Michael Burry, the investor immortalized in The Big Short, fired a single tweet on X on April 8 that rattled markets: “Anthropic is eating Palantir‘s lunch.” Though the post was soon deleted, major financial outlets had already captured screenshots. The next trading day, Palantir (NASDAQ: PLTR) tumbled more than 8%, closing at $129.25, dragging the iShares Expanded Tech-Software ETF (IGV) lower in sympathy.
Anthropic’s Growth vs Palantir’s: Four Months vs Twenty Years
Burry laid out three sets of numbers to puncture Palantir’s valuation narrative. First, Anthropic grew its annualized revenue from $9 billion at end-2025 to over $30 billion by early April 2026 — a feat accomplished in just four months. By contrast, Palantir took two decades to lift its ARR from $2 billion to $5 billion. Second, data from corporate spend platform Ramp showed enterprise AI adoption reached a record 47.6% in March, with Anthropic capturing roughly 73% of new enterprise AI spending; nearly a quarter of Ramp customers now pay Anthropic directly, versus just one in 25 a year ago. Third, in head-to-head competitions for new enterprise clients, Anthropic beats OpenAI about 70% of the time.
Anthropic’s Momentum and a Product Blow
Bloomberg confirmed on April 6 that Anthropic had reached $30 billion in ARR. The company’s enterprise customer base is growing even faster: the number of clients paying over $1 million annually doubled from 500 to 1,000 in just two months. With 80% of revenue from enterprises, Anthropic trades more like a pure B2B software company. A $380 billion valuation from its Series G in February 2026 marked the first time its ARR surpassed OpenAI’s (around $24-$25 billion). Market chatter suggests Anthropic could IPO as early as October 2026. On the same day Burry tweeted, Anthropic launched Claude Managed Agents, a cloud platform that lets enterprises deploy AI agents without touching underlying infrastructure — a direct assault on Palantir’s reliance on forward-deployed engineers. Days earlier, Anthropic also secured a Google TPU chip partnership with Broadcom.
Burry vs Palantir: A Long Feud
Burry’s animus toward Palantir is well documented. In Q3 2025, his Scion Asset Management disclosed put options on PLTR (the notional value was initially misread as $912 million; Burry later said he spent only about $9.2 million on the puts). He also bought puts on Nvidia that quarter. In November 2025, Scion deregistered as an investment adviser; that same month, Palantir CEO Alex Karp called Burry “bat shit crazy” on CNBC. Not backing down, Burry published a 10,000-word bear case on Substack in late 2025, arguing Palantir could be worth just $46 per share. His thesis zeroes in on Palantir‘s FDE (Forward-Deployed Engineer) model, which he calls a low-margin consulting business disguised as an AI/SaaS company. Palantir stations engineers — internally called “Deltas” — at client sites for extended periods, often in air-gapped environments, even on Airbus assembly lines. Burry contends this is billing for human labor, not product. Palantir’s own blog frames FDE as a “product discovery mechanism.” Ramp’s AI data now raises a structural question: as enterprise AI procurement shifts from “buy systems, deploy people” to “subscribe to cloud agents,” Palantir’s FDE-dependent stickiness could become a liability. Burry simply did what he does best — spot a lurking vulnerability and call it out.