AI News

Palantir CEO Karp: Billionaires cannot get 100 times richer in the AI era

HuxiuUpdated Nov 15 12:51

AI summary

In a podcast interview on "The Builders," Palantir CEO Alex Karp opposed directly levying a wealth tax, arguing that it would not help the poor but would instead force the government to use AI to assess private property and expand the surveillance state. He acknowledged that AI will exacerbate the wealth gap, with wealth in the tech circle potentially rising 10 to 50 times, and advocated "trading reduced waste for higher taxes." He also criticized large US companies for being angry at large model companies such as Anthropic, because they pay for services yet contribute proprietary data to train the models.

Body · AI translation

**Key Points**

**1. The traditional software playbook is a low-value addiction model**

Karp says the standard model is to make low-value but highly addictive products and push them through aggressive sales, like an "STD"; Palantir did the opposite, spending about three years with zero revenue to hone real value.

**2. Outsider status and dyslexia turn out to be advantages**

Karp believes dyslexia kept his brain from being "colonized" by mainstream language, letting him confront problems directly; top builders like Jobs and Musk all share outsider identities and a humanistic aesthetic.

**3. A wealth tax would expand the AI surveillance state**

Karp acknowledges AI will worsen wealth inequality, with tech fortunes possibly rising 10 to 50 times; but a wealth tax would require AI to assess private property, so he advocates "less waste in exchange for more tax."

**4. Large models trigger a corporate data-sovereignty backlash**

Companies pay for models yet contribute proprietary data to train them, and may even be turned into competitors, hence their anger at Anthropic and others; Palantir uses an "ontology" as a security layer to protect data.

**5. Military AI must be isolated from domestic surveillance**

This article comes from the WeChat public account: AGI 2050, author: AGI 2050, original title: "Palantir CEO In-Depth Interview: Billionaires Cannot Get 100 Times Richer in the AI Era," header image from: Visual China.

If OpenAI is Silicon Valley's visible AI upstart, then Palantir is one of the world's most powerful and most controversial data companies, serving the US government and intelligence agencies. After 9/11, "Silicon Valley venture capital godfather" and core PayPal Mafia figure Peter Thiel had an idea: could PayPal's methods for identifying fraud patterns be used to help intelligence agencies find threats in massive datasets? Palantir was born from this. Today, from thwarting hundreds of terrorist attacks (many of which would otherwise have occurred in Europe) and dismantling organized professional criminal groups to data applications in Ukraine's defense and on the Israeli battlefield, this hundred-billion-dollar-market-cap giant stands behind them all.

Yet the man at the helm of one of the world's most controversial tech companies is a complete "Silicon Valley outsider"—Alex Karp. He is not an engineer; he comes from a philosophy background, earned a doctorate at Goethe University Frankfurt in Germany, leans left politically, is deeply wary of totalitarianism, and even has dyslexia.

In this episode of the podcast The Builders, master biographer Walter Isaacson and Alex Karp engaged in a deeply thought-provoking exchange. Karp not only rarely dissected his family of origin, Heidegger's philosophy and the historical lessons of Nazi Germany, but also launched a full-throated critique of the hypocrisy of today's Silicon Valley tech giants: Why does the traditional software startup playbook produce products that are like an "STD"—of little value yet hard to shake off? Why are large American companies furious at large-model companies like Anthropic? When AI may make people in tech 10, 20 or even 50 times richer, while ordinary people's wages, even if they rise at twice the rate of inflation, are likely unable to keep up, why would levying a "wealth tax" instead push the whole society into a terrifying surveillance abyss?

(This content comes from The Builders podcast, aired November 10, 2026; the original video is 43 minutes long. Below is selected dialogue content, about 14,000 characters in total, with a reading time of about 20 minutes.)

Highlights from this episode:

The software startup playbook taught in schools is like an "STD": make a simple product of little value but addictive, hire a bunch of salespeople to push it hard, easy to catch and extremely hard to get rid of. We violated every software development convention, spending about three years with zero revenue, stubbornly grinding away at a truly valuable product.

The benefit of having dyslexia is that your brain has not been "colonized" by the language system of mainstream culture. Facing any technical problem, I face the problem as it actually is.

Top builders like Jobs and Musk are without exception "outsiders," and they combine humanistic aesthetic sensibility with technical instinct. Silicon Valley can produce great companies while Germany and France cannot, not because we are smarter, but because aesthetics, depth of understanding of the world and business intuition matter more than pure technology.

Young people don't want the government to know everything about them, and I agree; but when they think it doesn't matter if companies like Google know, they are being naive. They severely underestimate how tech products influence their thoughts—if commercial giants control all the data, they can control or greatly influence your thinking.

I firmly oppose directly levying a "wealth tax." You cannot let people who are already billionaires get 100 times richer while ordinary people get only 5% richer; but forcibly collecting a wealth tax not only fails to help the poor, it also forces the government to use AI to monitor the value of every chair in your home, further expanding the surveillance state.

Right now large American companies are absolutely furious at large-model companies like Anthropic. Companies spend heavily to buy AI services, only to find the product creates little value, they are paying, and their proprietary data is being used to train models. They ask: if you talk about ethics, how can you use my data to make a product that competes with me? I often have to explain on Dario's behalf that he is not the villain they imagine.

Silicon Valley is an extremely isolated island, born from a system built jointly by the military, government, universities and venture capital, yet today some people there are unwilling to participate in national defense. Even more absurd is that the bosses of AI labs, while their own fortunes soar, condescendingly tell ordinary people: "Your job is going away, your life is going to be ruined, from now on rely on my handouts."

My family of origin and dyslexia: a brain not fully "colonized" by the mainstream

Q: When writing biographies, I have a theory: a person's fundamental character often all comes from his father. Tell me about your father first.

Alex Karp: My father was a very smart man. His mother was a German-born Jew who died very early, and he was raised single-handedly by my grandfather; that experience shaped him profoundly. I think my parents were very much in love back then. But to be honest, it was no accident that my father married my mother—my mother was a Black woman so brilliant it was suffocating, but she also genuinely had a somewhat difficult personality. I think my father especially enjoyed this, because it stung his stepmother, who cared more about status. My mother's brilliance had an extremely captivating charm.

Put it this way: if my parents had lived out the lives I think they could have had, they never would have produced someone like me. I often feel a certain unease about one thing: it was precisely a family that ultimately fell apart in a messy divorce, and a pair of parents with such complicated personalities, that happened to forge a "special product" like me—someone naturally suited to building a company like Palantir. When you throw someone like me into a chaotic environment, making you feel from childhood that you can only rely on yourself, and even subconsciously or semi-consciously feel that you do not fit in with your parents, you develop a very deep confidence in your ability to survive, but you may also develop a latent distrust of external systems and structures.

Q: How did having dyslexia, in turn, give you wings to take off?

Alex Karp: You have to work backward from where you are now to see how it made you. Actually, until about six years ago, I kept the fact that I had dyslexia hidden, as if I were in the closet. Later I was just too tired and felt too frustrated, so I confessed to a good friend. But looking back now is interesting: the people who back then mocked everything we did and every word I said are now basically copying our model word for word. Why? Because they are always looking for a ready-made "success script."

Dyslexia is a very broad term. By my own definition, it means there is a weakened, non-one-to-one relationship between you and the classification system of written language. Even now, when I already have a PhD and even personally draft many documents inside Palantir, my mechanism for reading is still different from yours; my brain needs an extra "translation conversion" step.

Why does this matter? Of course, if given the choice, no one would want dyslexia. But the fact is, precisely because you have dyslexia, your brain has not been thoroughly "colonized" by mainstream culture. So whether facing mainstream culture, technical problems, or any complex predicament, I engage directly and nakedly with the problem itself, rather than looking at it through a linguistic framework instilled by others.

The madman with zero revenue for three years: breaking the software industry's "STD model"

Q: How did you manage not to be "colonized" by mainstream culture, and use that to build Palantir's counterterrorism product?

Alex Karp: Think about how we built that counterterrorism software back then. That product changed the course of European history. We thwarted hundreds of terrorist attacks, some of which would otherwise have happened elsewhere in the world, but many of which would otherwise have happened in Europe. Think about it: if those attacks had actually happened, what would Europe look like today? Later, our product also became an absolute cornerstone of Ukraine's national defense.

If you go to school to learn how to develop software and build a software company, the standard textbook will teach you: first make a simple product that actually has little value to customers but gets them addicted; then hire a large sales force and make many customers pay against their will. That is how traditional software companies are built. The derivative of this model is a simple but addictive thing—it creates no real value, but once you start using it, it is hard to shake off. This is exactly like an "STD": extremely easy to catch, but possibly very hard to get rid of completely. And in this standard process, you must avoid anything that looks like "grunt work" at all costs.

But what did we do? We spent a full three years polishing a product with real core value. So here comes the question: what is "value"? Our definition at the time was: you must both thwart terrorist attacks and at the same time protect civil liberties. Good God, with just that one goal, you offend two major camps at once: one camp wants to stop terrorist attacks, and the other wants to protect civil liberties.

So to build this product, you have to violate every unwritten rule of the software development field. And the biggest iron rule we violated was: for three full years, we basically had zero revenue. Three whole years, not a cent of revenue! You have to be completely insane to do such a thing. I do believe that at the time there had to be a neurodiverse person (meaning someone whose neural mechanisms differ from the norm) at the helm to withstand the pressure and fight this through to the end. Many business executives often ask me privately: "Karp, can you teach me how to be unafraid of making enemies and unafraid of being unpopular, like you?"

The outsider's superpower and Heidegger's political warning

Q: Clearly, you felt like an outsider from childhood. I have another theory about biography: being an outsider is actually a superpower, because you do not need to expend effort trying to please others. Did you ever try hard to fit in?

Alex Karp: No. But I think that human beings, as a species, are instinctively eager to belong. Most of us who do not try to fit in have actually understood, deep down, for a long time: even if we tried, we would be doomed to fail.

Q: How much has this outsider mentality helped you? By contrast, people like me—as a Jew in New Orleans—have always felt that I fit in quite well and have always wanted to get along with everyone.

Alex Karp: I don't think you're really just an insider. If you were a pure "insider," you could never have written those profound biographies of yours. Although I'm not the one interviewing you today, if I were interviewing you, I would certainly push back on the way you just described yourself.

Back to myself: without standing in the position of an outsider, it would have been almost impossible for Palantir to come into being, and it would have been very hard for my life trajectory to take the shape it did. It is almost impossible to do all this by following the "internal elite path." And I think that the traditional "internal elite track" has now been trampled into ruin and has completely failed. Look at the corporate culture, aesthetic orientation, and product form we built at Palantir—there is not a single thing that a traditional "insider" would do.

What is even more interesting is that precisely because being an "outsider" is now so valuable, countless "fake outsiders" who pretend to be outsiders have sprung up. And what about a real outsider like me? To be honest, I dream of giving myself an "insider vacation"—even if I were given eight weeks a year to experience what it is like to be a mainstream insider! I know that idea is utterly stupid. I even have a fantasy about reincarnation: in my next life, I would especially like to be reborn in Bavaria, Germany, as a Catholic veterinarian.

Q: Since we have brought up Bavaria, this leads to another related concept in human nature—the "instinct to collaborate with the enemy and submit." When you were a graduate student in Germany, your doctoral dissertation was about the Nazis, right?

Alex Karp: What I wrote at the time was a critique of the philosopher Heidegger. Whether you like him or not, it is hard to deny that Heidegger is one of the two or three most important philosophers of the past two hundred years or so, perhaps on a par with Wittgenstein. Heidegger developed a "phenomenology," which in plain terms is an elaborate theory explaining how all things in the world are endowed with value in relation to the world. His mentor Husserl was Jewish, and his mistress Hannah Arendt was also Jewish, yet Heidegger himself became an ardent supporter of Hitler.

I believe that Heidegger's philosophy is the most important political philosophy in the Western world today—unfortunately, in the wrong direction. Because most of the political movements in the West today that I call "regressive"—whether the far right (trying to retreat to some old order that never really worked in history) or the far left, which on the surface pretends to believe in Marxism but whose underlying logic is entirely Heideggerian.

This is also why I have said for a year that I am still an "unhappy Democrat," and why I warned the Democratic Party early on: there is a faction in your party that is not a "progressive" faction at all. They fly the banner of progressives, but what they want in their bones is a kind of communitarian economy akin to feudalism—that is not even socialism. And they pretend to be isolationists, when in fact they are simply singling out certain groups that make them uncomfortable and attacking them. This is fundamentally a right-wing party!

Anyway, I did indeed study Heidegger in depth back then. But another honest truth is that I also wanted to "break out of prison" from academia—halfway through my doctoral dissertation I understood completely that academia was not the right meal for me, and that I had to get out and actually build something. To use the name of your podcast, I am a "builder" to my core.

There is a very subtle common thread among builders. Look at the people you have written about—Steve Jobs, Elon Musk—and you will find that every case is a combination of "outsider identity" and "humanistic aesthetic inclination"; their humanistic attributes are in no way inferior to their technical attributes. Although most people in the tech world would not agree, I objectively believe that aesthetic intuition about what matters to the world, and business acumen, are far more important than technology itself. That is why great tech companies are born in the United States and, unfortunately, not in Germany or France. It is not because we Americans are smarter than they are.

Peter Thiel's invitation

Q: So why did you found Palantir back then? What was the original motivation?

Alex Karp: From my personal perspective, I founded Palantir because I met Peter Thiel when I was studying at Stanford Law School. I loafed around all day at law school and did nothing serious; Peter was much more diligent than I was. But we were in the same class group and lived in the same dormitory building—the dormitory specifically for poor students with little money. And both of us had extremely strong intellectual curiosity.

Then one day, he came to me and said: "We have an idea. We want to take a methodology we used at PayPal and move it into the intelligence field. Because this set of things has proven very effective in combating online financial fraud, and intelligence agencies are obviously extremely lacking in this kind of technical capability." When I heard it, I thought this was simply too interesting, and Peter himself was extremely persuasive and charismatic, so I agreed on the spot.

The reason I joined also has to do with my view of the American national project. Admittedly, you can view America through the lens of a blindly xenophobic chauvinism, but of course we have made many mistakes, some historical mistakes that are even hard to understand in hindsight today—for example, our failures in racial integration and mutual respect, and of course slavery, as well as our insufficient attention to social class and our failure to build a tighter national community; the list of mistakes can be very long. But in my view, and also in the view of most Americans, there has never been a national project on the scale of America. You simply cannot compare 1970s Sweden to America, right? Those are two completely different things; Sweden at the time was a highly homogeneous, monocultural society.

Google knows everything about you: Is Palantir engaged in mass surveillance?

Q: When many people hear Palantir mentioned, what comes to mind is cameras on every street collecting our information. Doesn't that violate our right to privacy? Or do we place too much importance on privacy?

Alex Karp: No. First of all, I think Europe gets one thing very right: Europe's constitutional order strictly protects personal privacy in both the commercial sphere and the government sphere. In the United States, by contrast, our Constitution mainly guards against "government overreach," but at the cultural level we are far less sensitive to "overreach by corporate giants."

Q: That's interesting. When I teach at university, all my students say they don't mind at all if Google knows everything about them, but they absolutely do not want the government to know everything about them. That mindset is strange.

Alex Karp: Exactly. I strongly support their vigilance against the government knowing everything about them, but they severely underestimate the danger on the other side. The logic in their subconscious is: "I don't care if a commercial company knows my data, because whether to use it is my own free choice." That is of course true—it really is their own choice—but they mistakenly believe that the views in their heads are entirely the product of independent thinking. In reality, a person's views are at best the product of an ongoing dialogue between "your own thoughts" and "the external mainstream consensus."

Young people today severely underestimate how much the technology products they interact with every day influence their thinking. Precisely because they underestimate this, they do not understand why they should be more vigilant about privacy issues in the commercial sphere. In fact, among the outsiders who criticize Palantir, there are some whose concerns I deeply sympathize with—it's just that they are aiming at the wrong target and have Palantir completely backwards.

You often hear people shouting clichés like "Palantir is building a surveillance state." But whether from the technical foundations or actual operations, what Palantir does is precisely the opposite of what they imagine, and our strictness in protecting privacy permissions exceeds that of any tech company on earth. But those critics are right about one thing: if a company really did control all the data, it really could control your thinking, or profoundly shape your cognition.

Q: But isn't Palantir part of the whole "surveillance culture"? This culture certainly has the benefit of maintaining security, but could the cost be the loss of personal autonomy?

Alex Karp: That of course depends on how you define the word "surveillance." If you are an enemy of the United States, firing on the U.S. military, or planning an attack on America, or if someone enters your country and wants to destroy the evil things you are doing, and you are obstructing them—although many people disagree, I freely admit it: yes, our products are used against those people.

Q: What if I'm just walking down the street in New Orleans and your system records in the background, "Look, Walter has walked over here now"?

Alex Karp: Well, domestic surveillance really is the problem. But strangely enough, Palantir is used deeply in France and some other parts of Europe (due to confidentiality contract restrictions, I can't name some places). There is an extremely core distinction here: "legally authorized surveillance" and "surveillance without legal authorization" are two completely different things. This distinction is crucial politically, morally, and technically.

The political and moral reasoning is obvious to everyone, but technically, building such a system is extremely difficult. For example, under the European legal environment, your software must ensure that: the system can retain relevant data only when a court has issued a search warrant; and once the warrant expires or is revoked, the system must be able to completely erase the data. The hardest technical pain point is: at the very second the warrant disappears, how should the relevant data be handled? Our products enable agencies to strictly follow this legal procedure.

Of course, some people will still say: "That is still surveillance—you are still surveilling a person the court has determined to be a criminal." This brings us to the extreme tug-of-war between the left and the right: the extreme caricatured position of the right is "who cares, just kill all the terrorists and criminals"—that is not the society most of us want to live in; and the extreme position of the left is actually equally counterproductive, as they shout, "No, even if we know they are hardened criminals, we must never surveil them!"

In fact, the place where our products exert the greatest power in Europe is in combating organized professional criminal groups, assisting border enforcement (because smuggling people into the country is entirely operated by professional criminals), and counterterrorism cases. When it comes to domestic intelligence, there must be a legal basis akin to "probable cause" and authorization under a judicial framework before you can lawfully access data. In practice, this is a mode of law enforcement that 90% of the public would wholeheartedly support. As for the remaining 10% who oppose even this kind of lawful law enforcement, I can only say that their naivety will ultimately push all of us toward some form of fascism. Just imagine: if all the terrorist attacks we foiled in Europe had actually occurred, today's Europe would long ago have been taken over by far-left or far-right fascism—the two are essentially birds of a feather—and by then, everyone would be under round-the-clock, no-blind-spot surveillance.

Q: Let us zoom out to a broader, more philosophical perspective, and not talk only about Palantir. Today's technology has an extremely vast data-collection capability: cameras on every street, visual tracking, facial recognition. Do you think this data explosion, combined with AI, will to some extent erode human individual autonomy?

Alex Karp: Without question, if you do not actively resist and regulate it, it not only could but inevitably will erode human autonomy. This brings us to the current great AI debate. In my view, AI is best understood as a kind of "raw material" that can be processed into a tool for good or forged into a weapon for evil.

One of the worst ways to abuse this powerful force is to use it for surveillance. I believe the vast majority of such abuse will occur in the commercial consumer sphere rather than the government sphere; but the risk does indeed exist in the government sphere as well. Therefore, I strongly support strict legislative restrictions on certain technologies.

Take a concrete example: for military-grade AI technology like what we use on the battlefield, the state should legislate outright to prevent such technology from being used domestically. On the battlefield, we must use it; we have no choice, because our adversaries are devoting all their efforts to developing these things; if we do not build them, we will be utterly crushed on the battlefield. Based on my observation of all of human history, if you are a true pacifist, you must possess an incomparably powerful military. The idea that peace can be bought by weakening one's own military power is either the fantasy of half-witted fools or something put forward by those who truly crave war.

Alex Karp: I believe that if the United States does not remain highly vigilant, we will ultimately be seriously affected by extremist forces at both the left and right ends. Take the debate over the "wealth tax," for example.

American society does indeed currently suffer from severe class injustice, and this must be addressed; moreover, the advent of AI will further intensify this rift. Although I firmly believe that AI will create more jobs, or jobs of slightly higher quality, especially for blue-collar skilled workers and those in vocational education, and many people will also benefit through retraining, even if the wages of these ordinary workers could rise at twice the rate of inflation, they are unlikely to keep up with the pace at which wealth is expanding in the tech and AI circles—because the people working in AI may see their net worth soar 10 times, 20 times, or even 50 times!

Faced with this rift, what is the reaction of some people today? It is to force through a policy that "only hurts the rich but does not help the poor at all," and, for various reasons, such a policy will ultimately deal a heavy blow to the middle class as well.

Q: Wait a moment. Since you also acknowledge that AI will lead to more extreme wealth disparity—just like a return to the Gilded Age of the 1890s, when technological prosperity likewise produced a huge wealth gap—shouldn't we then inject more wealth into the public pool of funds to narrow the gap?

Alex Karp: That is exactly what I mean! You must take action, but the question is: what exactly should you do?

Under my plan, for example, set a 50% inheritance tax, and then tell people: you can pay 25% now, but on the condition that for every extra penny of tax the government collects, it must at the same time identify administrative waste within the government system equal to half that amount!

Why? Because the left has a very reasonable criticism, with which I completely agree, and most people agree as well: you absolutely must not let people who are already billionaires become 100 times richer while everyone else in society becomes only 5% richer. This criticism is entirely valid. But at the same time, those who understand the laws of economics have an equally reasonable rebuttal: you cannot help the poor simply by collecting taxes, because half of the money the government collects is wasted through inefficiency! So my plan is: fine, let us tie these two pain points together and solve them at the same time. This is the same logic as when we had to both catch terrorists and protect civil liberties.

What is even more frightening is that, regarding the "wealth tax," for some reason everyone keeps ignoring a fatal logical consequence: in order to enforce a wealth tax, the government must establish an all-pervasive surveillance system! Because to tax net worth, the tax authority must figure out exactly how much that bench in your home in New Orleans, Walter, is worth.

Q: Let us set aside the practical operational issues for now; after all, assessing the value of every item in every person's home sounds operationally unrealistic...

Alex Karp: Sorry, I must interrupt and rebut you on this point: in the AI era, it is not difficult at all operationally! This precisely confirms the surveillance concern you just raised. With AI, assessing every person's entire household possessions is not unrealistic; especially when combined with various data, AI can calculate every item you own.

Q: Since it is technically feasible, where exactly is the problem with taxing the existing wealth of the rich?

Alex Karp: There are three problems: first, it does not help the poor at all; second, it will completely tear apart social consensus and destroy America's culture of technological innovation, because it involves the issue of asset liquidity; third, its inevitable logical consequence is to extend the tentacles of government surveillance into every corner of your private life.

Q: Then give me a "Palantir-style" solution. Since the wealth gap in society is too large, we need to rebuild social consensus. Apart from improving government efficiency, how exactly should the pie be divided?

Alex Karp: The core is to implement an exchange-based tax increase plan of "trading waste for tax revenue." Apart from the plan I just mentioned, any tax increase other than a wealth tax is acceptable, but it must be tied to a hard condition: for every 1 unit of additional tax revenue the government raises, it must identify 0.5 units of fiscal waste. That way, far more will actually be transferred to the government.

Because you are actually solving two completely different problems at the same time: the first problem is "unequal wealth," and the second problem is "how to substantively improve the lives of the poor." These two are not the same thing at all! I strongly support rich people like me paying more taxes, but three preconditions must be met: first, it must not destroy America's tech industry, and a wealth tax will definitely destroy it; second, it must not expand the boundaries of the surveillance state, and a wealth tax absolutely will expand them; third, the money collected must truly be put to use to improve the lives of the working class, the upper-middle class, and those below them.

If you are obsessed with only one end of this, you are merely creating social division. And in the end, the ones who pay for that division are the people at the very bottom of society with the least voice. In fact, in America, if you are truly poor, you basically have no channel to make your voice heard—because those politicians who claim to represent your interests are really only interested in "destroying the rich," not in structurally helping the poor escape their predicament.

Large Models Drain Corporate Secrets: Why Are Customers Furious at Anthropic?

Q: Let us talk about AI training data and digital surveillance. Recently you have been very strongly calling for "digital sovereignty." Now those leading large language model companies are not only frantically draining everyone's data to train models, often without permission, but are simply building their commercial empires on our shoulders.

Alex Karp: It is not just ordinary people—American enterprise customers today are furious, hopping mad, nearly having strokes! Why? Because these companies have suddenly come to their senses: they spent astronomical amounts of money buying large language model services, only to find that in engineering practice this stuff has created very little actual value for the enterprise.

What is even more crushing for them is that companies have suddenly discovered: while paying real money for the service, they are actually working for free for the large model companies, training models for them! How are large models trained? By directly feeding customer enterprise data into the models. And in the business world, what data is most valuable? Of course, the enterprise's core proprietary data! If you use these large model products completely naked, without any security protection layer, this is objectively what will happen.

Q: That sounds a bit like the management consulting firms such as McKinsey in the early years...

Alex Karp: Exactly, but what large models do is equivalent to magnifying McKinsey's model by 5 million times, and replicating it automatically at scale! And in the end, this whole thing will reappear in your business, while the results become theirs.

Precisely because of this, many enterprise customers are especially angry at Anthropic now. Funny to say, I now feel that 10% of every day is spent doing psychological massage for executives of major companies. I have to explain to them: "Dario (founder and CEO of Anthropic) really is not the cartoon villain you imagine, he is not Darth Vader from Star Wars; their team really has a set of moral philosophy they believe in." Then the executives retort directly: "Since they advertise themselves as so moral and so ethical, how can they take my money while stealing my core data to build a competing product that directly fights me for business and destroys me?!"

Q: So how does Palantir help enterprises fight back against this data plunder?

Alex Karp: First, the core secret of the "data sovereignty" we advocate—and also the most obvious point, but the hardest for competitors to copy—is that our commercial interests are truly aligned with those of our customer partners.

A lot of our work now is customer education. Companies come to us and ask: "How should we protect ourselves? How should the contract terms be written? How should we safely call closed-source and open-source models?" We have published document after document, and not just for customers and partners.

At the technical foundation, customers use what we call an "ontology" system architecture. When we first proposed this concept, it was likewise mocked, and now everyone is talking about "ontology." How to understand "ontology" in plain terms? You can think of it as a "contextual framework." This is actually exactly the same job Walter you are doing interviewing me right now: I blurt out all kinds of fragmented ideas, and you are responsible for putting them into a logical framework so that readers find it both interesting and understandable.

In enterprise software, the "ontology" layer plays two roles at once: first, it is an "alpha generator," capable of processing the rough raw material of large models into real, deployable business value; second, it is a rigorous "security shield" that lets enterprises call on external large-model capabilities without ever having their core secrets siphoned away or getting burned.

But at a deeper level, the essence of "digital sovereignty" is our conviction that the world should contain "sovereign independent institutions" and "sovereign independent individuals." Whether facing corporate giants or government agencies, every person should enjoy the natural protections of the First and Fourth Amendments in the digital world. And at the level of national security—for example, the extremely powerful systems we built for the U.S. Department of War (that is, the Department of Defense), or the systems running on the battlefields of Ukraine and Israel—these institutions possess their own unique datasets, their own ways of understanding data, and their own tactical logic for execution on the battlefield. These absolutely core secrets should not be understood by any external party.

Silicon Valley on an Island: Cannot Do Without the Military Yet Rejects It, and Does Not Care About Ordinary Americans

Q: Following on from this topic, do you think more American tech companies should step forward proactively to serve America's military strength and national ambitions? When you see Anthropic or even Google step forward and say, "No, we will absolutely not participate in military projects," what do you think?

Alex Karp: Silicon Valley is a wondrous, unique place that the whole world envies—and envies with good reason. But at the same time, Silicon Valley is also a highly isolated island.

You should know that, in all of American society, the only institution that the vast majority of Americans regard as uncontroversially "noble and worthy of respect" is the U.S. military. And Silicon Valley's rejection of the military is downright hypocrisy—without the U.S. military, Silicon Valley would not exist at all! Thank you very much for bringing up this bit of historical common knowledge: how was Silicon Valley born back then? It was basically the Vannevar Bush model you mentioned (Bush headed the U.S. Office of Scientific Research and Development during World War II), a product bred by the cross-fertilization of three parties: top universities (mainly Stanford University), various departments of the U.S. government, and the capitalists who invented the venture capital model.

So how did it later go off the rails? A lot happened in between, and we made a lot of mistakes, which made serving the government "not cool" in Silicon Valley. But I think Elon Musk, our Palantir, and the people who came out of Palantir (the "Palantir mafia") have helped restore that "cool" factor in defense tech and adjacent fields.

But afterward, a truly awful trend did sweep through the American tech world: in that small Silicon Valley circle, one way to prove to your peers that you are smart is to behave despicably and cowardly—you flatter others, make them feel smart, while quietly transferring their intellectual property into your own pocket. This business model achieved enormous success in Silicon Valley; its unit economics were astonishingly good—outrageously good! If you were a venture capitalist or an insider, you could earn 100x, 1,000x returns from this trick.

Of course, objectively speaking, this capital frenzy also produced some genuinely good things as a side effect. If Silicon Valley had not created such enormous excess wealth, a heretical "bastard misfit" like Palantir could never have raised money back then. It was precisely by riding the bandwagon of everyone in Silicon Valley getting rich that we barely scraped together enough money to survive—even though we were at the very back of the line, among the last to get funded.

Q: But let us return to the proposition that "technology and engineering should serve broader social value." How did Silicon Valley break this contract? Was it because they built their wealth on shallow, stupid consumer entertainment products with little real value?

Alex Karp: When you break that tacit, implicit contract between "wealth creation" and "social value creation," the other contracts in society collapse along with it. The implicit contract was originally this: society allows you to make a fortune—even if you are richer than a nation, we will not bother you—provided that you are genuinely creating real value for society as a whole. Once you violate that premise, society heads toward complete rupture.

I do not actually think the consumer products Silicon Valley makes are entirely without value. But put yourself in someone else's shoes: suppose you are an ordinary carpenter living in Louisiana today. Ask yourself honestly—do you feel that your life today has substantially, materially improved compared with 30 years ago? I do not think so. So whether or not it is objectively correct, this feeling among ordinary people is broadly right—in the eyes of ordinary Americans, the fact is: a tiny group at the top of the pyramid saw their wealth soar 100x, 200x, while your own life has barely moved in 30 years. And those people got rich off consumerism and selling ads!

Worse still, these tech products have sliced society into highly exclusionary "micro-subcultures," which is deeply divisive. Within each little circle, people can get along because their views are highly aligned; but the "threshold of disagreement" that can get you kicked out of the circle keeps getting smaller and more unforgiving. I am a case in point: many Democrats now refuse to acknowledge that I am a Democrat, simply because I think part of the party is going mad. It is getting harder and harder for our whole society to find common ground.

Good — that is the social backdrop, already bone-dry tinder and roaring flames. And then AI arrived! It was like pouring high-octane gasoline, from every direction, onto a blazing fire. Because what are the big shots now running AI labs telling ordinary people? They are saying: “Your future life is doomed — it’s going to be miserable!”

Just think about how absurd that is! In a normal political movement, the leaders tell the public: “Hey, come with us and the future will be better!” Now, instead, the people driving the AI movement are pointing straight at ordinary people’s noses and saying: “Your life is about to be a mess; you’re about to be completely unemployed. And after you’re unemployed, you can live on the money I hand out. Oh, and by the way, by then my personal net worth will probably be…”

Question: It looks like we have lost two precious things at the same time: first, the promise that technology can make ordinary people’s lives better, rather than just producing ad-driven consumerist toys; second, a grand national ambition and a sense of national greatness. Both of these seem to have disappeared.

Alex Karp: These two things are actually closely connected. Because if you really have a national ambition, you have to talk about the things that can concretely improve ordinary people’s lives today and tomorrow, not spend all day terrifying the public with how miserable the future will be, and certainly not build an elite dream that excludes ordinary people. The reason Silicon Valley is so wildly wrong on these major questions is largely that it does not talk to real ordinary Americans at all, and it does not care at all. Silicon Valley is an island; you just need to look cool on that island.

The convergence of the far left and the far right: the former Nazis I met in the German countryside

Question: Talk about the anti-government sentiment in Silicon Valley since the 1960s and 1970s. But now there is also an extremely strong anti-government sentiment in the American right-wing camp. Where did that come from? Is it dangerous?

Alex Karp: At Palantir, we have developed a habit: even when faced with an argument that sounds regressive and extreme, we first look for the reasonable elements behind it.

Put yourself in that position: if a tremendously powerful new technology is rising and could at any moment be used against you, directly or indirectly stripping you of your basic constitutional rights; at the same time, you have personally witnessed the government’s terrible performance in a series of major crises — for example, COVID, which I think most people would admit was managed in some respects as a complete mess; and on top of that, society’s “index of national purpose” has fallen to what may be the lowest level I have seen in my lifetime.

When you deeply distrust the government, when national purpose is low, when you think everyone who claims to have real convictions is a fraud, and when you find that the people doing best seem to have little to do with you — if you still did not distrust the government in that situation, that would be truly bizarre!

But once this anti-government sentiment goes to extremes, whether on the far right or the far left, it becomes extremely unconstructive. And to be clear, I must point out: the “extreme right” you just mentioned and the “extreme left” I criticized earlier are not essentially real right or real left at all — today’s far right is at heart left-wing, and today’s far left is at heart right-wing! Fascist movements often work like this; somehow they generate an indescribably powerful seductive force.

Question: Did you learn this deep insight while studying in Germany?

Alex Karp: When I was studying in Germany, I spent a lot of time with real “former Nazis,” because I desperately wanted to understand their true psychology at the time. To meet real former Nazis, you have to go to remote rural Germany — because in the big cities, you cannot find a single person who admits to having been a Nazi; everyone in the city will pound their chest and tell you, “We hated Hitler back then.” Only in the countryside will people tell you the truth.

After in-depth conversations with these former Nazis, I reached a conclusion: anti-Semitism was of course one of their motives at the time, but for the sake of historical accuracy I must honestly say that a large part of what drove them was Hitler’s powerful personal charisma, the malicious hijacking of what should have been a legitimate sense of national purpose, and the evil abuse of the patriotic feelings that human beings naturally need.

If you apply this historical lesson to today, why do today’s “regressive left” and “regressive right,” which on the surface are diametrically opposed movements, each have such lethal appeal within their own echo chambers? Because the right-wing camp has constantly been fed an extremely simplistic dogma: “Capitalism is always good and can solve every problem.”

But the correct narrative about capitalism should be: “Capitalism is better than the other alternatives humanity has tried.” Note that “better than the alternatives” and “can solve all problems” are completely different things! Traditional center-right institutions have sold capitalism as a panacea, which is obviously nonsense. Capitalism has indeed solved problems that other systems could not, which is why it is superior; but at the same time it also produces extremely serious “negative externalities.”

For example, it will create outrageous misallocation of wealth and polarization, which in turn directly affects political power and leads to the collapse of the entire social structure and shared narrative—because a person with $1 billion in a bank account and a person with only $100 in an account are simply not living in the same world! At its underlying logic, capitalism naturally puts a dollar price on every human life, and how much your life is worth depends directly on your net worth. Even if you are hit by a car on the street and it ends up in court, the amount the driver ultimately pays will probably differ depending on whether the person hit is you, Walter, or me, or an unemployed blue-collar worker. This is the result capitalism naturally produces.

Q: Finally, I want to follow up on the “value of a human being” you just discussed and ask an ultimate question: what does it really mean to be born human? Take Creech Air Force Base in Nevada, for example, where military drones are deployed, and your company Palantir also has deep cooperation in this area. There, “humans” are almost no longer in the decision-making loop. This makes me feel it is like a metaphor for an era: in the face of such powerful new technology, the question is not only to what extent humans must remain in the system’s decision-making loop, but also, if machines truly acquire intelligence in the future and even develop some kind of consciousness, where exactly does humanity’s place lie?

Alex Karp: This indeed touches the most core soul-searching question: when the core capital on which humanity has relied to build its entire moral system and to claim to be the crown of creation—namely our “cognitive intelligence”—is no longer an advantage exclusively enjoyed by humans, what exactly is still special about human beings? When we personally nurture a new species that surpasses us in intelligence, what will we face?

We firmly believe from the bottom of our hearts that in us humans, beyond the ability to speak and computational cognitive ability—abilities that will sooner or later be completely surpassed by machines—there exists something of a higher dimension. That thing is called “humanity”: it is artistic, individual, unique, irreplaceable, more valuable, and must be protected.

Source: The Builders, host Walter Isaacson in conversation with Palantir co-founder and CEO Alex Karp, recorded on August 6, 2026, released on November 10, https://www.youtube.com/watch?v=m4mV4IClDlMAI Industry Signal Channel: Frontier Technology. This content is published with the author’s authorization; the views represent only the author and do not represent Huxiu’s position. If you have objections or complaints about this manuscript, please contact tougao@huxiu.com. Everyone who is changing and wants to change the world is on the Huxiu APP[ ]

Source

Huxiu