Palantir CEO Alex Karp accused leading artificial intelligence laboratories of attempting to seize control of their partners' production capabilities, using language that compared them to forces that historically spurred Marxist socialism. In a shareholder letter published Monday following the company's second-quarter earnings, Karp—who holds a PhD in social theory—warned that frontier AI labs can't be trusted by enterprise customers. The letter accompanied financial results that Karp described as showing "Marxist overtones and undertones" in Palantir's business environment.
The data and analytics company reported revenue of $1.9 billion for the second quarter, representing a 93% jump compared to the same three-month period one year earlier. Palantir's profit reached $1.1 billion for the quarter. Karp noted in his letter that this single quarter's profit exceeded the company's total revenue during the corresponding period twelve months prior. The surge in results came as artificial intelligence adoption accelerated across Palantir's customer base, despite the company's concerns about competitors in the AI space.
According to Karp's shareholder letter, companies building large language models "intend, knowingly or otherwise, to capture the means of production of their purported partners." During a conference call with Wall Street analysts, he elaborated that enterprise customers who partner with AI labs are paying "for the right for them to migrate your IP, your know-how, your expertise to their model, so that they can build a competitive business that doesn't require your business or people." The report notes that Karp's underlying argument—that AI labs partnered with companies while simultaneously launching competing businesses in areas like design tools, healthcare operations, legal services, and drug discovery—has been echoed elsewhere, including by Microsoft CEO Satya Nadella.
The report explains that Palantir positions itself differently by offering model-agnostic AI and analysis software that lets governments and enterprises maintain control over their data and what the company calls AI "exhaust"—the prompts, orchestration, and context generated during use. This approach contrasts with frontier labs that, according to Karp's characterization during the analyst call, convince customers to sign up for arrangements that transfer intellectual property to the labs' models at the customer's expense. The report describes Karp's language as jarring, particularly his references to "tech bro patriot" rhetoric common in defense technology circles and his comment about people who "eat vegetables and don't support war fighters" believing they deserve to control production. Palantir's senior leadership consists entirely of men, the report notes.
The article concludes by observing that despite Karp's sharp rhetoric, neither AI laboratories nor Palantir qualify as economic villains or heroes beyond what any for-profit company represents. Palantir's record-breaking quarterly performance demonstrates that the artificial intelligence market is expanding rapidly enough to accommodate multiple business models and competitors simultaneously. The company's strong results suggest demand exists both for frontier model builders and for platforms that help organizations deploy AI while retaining control of proprietary information. As enterprises continue integrating AI capabilities, the choice between partnering directly with model developers versus using intermediary platforms like Palantir will likely depend on each organization's priorities around data sovereignty and competitive positioning. The philosophical intensity of Karp's framing may reflect broader anxieties about power consolidation as AI reshapes business infrastructure, though commercial success so far appears less zero-sum than his warnings might suggest.