Palantir posts record quarter as CEO Karp calls AI frontier labs 'Marxist'

Palantir's Record Quarter Comes With a Sharp Attack on AI Labs

Palantir delivered its strongest quarter on record, and its chief executive chose the moment to deliver one of his sharpest warnings yet about the companies building the world's most prominent AI models.

The software firm reported $1.9 billion in revenue for the second quarter, up 93 percent from the same period a year earlier, and $1.1 billion in profit. In the shareholder letter, CEO Alex Karp pointed out that the quarter's profit alone exceeded the company's total revenue in the same period the year before. The surge reflects how quickly governments and enterprises have shifted budget toward AI-related data and analytics software.

Karp used both the letter and the subsequent analyst call to cast the big AI frontier labs as a threat to the enterprises they claim to serve. 'Others, including many of those building large language models, intend, knowingly or otherwise, to capture the means of production of their purported partners,' he wrote, describing the dynamic as having 'Marxist overtones and undertones.' He argued that customers are effectively paying for the right to have their intellectual property, know-how and expertise migrated into a vendor's model, so that the vendor can build a competing business that no longer needs them.

Palantir's counter-position is explicit: it sells model-agnostic AI and analytics software that lets organizations keep control of their data and their AI 'exhaust' — prompts, orchestration and context — regardless of which underlying model they use. Karp's argument echoes a concern that Microsoft CEO Satya Nadella has also voiced about frontier labs moving into their customers' businesses. Whether or not the rhetoric lands, the quarter gives Palantir a powerful platform for the pitch.

What Karp's 'Marxist' Charge Actually Signals for the AI Market

Karp's Comments Are a Competitive Pitch, Not Just a Provocation

Karp's language is deliberately inflammatory, but it is not random. It is a positioning statement designed to draw a line between Palantir's model-agnostic platform and the business models of OpenAI, Anthropic and other frontier labs. The underlying observation — that AI companies have partnered with enterprises while simultaneously launching products in design, healthcare, legal and drug discovery — has become a common talking point among enterprise software executives, including Microsoft's Satya Nadella. The difference is that Karp expresses it in ideological terms and ties it directly to vendor lock-in and the migration of customer IP.

The Numbers Show a Demand Boom, With an Unanswered Question

Palantir's financials are the foundation of the argument. Revenue of $1.9 billion, up 93 percent year over year, and $1.1 billion in profit are record levels for the company, and Karp's point that quarterly profit surpassed prior-year quarterly revenue underscores how fast the business is scaling. The unanswered question is whether that growth is a function of Palantir's specific model-agnostic strategy or simply the rising tide of AI spending. If the tide is the main driver, frontier labs capture much of the same opportunity; if the data-control pitch is the driver, Palantir is taking share that might otherwise go to AI application builders.

Who Benefits If the Argument Resonates

If Karp's characterization persuades buyers, the clearest beneficiary is Palantir itself, because every enterprise that decides to keep data, prompts and contextual workflows inside its own environment is a potential customer for model-agnostic infrastructure. The frontier labs face a subtler cost: even a minority of enterprise buyers delaying or conditioning purchases on data-control terms would raise their customer acquisition costs and complicate their move into vertical applications. The two sides are not necessarily zero-sum, however. Palantir's own results suggest enterprise AI budgets are large enough to support multiple business models for now. What remains to be tested is whether the rhetoric matures into durable contract terms or fades as a marketing gesture.

What Enterprise Buyers and Investors Should Take From Palantir's Results

For enterprise technology buyers, Palantir's results and Karp's comments are a reminder that AI vendor selection is now also a data-control decision.

• Enterprises evaluating AI platforms should ask whether a vendor's architecture lets them retain ownership of prompts, orchestration and context, and whether that vendor operates competing services that could use customer data — the specific risk Karp attributed to frontier labs.
• Buyers should compare model-agnostic platforms like Palantir with integrated offerings from OpenAI and Anthropic on portability, switching costs and contractual data terms, not just raw model capability; Palantir's $1.9 billion quarter shows enterprises are already spending heavily on this choice.
• Investors should watch Palantir's next quarterly report for evidence on whether 93 percent revenue growth and $1.1 billion in profit reflect sustainable demand or a one-time surge, and whether Karp's public disputes with AI labs affect enterprise deal flow.