford-trained lawyer with a Haverford degree in philosophy, a man who had struggled with dyslexia since childhood, and the doctorate he was finishing at Goethe University Frankfurt would be judged on its ideas rather than its subordinate clauses. He declined. It was important to him, he said, that he be able to write it in German.
The title he put on it runs to twenty-one words in the original, the kind of compound-noun freight train that German academic prose exists to haul: Aggression in der Lebenswelt: Die Erweiterung des Parsonsschen Konzepts der Aggression durch die Beschreibung des Zusammenhangs von Jargon, Aggression und Kultur. Aggression in the lifeworld. The extension of Talcott Parsons' concept of aggression by describing the connection between jargon, aggression, and culture. The university's catalogue dates the dissertation to May 22, 2003. Sixteen days earlier and nine time zones west, on May 6, 2003, an old law-school sparring partner named Peter Thiel had incorporated a company in California and named it after a seeing stone from The Lord of the Rings.
Two documents, one month. One in German, about the way communities sharpen language into weapons and use it to decide who belongs. One in Delaware-flavored corporate English, about a piece of software that did not yet exist and would one day sit inside the targeting cycles of armies. It is difficult, two decades on, to read the first without hearing in it a job description for the man who would run the second.
That man is Alexander Caedmon Karp. By 2025 he was worth, at moments, more than $18 billion. His company, Palantir Technologies, carried a market value of roughly $430 billion, which The Economist calculated at more than 600 times its 2024 earnings and described, with characteristic restraint, as possibly "the most over-valued firm of all time." The same magazine, on whose parent company's board Karp once sat, had named him its CEO of the Year a few months earlier. Time put him on its 100 list and called him "the embodiment of a new kind of Silicon Valley billionaire: an unashamed techno-nationalist who evangelizes Western power." He has called himself a socialist, a progressive "but not woke," and, most recently, a "classic liberal," which is either a typo for classical or the most honest thing he has said.
He has also said that he felt different at "every institution I interacted with." He then spent twenty-two years building one that governments find very difficult to leave.
By the Numbers
The Seeing Stone
$12,000Inheritance from his paternal grandfather that seeded Karp's first firm, Caedmon Group
~$30MInitial cost of building Palantir, largely bankrolled by Thiel and his fund
20 yearsPalantir's run without a profitable quarter before its first GAAP net income in Q1 2023
$6.8BKarp's 2024 'compensation actually paid,' the highest of any U.S. public-company CEO
49.99%Voting interest guaranteed to Karp, Thiel, and Stephen Cohen through Class F shares
$10BCeiling of the U.S. Army Enterprise Service Agreement signed July 31, 2025
11Houses Karp owns around the world, by his own count; he has never learned to drive
Different at Every Institution
The 1967 clinic yearbook of the medical college that is now Sidney Kimmel, in Philadelphia, gives Robert Joseph Karp a single entry of the clipped, devotional kind that yearbooks specialize in. Muhlenberg College, B.S. 1962. Pediatric Society. Psychiatric Forum. "Married the former Leah Jaynes." That October, on the second day of the month, in New York City, the young doctor and his wife had their first son.
The household was, by any measure, a political one. Robert was Jewish, a clinical pediatrician who in later life would sit for a university-press podcast on redlining and lead poisoning, which is to say on the precise ways that a city's maps of who-lives-where get written into the blood of its children. Leah was an African American artist. Her brother, Gerald Jaynes, would become the A. Whitney Griswold Professor of Economics, African American Studies, and Urban Studies at Yale. In 2022 the artist Simon Denny published an essay on one of Leah Jaynes Karp's works, Reward for Information—Victim Eulogized Yesterday, noting with a curator's care that her son Alex's childhood fishing trip "may or may not be depicted" in it. The phrase is perfect. The boy may or may not be in the picture. He would spend his life arranging to be both.
The family settled in Philadelphia. There was a younger brother, Oliver, who goes by Ben, who would eventually move to Tokyo and teach at Temple University's campus there; their mother, in her late seventies, began dividing her year between her house outside Philadelphia and his apartment in Japan. Alex went, as his father had, to Central High School, graduating in 1985. He struggled to read. His parents took him to protests, many of them, for civil rights and for what were then just beginning to be called social justice causes, and what he says he inherited from his mother's side was "a strong affinity to fighting discrimination."
And then there was the matter of what he was. He has described it with an evenness that sounds practiced and probably is.
— Alex Karp, on his racial identityI view me as me. And I'm very honored to be honored by all groups that will have me.
The sentence that precedes it is the more revealing one: "some black people considered me black while some did not." Belonging, in this telling, is not a fact one possesses but a verdict others render, and render inconsistently. A child who learns this early learns one of two things. He can spend his life petitioning the jury. Or he can decide the jury has no jurisdiction and build a courtroom of his own.
Haverford, a small Quaker college on the Main Line, gave him a bachelor's degree in philosophy in 1989. He wanted, at the time, to be a social theorist — not a lawyer, not a banker, certainly not the chief executive of a defense contractor. But he went west anyway, to Stanford Law School, and there, in an institution he seems to have found exasperating in roughly equal measure to everyone else he befriended, he met the person who would make every subsequent thing possible.
Nine Days Apart in October
Peter Andreas Thiel was born on October 11, 1967, in Frankfurt am Main — nine days after Karp, on the other side of the Atlantic, in the city where Karp would one day write his German dissertation. His father, a chemical engineer, moved the family to Cleveland when Peter was one, then onward through a series of mining postings to South Africa and South West Africa, now Namibia. In Swakopmund he attended a German-language school that required uniforms and struck students' hands with a ruler; he changed elementary schools seven times; other children, a biographer would note, did not bother befriending him because they knew he would not stay. He read The Lord of the Rings more than ten times. At Stanford he co-founded The Stanford Review in 1987 to oppose the replacement of the Western Culture curriculum, and there he met René Girard, whose theory of mimetic desire would become his private scripture.
So: a German-born boy raised to be American, and an American-born boy who would teach himself to be German. Both outsiders at Stanford, both born in the autumn of 1967, both formed by a parent's itinerancy or a parent's politics into people who could not quite take an institution at its word. The symmetry is almost too neat to trust, and yet there it is in the birth records.
What bound them at law school was argument. Thiel, decades later, remembered the terms of their friendship with a precision that suggests he enjoyed losing to it.
— Peter Thiel, recalling Stanford Law SchoolHe was more the socialist, I was more the capitalist. He was always talking about Marxist theories of alienated labor and how this was true of all the people around us.
It is worth sitting with that image for a moment: a young man in a Stanford law classroom, surrounded by the future partners of America's great firms, diagnosing them, out loud and to their faces, as alienated laborers. This was the same man who would one day describe himself as defending "our belief that the West is a superior way to live." The distance between those two positions is not as great as it looks. Both are verdicts on the people in the room, delivered by someone who has decided he is not quite of the room.
Karp took his J.D. in 1992. When he announced that he intended to go to Frankfurt for a doctorate in social theory — a path that led, as far as anyone could see, nowhere lucrative — Thiel was the only person who supported the decision. Thiel, meanwhile, clerked for an appellate judge, lasted less than a year at Sullivan & Cromwell, traded currency options at Credit Suisse, wrote speeches for William Bennett, and went home to California to start a fund. The two men's lives proceeded on separate tracks that kept, unaccountably, touching.
The money came from a dead man. Karp received an inheritance of $12,000 from his paternal grandfather and used it to found a London-based money-management firm he called Caedmon Group — Caedmon being his own middle name, and also, as it happens, the name of the first poet in English, an unlettered cowherd who, as Bede tells it, could not sing at feasts until he was given a song in a dream. After some early successes in the stock market, wealthy Europeans gave the firm money to manage. Karp later said his ambition at the time was modest: to accumulate $250,000 and return to Berlin to read and write as a dilettante. His Wikipedia entry, in an unusually tart editorial aside, notes that he "failed to explain how it was possible to build 'a London-based money management firm' attracting 'high net-worth investors' with $12,000." It is a fair question. It may also be the first appearance of a pattern that would recur for twenty years: an enterprise whose actual size and its apparent reach were not quite the same number, and a founder untroubled by the gap.
For a time, Karp helped Thiel raise money for Clarium Capital, the macro hedge fund Thiel launched with $10 million of his PayPal proceeds. The philosopher was learning to sell. He would claim, later, to despise salespeople.
Writing It in German
He lived in Frankfurt from the mid-1990s until 2001, and he considered staying for good. He has said that before he went to Germany he had underestimated how German his own upbringing had been — his paternal family's recent ancestry was German, and something in the household, some rhythm of seriousness or argument, turned out to have a native country. He worked as a research associate at the Sigmund Freud Institute. He wrote, with Karola Brede as a referee, a dissertation on Parsons, aggression, and jargon. He wrote it, against her advice, in German.
The city gave him a political life too. Brede has recalled that Karp always possessed a kind of political agility she associated with liberalism. As a young man he felt an affinity for the FDP, Germany's free-market liberal party. He lived on the same street as Michel Friedman — lawyer, television host, member of the Christian Democratic Union, one of the most recognizable Jewish public figures in postwar Germany — and Karp tried to persuade Friedman that he had chosen the wrong party. One imagines the conversations on the sidewalk: the American doctoral student lecturing the German celebrity on German politics, in German. It is a small scene and an entirely characteristic one. Karp has never met a room he did not want to reorganize.
As a doctoral student, he publicly criticized the novelist Martin Walser, who in 1998 had argued that Germany's culture of Holocaust remembrance had become a moral cudgel, wielded by the liberal intelligentsia to suppress a reunited nation's ordinary feelings. Karp, the son of a Jewish father, the grandson of German emigrants, took the side of remembrance.
Hold that thought. It will come back inverted.
In 2001 he left. He went back to the United States, he said, for work. Two years later the dissertation was catalogued and the company was incorporated, and within a year after that, Thiel asked him to run it. By then Karp had no technical training and no operating experience of any kind — his biographer Michael Steinberger is blunt that he had "no previous technical or business training," though his philosophy background helped him reckon with ethical and political concerns. What he had was German, a theory of how language and aggression bind communities together, and the trust of a man who had backed him when no one else would.
He never really stopped living in German. Steinberger reports that "no matter where [Karp] is, he still speaks German for half the day." He travels with Austrian assistants who are also his closest confidants; two or three are almost always with him. He would later serve as co-managing director of Palantir Technologies GmbH in Frankfurt. The language he insisted on for his dissertation became the language of his private life, a portable homeland carried into conference rooms in McLean and Denver and Miami — a lifeworld, as Parsons and Habermas would have said, that he could unpack wherever he landed.
A Seeing Stone With No Salesmen
There are at least four versions of how Palantir was founded, and the company has never seemed eager to settle on one. They agree on the bones. In 2004, Thiel bankrolled a prototype built by Nathan Gettings, an engineer from PayPal, and two Stanford students, Joe Lonsdale and Stephen Cohen. The idea, as Thiel would later describe it, was to take something like PayPal's fraud-recognition systems and use it to "reduce terrorism while preserving civil liberties" — a phrase that would come to sound, depending on one's politics and the year, either like a mission statement or a punchline.
The cast is worth meeting. Lonsdale, who would later found 8VC and the anti-drone company Epirus, had followed Thiel's path as editor-in-chief of The Stanford Review — a mentee in the most literal sense, a young man who had inherited the master's newspaper before inheriting his company. Cohen would stay longest, becoming president and holding, alongside Thiel and Karp, one third of the special shares that guarantee the founders near-total control. Gettings, the PayPal engineer, would go on to run RoboteX, a Thiel-initiated company selling robots to police departments, firefighters, and SWAT teams, on whose board Karp and Thiel both sat. Each of them, in his way, was a man who had been given a key by Peter Thiel.
Karp was hired as chief executive in 2004. The choice baffled people then and baffles some still: a philosopher with a Frankfurt doctorate and a money-management firm of disputed provenance, running a startup whose customers would be spies. Venture capitalists were not interested. According to The Wall Street Journal, there were three investors besides Thiel and Karp, and little appetite from the major firms, so Thiel and his fund largely covered the initial cost of roughly $30 million. In-Q-Tel, the CIA's venture arm, put in about $2 million. Oakhouse Partners led a $7.5 million Series A in June 2006; REV led a $10.5 million Series B that November.
One decision from those years matters more than any of the money. Palantir declined to give board seats to its investors — including In-Q-Tel, the intelligence community's own fund — and kept declining through every subsequent round. The philosopher who had lectured Michel Friedman about his party affiliation was not going to be lectured by anyone about his company.
For two years, the Journal reported, Palantir revised its software continuously according to the demands of intelligence analysts to whom In-Q-Tel had introduced it. The company's theory, articulated in a 2010 blog post that took Garry Kasparov's experiments in human-computer chess as its text, was that computers alone could not defeat an adaptive adversary. It called its alternative "intelligence augmentation": human analysts, aided by software, exploring data from many sources at once. It was, in its way, a Frankfurt School idea dressed for Langley — a conviction that the human subject, not the machine, remains the site of judgment. (A 2020 essay by the scholar Moira Weigel would be titled, simply, "Palantir Goes to the Frankfurt School.")
The offices got names from Tolkien. The Shire in Palo Alto. Rivendell in McLean, Virginia, a short drive from the CIA. Grey Havens in London. Gotham, the defense and intelligence platform, shipped in 2008. A VentureBeat profile in 2009 described a company that had grown from zero to 150 employees in five years and was known across the intelligence community mostly by word of mouth. Palantir openly mocked the idea of a sales staff; it would be January 2019 before Bloomberg could report that it had stopped mocking salespeople and started hiring them. In their place were engineers sent to live inside the customer — "forward-deployed," in the company's martial vocabulary. In 2009, JPMorgan had 120 of them helping a bank executive named Peter Cavicchia III use Palantir's Metropolis product to watch employees' emails, browser histories, and the GPS locations of company phones, flagging signs of disgruntlement for an insider-threat team that might then follow them after hours. The seeing stone, as Tolkien readers will recall, was a dangerous instrument. Those who looked into it tended to see what someone else wanted them to see.
Twenty Years Without a Profit
The early years produced moments of glamour. On June 18, 2010, Vice President Joe Biden stood at the White House beside the budget director Peter Orszag and credited Palantir's software with helping the Recovery Accountability and Transparency Board catch fraud in stimulus spending; the capability, he announced, would be deployed next at Medicare and Medicaid. Fourteen years later the Financial Times would describe Karp as "a major Biden donor." Some debts get repaid in the currency they were issued in.
By 2011, analysts estimated Palantir's revenue at around $250 million. By 2013 a leaked document showed at least twelve U.S. government groups among its clients: the CIA, NSA, FBI, DHS, CDC, the Marine Corps, the Air Force, Special Operations Command, West Point, and others. The databases of the CIA and FBI, TechCrunch reported, had been siloed until Palantir linked them. The company was valued at $9 billion in December 2013, $15 billion in November 2014, $20 billion by the end of 2015 after an $880 million round. It was, by any Silicon Valley yardstick, a triumph.
It was also not making money. It would not make money — not one GAAP-profitable quarter — for twenty years.
The single largest customer that refused to buy was the United States Army, which had its own data system, the Distributed Common Ground System, known as DCGS-A, and did not want anyone else's. Soldiers in Afghanistan reported that Palantir did a better job predicting where improvised explosive devices would be buried. In 2012 the Army Test and Evaluation Command produced a favorable report on Palantir and recommended installing it; the command was ordered to modify the report because of sensitivities "about purchased systems in use that came outside the normal acquisitions process." In 2013 the Army canceled an independent MITRE evaluation that had also come back mostly positive. Fortune would later observe that the distrust ran both ways, and that Palantir approached the Pentagon with the barely concealed attitude that the Pentagon was incompetent at building data platforms. Which was, of course, precisely what Palantir believed.
So in 2016 Karp did what almost no software startup does to a prospective customer the size of a nation-state. He sued it. Palantir argued that the Army had discriminated against a commercial product in favor of a costlier, riskier in-house one — and won. The Army appealed in 2017; the matter was settled when Palantir beat Raytheon head-to-head and took the contract in 2019. The same year, it picked up the Pentagon's Project Maven, the drone-imagery AI work that Google's employees had shamed their own company into abandoning. Thiel, by this point, had his own complaint about Karp — that he always seemed to choose the hard, complex way when a straighter road would have done. Thiel left the company in his hands anyway.
There were embarrassments along the way, some of them severe. In 2010, a law firm reportedly asked Palantir, HBGary Federal, and Berico Technologies to draft a response to "the WikiLeaks Threat"; when Anonymous published HBGary's internal documents in early 2011, the plan was revealed to propose Palantir software as "the foundation for all the data collection, integration, analysis, and production efforts," alongside slides suggesting disinformation and the disruption of Glenn Greenwald's support for WikiLeaks. Karp cut ties with HBGary and apologized to "progressive organizations ... and Greenwald ... for any involvement that we may have had in these matters." An employee was placed on leave pending an outside review. He was later reinstated. The software firm i2 sued, alleging Palantir had obtained its code through a private-investigation cutout arranged by Shyam Sankar, then director of business development and later chief technology officer; i2 settled for $10 million in 2011. In September 2016 the Labor Department sued Palantir for routinely eliminating Asian job applicants; it settled in April 2017 for $1.7 million without admitting wrongdoing.
And there was the mystique, which the company did more to cultivate than to correct. Karp announced in 2013 that Palantir would not go public, because doing so would make "running a company like ours very difficult." The story that Palantir had helped find Osama bin Laden circulated widely; the war-studies scholar Anthony King has called it at best an exaggeration, part of the propaganda used to consolidate what was, by 2011, a small foothold in the defense world. When Bloomberg ran a 2018 feature under the headline "Palantir Knows Everything About You," The New Republic later observed, the company's leaders did nothing to dispute its excesses — the story helped the valuation, which reached $40 billion by January 2021. In 2017 BuzzFeed News reported that the relationship with the intelligence agencies was far rockier than the legend suggested: the NSA preferred its own engineers, and the CIA had been so irritated by publicity linking it to Palantir that it tried to cancel the contract. Asked about the FBI, Karp was serene. Palantir's position there was secure, he said, because "They'll have no choice."
In 2018, Morgan Stanley valued the company at $6 billion. The same year, The Wall Street Journal reported that bankers had floated $41 billion. Two numbers for one company, different by a factor of seven — the Caedmon question, asked again, at scale.
The Uses of a Pariah
In May 2022, The Times of London ran an interview with Karp under a headline that could have served as his corporate epitaph: "War is here — you need a pariah on your side." Palantir had been, for most of its existence, the company that other Silicon Valley companies pointed to when they wanted to feel clean. Karp had decided, at some point, to stop minding.
The clearest case was immigration. In 2014, Immigration and Customs Enforcement awarded Palantir a $41 million contract to build Investigative Case Management, a system for tracking the personal and criminal records of immigrants. For years the company's line was that because the contract ran through Homeland Security Investigations, ICE's criminal-investigations arm, Palantir played no part in deportations. In 2017, documents obtained by The Intercept and the advocacy group Mijente showed that ICE considered the software "mission critical" and had used it in a joint operation targeting the family members of undocumented children who had crossed the border. In August 2019, HSI led raids on food-processing plants in Mississippi that ended with nearly 700 arrests. A few months later, Karp acknowledged on CNBC that Palantir's blanket denials had been misleading. That same year, he renewed the ICE contract for $50 million over three years.
He has described family separation as "a really tough, complex, jarring moral issue," and said he favors "a fair but rigorous immigration policy." Amnesty International faulted Palantir in 2020 for failing to conduct human-rights due diligence. In April 2025, ICE paid the company $30 million to build the Immigration Lifecycle Operating System, designed to speed deportations; when the project was renewed that September, some employees asked internally whether the contract should end if ICE's use veered into "extrajudicial actions." A separate tool, called ELITE, built a map of potential deportation targets drawing in part on data from the Department of Health and Human Services, which includes Medicaid. Paul Graham, the co-founder of Y Combinator and not a man given to activism, wrote that Palantir was "building the infrastructure of the police state." Karp's own answer to critics of the policy, in a later context, was characteristically compressed: "You have an open border, you get the far right." The son of civil-rights marchers had arrived at a theory in which enforcement is the antifascist position.
Then there was Ukraine. In June 2022, Karp became the first chief executive of a major Western company to meet Volodymyr Zelenskyy after the Russian invasion. Palantir initially supplied its software and services pro bono, including paying for the new staff it hired inside the country; as costs climbed, Ukraine's allies stepped in. What followed was a kind of total integration. Ukraine came to use Palantir for air defense, mine warfare, and logistics, but also to manage its wartime economy, its schools, its refugees, its demining. Gotham fused drone video, intercepted communications, satellite imagery, and civilian tips submitted through an app called E-Enemy into a single battlefield picture that helped commanders choose targets — shortening, in the military idiom, the kill chain. With Ukraine's Ministry of Defense, Palantir organized the Brave1 Dataroom, where 100 companies trained more than 80 AI models on real battlefield data. Mykhailo Fedorov, later Ukraine's defense minister, called Palantir an "effective IT weapon" that the Ukrainians had and the Russians did not, and recalled that Karp had been struck by how Ukraine folded the technology "into the general logic of warfare." The Austrian analyst Markus Reisner put it more bluntly in June 2026: Ukraine had become Palantir's laboratory.
Israel followed. In January 2024, Palantir signed a strategic partnership with the Israel Defense Forces to support its "war-related missions." Karp's public support for Israel after October 7 was emphatic enough that, by his own account, several employees left. Palantir announced it would set aside 180 positions for Jewish college graduates, citing campus antisemitism. In 2025 he received the Lamplighter award from Rabbi Levi Shemtov. In a 2024 interview he described himself as Jewish — a fact his father's yearbook had implied fifty-seven years earlier, and which the son had rarely put forward as the headline of his identity until the world forced the question.
The pariah strategy had a commercial logic, and it is worth stating plainly. When everyone else in Silicon Valley wanted to be liked, the company that did not care about being liked could take the contracts no one else would touch. Maven, after Google fled. ICE, after the walkouts. Kyiv, while competitors waited to see how the war broke. Karp had once diagnosed his law-school classmates as alienated laborers. He had built a company that ran on the opposite fuel: a workforce, as he would later put it, with a warrior culture that valued excellence over money, held together by the conviction that the rest of the industry was too comfortable to do what needed doing.
Jargon and Aggression
In 2018, Karp said he was a socialist. He said he had voted for Hillary Clinton. He described himself as a progressive "but not woke." In 2024 he told an interviewer he was "not thrilled" with the direction of the Democratic Party but would still be "voting against Trump." In 2025 he called himself a "classic liberal." By November 2025, talking to Wired, he was imagining an argument with a family member in which he would be "pointing out that Trump's decisions on AI, and his decisions on the Middle East, are very different than people in the Democratic Party would have made, and very good." If the party's Zohran Mamdani wing took over, he said, he would consider the party to have left him.
The editor of Current Affairs, Nathan J. Robinson, wrote in 2024 that Karp's socialism "seems to have some idiosyncratic personal definition in mind that has nothing in common with the socialist tradition." That is almost certainly right, and also beside the point. Karp's politics have never been a set of positions so much as a set of enemies, and the enemies have changed while the posture has not. In the Stanford classroom the enemy was the comfortable professional class. In the Frankfurt street it was the wrong party. In Palo Alto it was Facebook — when Palantir expanded into a former Facebook office, it painted over a famous wall associated with its predecessor, a small act of exorcism with real symbolic weight, given that Thiel had been Facebook's first outside investor and a member of its board. In August 2020, moving the company's headquarters to Denver, Karp wrote that the "engineering elite of Silicon Valley ... do not know more about how society should be organized or what justice requires."
And then, after October 7, the enemy became the university. In December 2023, as demonstrations convulsed Columbia, he said: "There is literally no way to explain the investment in our elite schools, and the output is a pagan religion — a pagan religion of mediocrity, and discrimination, and intolerance, and violence." At a Washington AI expo in 2024 he said that "the peace activists are war activists" and suggested protesters try an exchange program in North Korea. He has called "woke" thinking the central risk to Palantir and to the country. In 2025 he added a symmetrical demonology: a woke left, by which he meant the Mamdani wing, and a woke right, which believes that "everything is a conspiracy, any use of technology is actually going to only be used to eviscerate and attack us." Both, he said, were opposed to meritocracy.
Here is where the dissertation returns. Its subject was the connection between jargon, aggression, and culture — the way communities develop private vocabularies that mark insiders, and the way those vocabularies discharge hostility outward. "Pagan religion." "Woke left." "Warrior culture." It would be glib to say Karp became the phenomenon he studied. It would not be entirely wrong.
The fullest statement of the late Karp arrived in February 2025, in a book he wrote with Nicholas Zamiska, a Palantir executive and his longtime collaborator: The Technological Republic: Hard Power, Soft Belief, and the Future of the West. It argues that Silicon Valley grew complacent building consumer distractions while the West's ambition withered, and that software companies must partner with the state to win the artificial-intelligence arms race. In a letter to investors Karp had already quoted Samuel Huntington's The Clash of Civilizations to the effect that the West rose not through the superiority of its ideas or religion but through "its superiority in applying organized violence," a fact Westerners forget and non-Westerners never do. He told Maureen Dowd of The New York Times that the goal was to "scare the crap out of your adversaries," and asked: "Do the Chinese, Russians and Persians think we're strong?" Dowd found the interview "brim[ming] with American chauvinism." He predicted a three-front war against China, Russia, and Iran, argued for autonomous weapons, and said he favored conscription. Mark O'Connell, reviewing the book in The New York Review of Books, concluded that Karp's self-presentation as an unconventional thinker was "ultimately in service of a consolidation of state power and business interests."
The book also contained a reversal that almost no one in America noticed and almost everyone in Germany did. Decades after Karp, the doctoral student, had attacked Martin Walser for calling Holocaust remembrance a moral cudgel, The Technological Republic embraced Walser's idea, describing it as representing "the forbidden desires and feelings of a nation." In April 2026, Palantir published a manifesto distilled from the book. One of its summarized arguments read: "The postwar neutering of Germany and Japan must be undone." Another: "No other country in the history of the world has advanced progressive values more than this one." Critics in several countries called it supremacist. Karp, by then, had heard worse.
— Alex Karp, Reagan National Defense Forum, December 2023Somehow the corporate elite of this country thinks when it's time to make money, you stand up, and when it's time to stand up, you go play golf. And we've got to change that. That's our fault.
There is a coda to the politics, and it is a strange one. Palantir has become, among other things, a landing place for former socialists. Laura Rudas, its executive vice president of strategy, was once secretary-general of Austria's Social Democrats and a member of parliament; she left for Palantir despite having just won reelection. Julie Martinez, the company's global data protection officer, was appointed spokesperson of France's Socialist Party in September 2025 while still on Palantir's payroll. In September 2026, Tom Watson, who had served as deputy leader of the British Labour Party under Jeremy Corbyn, joined as a senior vice president. It is as if the company had become a sanctuary for a particular European type — left-wing by formation, statist by conviction, disillusioned by its own side — that its chief executive recognizes, perhaps, from the mirror.
Burning the Short Sellers
The refusal to go public lasted seven years. On September 30, 2020, Palantir listed on the New York Stock Exchange — not through a conventional IPO but a direct listing, which allowed existing shareholders to sell without new money being raised and without bankers setting the price. It advertised the ambition of becoming the "default operating system across the US." The listing came with a governance structure that made the old refusal to grant board seats look timid. A class of 1,005,000 Class F shares, held in a Founder Voting Trust and owned equally by Thiel, Karp, and Cohen, guaranteed the three men a 49.99 percent voting interest regardless of how many other shares they owned or how many the company issued. Insiders' Class B shares carried ten votes each; public Class A shares carried one. When critics accused the founders of wanting to be "emperors for life," they replied that the structure was legal and had been approved by a majority of investors. Both things were true.
That first year The New York Times ranked Karp the highest-paid chief executive of any public company, at $1.1 billion. The company was still losing money — $1.166 billion in net losses in 2020 alone, on $1.093 billion of revenue. In February 2023 it reported its first quarter of positive GAAP net income: $31 million. Twenty years. A generation of analysts had come and gone waiting.
What came next was a vindication so steep it began to look like a different kind of problem. In April 2023 the company launched its Artificial Intelligence Platform, which plugged large language models into customers' private networks and let them build "agents" against a digital model of their operations; Karp promised that in military use the AI would not carry out targeting on its own, but would require human oversight — intelligence augmentation, still. Commercial revenue surged. In September 2024, S&P announced Palantir's inclusion in the S&P 500; the shares rose 14 percent the next trading day. In November the listing moved to the Nasdaq. Revenue for 2024 reached $2.87 billion, with $462 million in net income. In the fourth quarter of 2025, revenue hit $1.4 billion, up 70 percent year over year, with U.S. commercial revenue up 137 percent to $507 million. On July 31, 2025, the Army — the same Army Karp had sued — consolidated 75 separate contracts into a single Enterprise Service Agreement worth up to $10 billion over ten years.
In 2024, Karp's "compensation actually paid" came to almost $6.8 billion, the highest of any CEO of a public American company. He sold $1.9 billion of Palantir stock that year and was criticized for it. He had a theory about who was doing the criticizing. He said he loved "burning the short sellers," compared them to cocaine addicts, and said they "just love pulling down great American companies." In February 2025, promoting his book, he said: "I love the idea of getting a drone and having light fentanyl-laced urine spraying on analysts that tried to screw us." It is difficult to imagine any other chief executive of an S&P 100 company saying this. It is equally difficult to imagine it costing him anything, which is perhaps the point.
The numbers have a vertiginous quality. Fortune noted in 2025 that Palantir, worth more than $400 billion, would not even qualify for the Fortune 500 on revenue. Its headcount that August was about 4,100, and Karp wanted to cut it to 3,600. By November its shares were the most expensive in the S&P 500, at 85 times expected forward sales. In 2024 Forbes ranked Karp 1,143rd among the world's billionaires, at $2.9 billion; a year later he was, at moments, among the 200 richest people alive. The Institute on Taxation and Economic Policy reported that Palantir paid no U.S. federal income tax for 2025 on roughly $1.5 billion of domestic income, largely through immediate deductions for research and development. The same year, Palantir helped fund the demolition of the White House's East Wing and the planned construction of a ballroom. In February 2026, the company announced it was moving its headquarters again, from Denver to Miami. Business Insider reported that an entity tied to Karp had bought a $46 million Miami mansion the previous June, months before the move was announced.
The philosopher who had wanted $250,000 to read in Berlin had become a person for whom $250,000 is a rounding error on a rounding error. The interesting question is whether he noticed the change. He has given one answer to it, about cars, which we will come to.
The Continent That Won't Quite Buy
If America made Palantir rich, Europe is where Karp seems to want to be understood, and where he is most consistently refused.
He built the network deliberately. He sat on the board of the publisher Axel Springer from 2018 to 2020 and later on its shareholders committee; he sat on the board of the chemical giant BASF until 2020. He works through the Atlantik-Brücke, the transatlantic friendship society. He sits on the steering committee of the Bilderberg Meeting and appears at the Munich Security Conference. In Leipzig he helped sponsor an exhibition called Dimensions: Digital Art since 1859 and said at its opening: "We can't leave the entire future to the Americans." The man who said it is an American.
In Germany, the reception has been schizophrenic in the clinical sense — two systems that cannot hear each other. The state of Hesse, Thiel's birthplace and the site of Karp's doctorate, became in 2017 the first German state to deploy Palantir's software for its police, as Hessendata. Bavaria followed with a version called VeRA and lobbied for nationwide adoption. In 2023, Interior Minister Nancy Faeser stopped federal use; the same year a German court ruled the software's legal basis unconstitutional, and the law was rewritten rather than the software abandoned. A 2025 YouGov poll found 51 percent of Germans at least partly supportive of federal agencies using Palantir; a petition against it on Campact gathered more than 400,000 signatures. A 2024 documentary by Klaus Stern, Watching You: The World of Palantir and Alex Karp, examined his German network; Karp declined to participate. In September 2025 he told German leaders, in effect, to stop being German — to "talk about the crisis," empower founders rather than the political class, carve out special technology zones with laxer labor rules, build on industrial strengths in cars and chemicals and machinery, and confront a pacifism he called the country's Achilles' heel. He said he saw Palantir as a continuation of Germany's cultural legacy and virtues. Michel Friedman's old neighbor had graduated from correcting one man's party to correcting a nation's.
Britain bought more readily. Palantir's UK operation is run by Louis Mosley, grandson of Oswald Mosley, the leader of the British Union of Fascists — a man whose surname is itself a verdict rendered by others, in a country that renders such verdicts with exceptional memory. Mosley was quoted internally describing Palantir's strategy for entering the National Health Service as "Buy our way in," acquiring smaller companies with existing NHS relationships in order to "take down a lot of political resistance." In November 2023 NHS England awarded Palantir a £330 million contract for its Federated Data Platform; the contract was first published with more than 70 percent of its pages redacted. In December 2025, the Ministry of Defence gave Palantir a £240 million direct award, with no other bidders considered, on the grounds that switching suppliers would be technically difficult and no one else could do the job.
Elsewhere, the doors began closing. In May 2026, London's mayor Sadiq Khan blocked a Metropolitan Police contract worth up to £50 million, citing ethics and value for money. That same month the Bundeswehr said it would not award Palantir national contracts for now; the head of its cyber command said it was "currently inconceivable" to give industry access to the national database. In June 2026, after the U.S. government abruptly cut off access to Anthropic's newest models, France announced that its domestic intelligence service would replace Palantir with a French company, ChapsVision, citing strategic dependence. Spain quietly told state-linked companies to avoid it. Switzerland had rejected larger deals for fear that data might reach American intelligence.
And yet NATO itself acquired the Palantir Maven Smart System on March 25, 2025, and it reached full operational capability in June 2026. Admiral Pierre Vandier, the alliance's supreme allied commander for transformation, said there was currently no competitor, and that any European alternative would need to prove itself in months or years, not a decade. Europe's politicians say no; Europe's generals say there is no one else to say yes to. Karp, the American who writes in German, is caught in the middle of a continent's argument with itself about whether it can afford to belong to his West.
Too Poor, Then Too Rich
He lives, officially, in Lyman, New Hampshire, a town of a few hundred people in the White Mountains. In August 2021 the Concord Monitor reported that he had given $180,000 to an ousted hermit known as River Dave — the kind of gesture that is either deeply private or calculated, and which in Karp's case is probably both. He owns eleven houses around the world. On November 8, 2022, he moved his residence from Switzerland to Liechtenstein, a principality famous for admitting almost no one; in June 2026, Süddeutsche Zeitung reported he was building a house on a 6,000-square-meter plot in Schaan. Liechtenstein's officials declined to discuss it. The founder of the company named for the seeing stone has a talent for not being seen.
As of 2025 he was in long-term relationships with two women and had been described as "geographically monogamous." He has said the thought of having children "gives [him] hives." He speaks fluent German and passable French. He jogs, skis, and is known around Palantir for an intensity that its admirers call conviction and its detractors call performance. A small shelf of quick paperback biographies has appeared to explain him — ALEX KARP: The Philosopher CEO Who Built Palantir and THE MASTERMIND OF PALANTIR: The Life of Alex Karp among them — while the serious attempt, Michael Steinberger's The Philosopher in the Valley, arrived in November 2025 with a subtitle that included the phrase "the Rise of the Surveillance State," which suggests not every encounter between author and subject was warm.
He has never learned to drive. Asked why, he gave an answer that reads, in retrospect, like the most efficient autobiography any billionaire has offered: "I was too poor. And then I was too rich." There was no middle. There was the Philadelphia boy at the protests and the Frankfurt student with $12,000, and then there was the man with Austrian assistants and eleven houses, and in between there was nothing — no years of ordinary ownership, no license, no learning to steer.
Somebody else steers. That is how it has always worked. Thiel steered the money; the analysts at Langley steered the software for two years; the forward-deployed engineers steered the customers; the generals in Kyiv steered the targeting. Karp's genius, if that is the word, has been to sit in the place where all of that steering converges and to give it a vocabulary — a jargon, a culture, an aggression — that makes the people inside feel they belong to something that the people outside cannot understand.
Steinberger reports a detail that is easy to pass over. In every Palantir branch — in Denver and Miami and the Shire and Rivendell and the Grey Havens, in all the Tolkien-named buildings around the world — Karp has his own office. And when he is there, German is often spoken.
Part IIThe Playbook
What follows is an attempt to extract, from Karp's twenty-two years at Palantir, the operating principles that are genuinely transferable — and to be honest about which of them carry costs that an operator should weigh before adopting. Some are about structure and control. Some are about time. A few are about the uses and dangers of being disliked. None is about working hard, which Karp assumes and so will we.
The Principles
- 1.Treat alienation as a method, not a wound.
- 2.Never sell the board seat.
- 3.Let the user write the product before you sell it.
- 4.Choose the hard problem on purpose.
- 5.Pick a foil and paint over its wall.
- 6.Sue the monopoly buyer when the process is rigged.
- 7.Treat patience as a balance-sheet asset.
- 8.Go where the war is.
- 9.Take the contracts that cost you friends — and know the price.
- 10.Make the culture a filter, not a perk.
- 11.Manage mystique like debt.
- 12.Stay small enough to be indispensable.
Principle 1
Treat alienation as a method, not a wound
Karp felt different at "every institution I interacted with." The ordinary response to that feeling is to spend a career earning admission — to become the most fluent insider precisely because one began outside. Karp did the opposite. At Stanford Law he diagnosed his classmates as alienated laborers. In Frankfurt he told a CDU celebrity he was in the wrong party. In Palo Alto he defined Palantir against Facebook and, eventually, against the entire "engineering elite of Silicon Valley." The outsider's vantage became a standing analytical tool: a way of seeing what an institution takes for granted because one never took it for granted oneself.
The operational payoff is real. Palantir's core insight — that the CIA's and FBI's databases were siloed because each agency's culture treated its data as identity — is the kind of observation insiders struggle to make. The cost is that a posture of permanent estrangement can harden into a politics of permanent enemies, which is roughly the arc Part I describes.
Tactic: Write down the three assumptions your industry treats as too obvious to discuss, and build your first product around the one you most suspect is false.
Principle 2
Never sell the board seat
From its first rounds, Palantir refused to give investors board seats — including In-Q-Tel, the fund through which its most important early customer was investing. It kept refusing through every round, and at the direct listing in 2020 it formalized the position with Class F shares guaranteeing the three founders 49.99 percent of the vote regardless of dilution. Bloomberg's later assessment captured both edges: the structure lets management "stay the course on long-term plans" and resist pressure, but it also "insulates an increasingly powerful company from a measure of accountability."
For an operator, the lesson is not that every founder should seek permanent control. It is that control is cheapest to retain at the beginning and almost impossible to reclaim later. Palantir could survive twenty unprofitable years, sue the U.S. Army, and take politically radioactive contracts precisely because no investor with a board seat could force a different path.
Tactic: Before your first priced round, decide in writing which governance rights you will never trade, and treat any term sheet that touches them as a decline.
Principle 3
Let the user write the product before you sell it
For two years, according to The Wall Street Journal, Palantir revised its software continuously according to the demands of intelligence analysts. The company mocked the very idea of a sales force until 2019. In place of salespeople it sent forward-deployed engineers to live inside customers — 120 of them at JPMorgan in 2009. The model inverted the enterprise-software orthodoxy of building a product, then hiring a sales team to push it.
The method has a philosophical root that is easy to miss. Palantir's stated theory was "intelligence augmentation" — the conviction that human analysts, not autonomous machines, should remain at the center. A company that believes the human is the locus of judgment will naturally build by watching humans work. The JPMorgan example is also a warning: engineers embedded deeply enough in a customer will build whatever the customer asks for, including the surveillance of that customer's own employees.
Tactic: For your first major customer, embed an engineer on site for a full quarter with no quota other than shipping what the user actually needs, and keep a written log of every request you refuse and why.
Principle 4
Choose the hard problem on purpose
Thiel's longstanding complaint about Karp was that he always chose the hard, complex way when a simpler path might have sufficed. Thiel nonetheless left the company in Karp's hands. The complaint is the strategy. Integrating classified data across agencies, fighting the Army's procurement system, deploying software in an active war zone — each was harder than the alternatives, and each produced a position competitors could not easily replicate. By 2025 the UK Ministry of Defence was justifying a £240 million direct award on the grounds that no other supplier could provide a similar service, and NATO's transformation commander was saying flatly that Palantir had no competitor.
Difficulty, in this reading, is a moat dug in advance. The obvious risk is that a founder who prizes difficulty will sometimes choose it where it creates no advantage at all. The test is whether the hardness accrues to you or merely to your schedule.
Tactic: For each major strategic choice, ask whether the harder path creates switching costs for customers or only costs for you, and take it only in the first case.
Principle 5
Pick a foil and paint over its wall
When Palantir took over a former Facebook office in Palo Alto, it painted over a famous wall associated with the previous tenant. Karp criticized Google and Facebook for their attitude toward data protection. In 2020, moving to Denver, he declared that the engineering elite of Silicon Valley "do not know more about how society should be organized or what justice requires." Later the foil became "woke" culture and elite universities. Each foil did the same work: it told employees and customers what Palantir was by vividly showing what it was not.
A foil is cheaper than a mission statement and more motivating. It also has a half-life. Karp's foils have grown progressively larger — a company, then a region, then a political culture — and each escalation narrows the pool of people who can work with you.
Tactic: Name one competitor or industry norm whose opposite you embody, make the contrast physical in your office or product, and revisit the choice every two years to ensure the foil hasn't outgrown your business.
⚔
Conventional Wisdom vs. the Karp Approach
Where Palantir's operating choices diverged most sharply from Silicon Valley defaults.
| Valley default | Karp's choice | Evidence |
|---|---|---|
| Give lead investors board seats. | Refuse all investor board seats. | Declined even In-Q-Tel; Class F shares at listing. |
| Hire sales early. | Embed engineers; mock sales. | No sales hiring push until 2019. |
| Never sue a prospective customer. | Sue the U.S. Army. | Won in 2016; Army contract by 2019. |
| Go public when the market is hot. | Wait, then list directly. | Refused IPO in 2013; direct listing in 2020. |
| Avoid controversial customers. | Take them openly. | Maven after Google; ICE; IDF partnership. |
| Stay in Silicon Valley. | Leave it, publicly. | Denver in 2020; Miami in 2026. |
Principle 6
Sue the monopoly buyer when the process is rigged
The U.S. Army had its own system, DCGS-A, and an acquisitions bureaucracy determined to protect it. When the Army Test and Evaluation Command produced a positive report on Palantir in 2012, it was ordered to modify it. When MITRE's evaluation came back mostly positive in 2013, the Army canceled it. Karp's response — a 2016 lawsuit arguing the Army had illegally discriminated against a commercial product — was nearly unthinkable for a vendor. Palantir won. After beating Raytheon head-to-head, it won the contract in 2019. Six years later the Army consolidated 75 contracts into a single agreement with Palantir worth up to $10 billion.
The principle generalizes beyond government. When a dominant buyer's process is structurally biased toward an incumbent, quiet persistence may never succeed. Litigation, in that circumstance, is not a breach of the relationship — it is the only route to having one. But it works only if the evidence is overwhelming and your product is genuinely better; Palantir had the soldiers' testimony from Afghanistan and two suppressed favorable evaluations.
Tactic: If a dominant buyer is excluding you through process rather than merit, assemble the documentary record of suppressed evaluations before you escalate, and escalate only when that record would persuade a neutral judge.
Principle 7
Treat patience as a balance-sheet asset
Palantir was unprofitable for twenty years. It lost $1.166 billion in 2020 alone. In 2013, at a $9 billion valuation, Karp said going public would make running the company "very difficult" and declined. The first GAAP-profitable quarter, $31 million, came in early 2023. Two years later the company was worth more than $400 billion.
What made the patience affordable was the combination of Principle 2 and a chairman with deep pockets: Thiel and his fund covered most of the initial $30 million, and founder control meant no investor could force an early exit. Patience is not a virtue a founder can simply decide to have. It is a structure built in advance — capital from people who share the time horizon, governance that protects it, and customers whose contracts last longer than the funding cycle.
Tactic: Match your capital sources to your product's true time-to-maturity, and decline money from anyone whose fund life is shorter than your realistic path to profit.
Principle 8
Go where the war is
In June 2022, Karp became the first CEO of a major Western company to meet Zelenskyy after the invasion. Palantir provided software and services pro bono at first, even paying for the personnel it hired inside Ukraine. Within a few years Ukraine was using Palantir for air defense, logistics, demining, refugee management, its education system, and the Brave1 Dataroom, where 100 companies trained more than 80 models on real battlefield data. Ukraine's defense minister called it an "effective IT weapon." An Austrian analyst called the country Palantir's laboratory.
The commercial logic is that the moment of maximum need is the moment of minimum competition, and that being present first at a crisis creates loyalties — and dependencies — that outlast it. Commentators described the arrangement as a "software-based guarantee" carrying strategic dependency for Ukraine. Operators should note both halves: the first mover at a crisis gains extraordinary learning and lock-in, and the customer eventually notices the lock-in.
Tactic: When your target customer faces an acute crisis, show up early with your best people and a willingness to absorb initial cost, and document every lesson the crisis teaches your product.
Principle 9
Take the contracts that cost you friends — and know the price
Palantir took Project Maven after Google abandoned it. It renewed its ICE contract for $50 million in 2019 even after acknowledging that its earlier denials of involvement in deportations had been misleading. It signed a strategic partnership with the IDF in January 2024 and lost employees over Karp's stance. Each choice traded reputation in one constituency for durable revenue and loyalty in another.
The honest accounting matters here. The HBGary episode in 2010 shows Karp apologizing specifically and severing ties when a proposed engagement crossed a line he recognized. The ICE episode shows the reverse: misleading denials, followed by acknowledgment, followed by renewal. The second path cost Palantir credibility it has never fully recovered. The principle is not "take every contract." It is that if you take the controversial contract, own it from the first day; the denials do more damage than the deal.
◷
Decision Timeline
The pivotal calls that shaped Palantir under Karp.
2003
Thiel incorporates Palantir on May 6; Karp's Frankfurt dissertation is catalogued sixteen days later.
2004
Karp hired as CEO; Thiel funds most of the ~$30M build; investors denied board seats.
2011
Karp apologizes to Greenwald and "progressive organizations" after the HBGary/WikiLeaks plan surfaces.
2013
At a $9B valuation, Karp refuses to pursue an IPO.
2016
Palantir sues the U.S. Army and wins.
2019
Takes over Project Maven from Google; renews ICE contract for $50M.
2020
Moves HQ to Denver; direct listing on the NYSE with Class F founder shares.
2022
Karp meets Zelenskyy in Kyiv; Palantir works pro bono at first.
2023
First GAAP-profitable quarter; AIP launched with human-oversight pledge.
2025
Army consolidates 75 contracts into a $10B agreement; The Technological Republic published.
2026
HQ moves to Miami; France and London's mayor reject Palantir while NATO's Maven system reaches full capability.
Tactic: Before signing any contract you expect to be controversial, draft the public statement you would issue if it leaked tomorrow — and if you cannot write it without evasion, don't sign.
Principle 10
Make the culture a filter, not a perk
Karp has described Palantir's "warrior culture," which values excellence over money. In 2025 its fellowship program gave participants a four-week seminar on Western civilization and whether it is worth defending. After October 7, Karp's public positions drove some employees out — a cost he acknowledged openly and seemingly accepted as a feature. The company recruited heavily among former military officers and former politicians, including, strikingly, a number of disillusioned European socialists.
The point is not the specific values; it is that Palantir's culture functions as a sorting mechanism. People who would be uncomfortable doing intelligence and defense work leave early or never apply, which reduces the internal friction that convulsed Google over Maven. A culture that repels the wrong people is more valuable than one that attracts everyone.
Tactic: Identify the single kind of work your company will do that some talented people would refuse, and state it explicitly in recruiting so the filtering happens before the offer, not after the walkout.
Principle 11
Manage mystique like debt
Palantir's founders cultivated an aura — the bin Laden legend, the CIA connection, the silence after Bloomberg's "Palantir Knows Everything About You" — that helped valuations and recruitment. The aura also had a bill. The CIA was irritated enough by the publicity to try canceling its contract. Conspiracy theories about a central database that Palantir does not, in fact, maintain have fed opposition in Germany, Denmark, and Switzerland. The European argued that Thiel and Karp caused themselves trouble by cultivating a mysterious aura that provoked backlash against software useful in both civilian and defense settings.
Mystique behaves like leverage: it amplifies returns on the way up and amplifies losses when the market — or a customer, or a government — turns. The founder's job is to know how much of it is on the balance sheet.
Tactic: List the three most flattering exaggerations circulating about your company and publicly correct at least one before a customer does it for you.
⚖
Europe's Split Verdict
How European institutions have responded to Palantir, as reported through 2026.
| Institution | Decision | Status |
|---|---|---|
| NATO | Acquired Maven Smart System, March 2025; full capability June 2026 | Adopted |
| UK Ministry of Defence | £240M direct award, December 2025 | Adopted |
| NHS England | £330M Federated Data Platform, contract heavily redacted | Contested |
| Hesse and Bavaria police | Hessendata since 2017; VeRA pushed nationally | Contested |
| Bundeswehr (national) | Declined national contracts, May 2026 | Rejected |
| London Metropolitan Police | £50M contract blocked by Sadiq Khan, May 2026 | Rejected |
| France (DGSI) | Replacing Palantir with ChapsVision, June 2026 | Rejected |
Principle 12
Stay small enough to be indispensable
In 2025 Palantir was among the most valuable companies on earth and too small, by revenue, to make the Fortune 500. It had about 4,100 employees in August of that year, and Karp wanted to cut to 3,600. This is not an accident of growth but a model: a small number of deeply embedded engineers, a software platform that becomes the integration layer between a customer's existing systems, and contracts justified — as the UK MoD put it — on the grounds that switching would be too difficult.
The principle is that leverage comes from position in the customer's architecture, not from headcount. The risk, visible in France's June 2026 decision citing "strategic dependence," is that indispensability eventually becomes the customer's chief complaint. The company that cannot be replaced is the company the customer most wants to replace.
Tactic: Measure your strategic position by how many of the customer's other systems route through yours, not by seats sold — and invest in interoperability before the customer starts calling your indispensability a dependency.
Free playbook
Get The Business Model Playbook
58 business models, one visual page each: how the money flows, the metrics that matter, and who runs it. Free when you join the Faster Than Normal email.
Free. No spam. Unsubscribe anytime.
Part IIIQuotes and Maxims
In their words
— Alex Karp, explaining why he never learned to driveI was too poor. And then I was too rich.
— Alex Karp, to The New York Times, 2024Are we tough enough to scare our adversaries so we don't go to war? Do the Chinese, Russians and Persians think we're strong?
— Alex Karp, on immigration and European politicsYou have an open border, you get the far right. And once you get them, you can't get rid of them.
— Alex Karp, on Palantir's position at the FBI, as reported by BuzzFeed News in 2017They'll have no choice.
— Alex Karp, on Palantir's worldview, to The New York Times, 2024a consistently pro-Western view, that the West has a superior way of living and organizing itself
Maxims
- Feeling like an outsider is data. Use the dissonance you feel inside an institution to see what its members cannot.
- Control is cheapest at the beginning. The board seat you refuse in year one is the lawsuit you can file in year twelve.
- Build where the user sits. Two years of revising for analysts beat any sales deck Palantir could have written.
- Hard is a moat only if it's hard for them. Choose difficulty that raises your customer's switching costs, not just your own.
- A foil clarifies faster than a mission. Paint over the other company's wall and everyone knows who you are.
- Patience must be engineered. Twenty unprofitable years are survivable only with aligned capital and protected governance.
- Arrive first at the crisis. The customer remembers who came when no one else would.
- Own the controversial contract on day one. Denials cost more than the deal.
- Let culture do the hiring. A company that repels the wrong people avoids the walkout later.
- Indispensability becomes dependency. The day a customer can't leave is the day it starts planning to.
