Alex Karp on AI and the Middle Class: Why a Wealth Tax Is the Wrong Fix

In a conversation for The Builders, Walter Isaacson's interview series, Alex Karp, co-founder and CEO of Palantir, talks about how dyslexia and an outsider's childhood shaped the company, why its products took three years to earn revenue, where he draws the line on surveillance, and why he thinks AI will widen the wealth gap, and why a wealth tax is the wrong answer. The interview was taped on August 6, 2026.

This article is our own summary, with short quotes, from the video's automatic transcript. Watch the full conversation above for everything that was said.

The short version

• Three years with no revenue. To build an anti-terror product that both stops attacks and protects civil liberties, Palantir "had to violate every single norm" of software, starting with three years without revenue. Karp says it took a neurodivergent founder to fight for that.
• Dyslexia as an edge. He kept his dyslexia private until about six years ago. His view: if you are dyslexic, you are "not colonized by the broader culture," so you deal with problems as they are.
• Real products, not addictive ones. He contrasts Palantir with software that is "simple but addicting": low value, hard to get rid of, sold by large sales teams.
• Surveillance needs a legal mandate. Palantir is used heavily in Europe against organized crime, people smuggling and terrorism. He says the key line is whether data use has a warrant or legal mandate, and AI built for the battlefield should be legally barred from domestic use.
• AI will widen the gap. He expects AI to create more and better vocational jobs, but says even wages rising at twice inflation won't keep pace with people in AI who could get 10 to 50 times wealthier. "You can't have people becoming 100 times wealthier when they're already billionaires and everyone else is becoming 5% wealthier."
• His alternative to a wealth tax. Raise taxes on people like him, on condition that for every unit of tax raised, half a unit of government waste is found. A wealth tax, he argues, wouldn't help the poor, would break tech and would require surveillance of everyone's belongings.
• Enterprises are angry about LLMs training on their data. He says companies are "livid" to find models trained on their proprietary data while they pay, and Palantir's ontology is meant to make models useful while protecting that data.

What he said, by topic

Family and being an outsider

Isaacson's theory of biography is that it is all about the father. Karp describes his father, shaped by losing his German Jewish mother early, and his mother as an "excruciatingly talented," difficult Black woman. A messy divorce left him with a deep trust in his own ability to survive and "a latent mistrust of structures." He says nothing Palantir has built in culture, aesthetics or products is something an insider would do, and jokes that his rebirth fantasy is to be a Catholic Bavarian veterinarian.

Philosophy and why he left it

Karp wrote his doctoral work as a critique of Heidegger, whom he calls one of the most important philosophers of the last two centuries and who became an ardent supporter of Hitler. He argues that many regressive movements on the far right and the far left are Heideggerian rather than what they claim to be, and calls himself "still an unhappy Democrat." Halfway through the dissertation he realized he wanted to build something instead. Great builders, he says, combine an outsider's view with an aesthetic sense that is as much humanities as technology.

Founding Palantir

He met Peter Thiel at Stanford Law School, in the same section and dorm. Thiel proposed bringing the fraud-detection methods PayPal had used to the intelligence world, and Karp said yes. His reason was the size of the American project, anchored in a Constitution he says is still fresh: he singles out the First, Second and Fourth Amendments.

Privacy and surveillance

He says Europe got privacy right by protecting it in both commercial and government contexts, while Americans worry about the government but not enough about companies. His students don't mind Google knowing everything about them, which he thinks underestimates how much tech products shape their opinions. On Palantir, he says critics get the company wrong but are right that companies controlling all data can heavily influence what people think. The hard technical problem, he explains, is handling data correctly when a warrant expires, and Palantir's products are built to follow those rules.

Strong military, restricted technology

He argues that if you are anti-war you need a strong military, and that the US has no choice but to build battlefield AI because adversaries are. But he wants legislation preventing that technology from being used domestically. He warns that fascist societies have "no crime" because the only criminal band is the government, and that people with unusual opinions, like him, are the first to be targeted.

AI and inequality

Karp expects AI to exacerbate class inequities. He calls the left's critique legitimate, but says a wealth tax hurts the wealthy without helping the poor and would hurt the middle class. When Isaacson suggests AI could make assessing wealth practical, Karp agrees, and says that is exactly the problem: enforcing it would mean AI imputing the value of everything you own. His proposal ties higher taxes for people like him to identified waste. He says the inequity of wealth and the condition of poor people are two separate problems, and that solving only one divides society.

Data sovereignty and the LLM labs

He says US enterprises are paying for model products while those models train on their most valuable, proprietary data. He spends part of his day explaining to companies that Anthropic's Dario Amodei is not the caricature they think. Palantir publishes guidance with its clients on contracts and on working with open and closed models, and its ontology, a concept he says was first derided and is now widely copied, both makes raw model output valuable and acts as a protective layer.

Silicon Valley's broken covenant

He says Silicon Valley would not exist without the US military and calls it a highly insulated island. The tacit deal was that people could get very wealthy because they created value for society. He says a carpenter in Louisiana doesn't feel his life is materially better than 30 years ago while a small group got 100 or 200 times wealthier, and that AI lab leaders telling people their lives will get worse is like "dumping gasoline" on a fire.

What it means to be human

Asked what remains special about humans when machines surpass our cognitive abilities, Karp points to the US Constitution, where rights are presupposed by a higher source rather than granted by parliamentarians. That leaves room for something in people beyond intelligence, which he calls artistic, individual and particular, and which he says the country should fight to protect.

What it means for your business

1. Build for value, not lock-in. Karp's critique of "simple but addicting" software is a useful test: would customers keep your product if leaving were easy?
2. Protect your proprietary data before using LLMs. Read model contracts closely and put a layer between your data and the model that keeps it from becoming someone else's training set.
3. Design for the rules, not around them. Products that handle consent, warrants or retention correctly can be sold where competitors cannot.
4. Show people how AI makes their life better now. Karp's warning is that doom talk about jobs breeds a backlash that hurts everyone building with AI.
5. Expect the wealth-gap debate to reach your industry. Tax and regulation on AI gains are coming into the political conversation; plan for them.

Video details

• Title: Alex Karp on Why AI Will Hurt the Middle Class: You Can't Have Billionaires Becoming 100x Wealthier
• Host: Walter Isaacson
• Guest: Alex Karp, co-founder and CEO of Palantir
• Published by: The Builders with Walter Isaacson on YouTube (interview taped August 6, 2026)
• Watch: youtube.com/watch?v=m4mV4IClDlM
• Note: quotes come from the video's automatic transcript and are lightly cleaned. Check the video for exact wording.

Want to put this to work?

We help owners turn AI trends into a plan, a product and a launch. Book a free call or see our services.