Kevin PaffrathFounder & CEO

Karp's argument is that when you use a closed model and pay for tokens, the real reason those tokens are discounted is not to grow the market. It is to get access to your intellectual property so their models get better. There was functionally no honest disclosure of this. His claim is that these companies have exposed themselves to unlimited liability — both from enterprise clients whose IP was migrated without their knowledge, and from the broader global economy if those models contribute to destruction of value at scale. He says the only way you deal with that level of liability is to go to the government and give them 50 percent of your business. That is his nationalization thesis.

No. I think it is convenient fear-mongering designed to sell Palantir products. Who gets a 26-minute segment on CNBC unless they are saying the world is going to end and the biggest AI companies are going to get nationalized and their investors are going to lose everything? That narrative got him the airtime. Palantir's pitch is essentially that you should use their sovereign AI approach instead of Claude or GPT because your data stays protected. That is a legitimate product pitch, but the doom framing is exaggerated.

Palantir is arguing for a layered architecture they call sovereign AI. Rather than putting everything you have into a closed model that improves on your IP, Palantir builds an application layer on top of open-weight models or on-premise systems that you control. Their ontology layer sits between your data and the LLM. You get the intelligence without handing over your intellectual property to an external company. They have a lot of forward-deployed engineers who help clients set this up. It is a smart system. The fear-mongering is the marketing, but the underlying product is real.

Karp said the first lesson in poker and in life is that if you think the other person is the mark, you are the mark. He was applying that to enterprise clients of OpenAI and Anthropic. These clients think they are getting discounted tokens because these companies want to grow the market. Karp says no, the enterprise clients are the mark. They are the ones giving away their IP. Their data trains the models, and now those models work against them. He is saying the investors in OpenAI and Anthropic are also the mark because they are funding a liability time bomb.

The financials are genuinely strong. They have about 9.4 billion in free cash after paying around 500 million in operating costs. Revenue is up about 92 percent. Costs are only up 53 percent, so they have massive pricing power. The growth rate on Wall Street expectations is sitting around 47 percent annually. That gives them roughly a 23 PEG ratio right now, with a PE of about 109. My own forecast is closer to 50 percent growth over the next four years. At that rate, you could justify a three PEG, which puts this closer to a 250 dollar stock. I do not think it is fully priced.

This is the clearest concrete example. Dario Amodei literally described at Dreamforce how executives at Anthropic now access all their Salesforce data just by talking to Claude. They do not log into Salesforce anymore. Claude pulls the deal data, win-loss probability, themes, all of it. So what happens to Salesforce's value proposition long term? That is what Karp calls the vampire effect. Claude is slowly extracting value from enterprise software companies by acting as the interface layer. Over time, you do not need the CRM. You just need the model. That is the market cap destruction Karp says will generate massive lawsuits.

No. That is the most extreme version of his argument and it took him about 12 minutes to get there, then one minute at the end to walk it back to the more reasonable point, which is that we cannot let China get ahead on AI. If you genuinely believed what he was saying in the first 12 minutes, you would not buy any stock at all. The fact that he is telling you to buy Palanteer in the same breath tells you that the doom scenario is marketing, not his actual belief.

I like it. I bought it at 118 and I sent that buying alert to our alpha members at meetreinvest.com. I would still buy more at current prices because the fundamentals support it. The pricing power is exceptional, the balance sheet is strong, and they are highly selective with customers to the point where there is a waitlist. The arrogance you see from Karp is somewhat earned given the numbers. I just would not buy it because of the nationalization thesis. I would buy it because the PEG is reasonable for the growth rate.

The last point, not the first twelve minutes. The real conversation is about making sure Western democracies maintain AI dominance over China. Palantir's work with the US military gives them genuine credibility there. If you overregulate AI right now and slow down American labs, China gains ground. That is the serious geopolitical argument buried inside what started as a 26-minute doom advertisement. He gets there eventually, but it takes a while.

It is what comedians do. They compliment someone in a very extreme way first and then deliver the punchline. Karp says Dario is brilliant, ethical, thoughtful. Then he says, but Dario is the mark and he does not even know it and he is going to get nationalized. The setup is the compliment. The punchline is the takedown. He did the same thing with the investors in OpenAI and Anthropic. He called them greedy, said they think the businesses are the mark, then flipped it and said no, you are the mark. It is effective rhetoric, but you have to recognize it for what it is.

If you think the other person is the mark, you are the mark. Karp is applying this to everyone involved in the closed AI model ecosystem. The enterprise clients think they are getting a deal on tokens. The investors think the enterprise clients are the mark and that OpenAI and Anthropic are going to extract all the value. But Karp's argument is that the investors themselves are the ones who have made a mistake driven by greed. They did not mind when IP was migrating from every business into an entity they owned. They did not think through the next step, which is that value then migrates to the government. That is the poker lesson playing out at a macro scale.

I do not think it is a coincidence that this is happening roughly a month before the Anthropic IPO. The framing is essentially buy Palantir instead of investing in Anthropic because Anthropic is going to get nationalized and go to zero. That is a very convenient message right before a competing company is about to raise a massive amount of public capital. It does not mean he is wrong about the product or the business fundamentals, but the timing tells you something about the strategic motivation behind going on CNBC and saying all of this in September 2026.

There are three distinct layers. The first is studying language and heuristics, which is the early Palantir approach to understanding unstructured data. The second is an ontology layer where data gets combined with algorithms and deterministic solutions. The third layer is where they now incorporate LLMs. Palantir calls the combination of these three layers sovereign AI. It is different from just calling the OpenAI API or using Claude directly because at no point in that stack does your raw proprietary data leave your environment. You are using the model as a tool rather than feeding the model as a training source.

She bit hook, line, and sinker. That is the most telling part of that entire interview to me. A sophisticated financial journalist sat there for 25 minutes and basically validated his framing. When he said that there are liability issues that make these companies uninsurable, she jumped in and said yes, that is going to drive a lot of liability. She became part of the marketing. That is how you know Karp is genuinely skilled at this. He did not just sell the narrative to retail investors watching YouTube. He sold it to the on-air talent at one of the most watched financial networks in the world. That is a 26-minute advertisement that cost him nothing.

According to Karp's logic it would happen in stages. First the business valuation collapses when you announce you are going to the government to hand over 50 percent of the company. Then it collapses again when government officials actually join the board because the alignment between commercial incentives and mission becomes impossible. Then it might collapse a third time across the broader market because these companies are so deeply embedded in the financial projections of hundreds of public companies that their failure becomes systemic. He is essentially describing a contagion scenario. I think that is deeply exaggerated, but if you take his premises at face value, that is the sequence.

Look, I am a financial analyst who built his brand on YouTube. Part of what I do is be transparent about everything, including the weird stuff. The coupon codes, the alpha membership at meetreinvest.com, the Reinvest Terminal launching in November — that is all part of how this business works. I think being upfront about it is actually the more honest approach. I tell you I own Palantir stock before I tell you I think it is undervalued. I tell you there is a coupon expiring tomorrow. The alternative is to pretend you have no financial interest in what you are recommending, which is what a lot of people in this space do, and that is where you should be more skeptical, not less.