Palantir posted $1.9 billion in second-quarter revenue, up 93% year-over-year, and $1.1 billion in profit — and used the shareholder letter to accuse the frontier AI labs it competes with of behaving like the capitalists who gave rise to Marxist socialism. CEO Alex Karp argued in a Monday letter and on the accompanying analyst call that companies including OpenAI and Anthropic are structurally set up to absorb their own customers' businesses. The framing is unusual for a company reporting record numbers.
The quarter itself is the strongest in Palantir's history. Karp told shareholders the company booked more profit in a single quarter than it did in total revenue in the same period the year before. AI adoption is the tailwind, not the threat — enterprises and governments buying Palantir's software to orchestrate models and data are the reason the top line nearly doubled.
“There are Marxist overtones and undertones to our business”— Alex Karp, Palantir CEO
Karp's argument is that frontier model providers are a different animal from Palantir. Palantir sells model-agnostic AI and analysis software, letting customers control their own data and what Karp called their AI 'exhaust' — prompts, orchestration, and context. The frontier labs, in his telling, are collecting the same material from their enterprise customers and folding it back into their own models.
Key facts
- 01Palantir reported $1.9 billion in Q2 revenue, up 93% year-over-year, and $1.1 billion in Q2 profit.
- 02Karp said Palantir booked more profit in Q2 than it did in total revenue in the same quarter a year earlier.
- 03In the shareholder letter, Karp accused frontier LLM builders of trying to 'capture the means of production' of their enterprise partners.
- 04Palantir's pitch is model-agnostic AI software that lets customers keep their data, prompts, and orchestration in-house.
- 05Microsoft CEO Satya Nadella has voiced a related concern about AI labs competing with the partners they sell to.
In the letter, Karp cast this as a structural conflict, not a marketing gripe. He compared the dynamic to a Marxist critique of capitalism, arguing that LLM providers are positioned to absorb the workflows and expertise of the very companies paying them.
“Others, including many of those building large language models, intend, knowingly or otherwise, to capture the means of production of their purported partners.”— Alex Karp, Palantir CEO
On the analyst call he sharpened the point in the tech-defense idiom Palantir has adopted since expanding its Pentagon business. He asked whether enterprises really wanted to fund a future where a 'small, tiny group of people' in one place ended up owning the means of production of the American economy while everyone else absorbed the cost.
He also mocked the underlying transaction, calling enterprise LLM spend 'token self-pleasurings … at real cost like other forms of self pleasure.' The line drew attention on Wall Street, but the substantive claim underneath is about IP flow: that customer prompts, fine-tuning data, and integration know-how end up training a competitor.
That competitor concern is not fringe. Microsoft CEO Satya Nadella has voiced a similar worry about AI labs launching products that compete directly with the customers licensing their APIs. The pattern Karp is describing — model providers partnering with a healthcare firm, a design tool, a legal-tech vendor, then releasing overlapping first-party features — has played out repeatedly across 2025 and 2026.
Palantir's positioning follows from that critique. The company sells to governments and enterprises that specifically do not want their operational data leaving their environment, and it charges accordingly. The 93% growth rate suggests the pitch is landing, even as OpenAI and Anthropic post their own outsized numbers selling to a different slice of the same market.
The counterweight is that Karp's framing is convenient. Palantir competes with the labs for enterprise budget, and casting them as extractive gives Palantir's sales team an easy contrast. OpenAI and Anthropic have not, in practice, cornered Palantir — the record quarter is evidence that both models of AI infrastructure can grow simultaneously, at least for now. Neither lab has responded publicly to Karp's letter.
Palantir's take is worth reading as a market signal rather than a moral verdict. The interesting question for enterprise AI buyers over the next 18 months is not whether frontier labs are villains — they are for-profit companies competing for margin — but whether the model-agnostic orchestration layer Palantir sells becomes the default enterprise pattern, or whether the labs succeed in pulling more of the workflow directly into their own platforms. Palantir's $1.9 billion quarter is one data point that the orchestration layer has real budget behind it.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
