Author: [email protected]

Michael Burry has built a career on being early — and loudly wrong before being right. The founder of Scion Asset Management, immortalized for his prescient bet against the U.S. housing market ahead of the 2008 financial crisis, turned his sights on Palantir Technologies (PLTR) this week with a pointed post on X that sent the stock tumbling roughly 7% before he quietly deleted it.

The claim was simple and blunt, as Burry tends to be: Anthropic, the AI startup behind the Claude platform, is “eating Palantir’s lunch.”

Whether he’s right is a separate question. What’s not debatable is that the market paid attention.

What Burry Actually Said

Burry’s thesis centered on Anthropic’s explosive revenue growth — from $9 billion to $30 billion in annual recurring revenue (ARR) in a matter of months — as evidence that enterprise customers are gravitating toward AI solutions that are “easier, cheaper, and more intuitive.” His argument frames Palantir less as a high-growth technology company and more as a labor-intensive consulting business, pointing to the company’s reliance on Forward Deployed Engineers (FDEs) — Palantir staff embedded inside client organizations for months at a time to implement and maintain its platforms.

That model, Burry argued, is structurally vulnerable as direct AI integrations become more accessible. “It took $PLTR 20 years to get to $5 billion,” he noted, while Anthropic is scaling at a pace that suggests the market may be ready to reward the brains of the AI revolution over the operating systems built around it.

This isn’t a new position for Burry. He disclosed a significant short position in Palantir via long-dated put options as far back as September 2025.

The Bull Case: Palantir’s Moat is Real

Not everyone on Wall Street is ready to write Palantir’s eulogy. Wedbush analyst Dan Ives maintains an Outperform rating with a $230 price target, arguing that Palantir occupies a uniquely defensible position at the intersection of AI and federal government infrastructure. The argument: you cannot run sophisticated AI on sensitive government data without the kind of secure, structured, and compliant data architecture that Palantir provides.

That argument gained added texture this year when the Trump administration banned Anthropic from Pentagon systems following a dispute over AI safety guardrails. Palantir was reportedly ordered to remove Claude from its Maven Smart System and rebuild parts of the platform. That incident, while disruptive in the short-term, arguably underscores the stickiness of Palantir’s enterprise relationships — and the risk that pure AI model providers face in regulated environments where trust, compliance, and security clearances matter as much as raw capability.

Palantir has also posted ten consecutive quarters of accelerating revenue growth, a track record that speaks for itself regardless of how the competitive landscape evolves.

The Bear Case: Valuation Leaves Little Room for Error

Where the bull case gets complicated is on valuation. Morgan Stanley analyst Sanjit Singh, while acknowledging Palantir’s standing as a “clear winner through the first stage of the AI cycle,” has flagged that the stock currently trades at roughly 38 times 2027 sales. At that multiple, even strong execution may not be enough to drive meaningful upside. The bar is simply very high.

Burry’s consulting-business critique also has some factual grounding. Palantir’s 10-K does categorize its FDE deployments under professional services — a labor-driven revenue model that is inherently harder to scale than a software subscription or API-based product. As Anthropic and similar companies lower the barrier to deploying enterprise AI, the question of whether Palantir’s hands-on model remains a differentiator or becomes a liability is a fair one to ask.

The Bigger Picture

What this week’s episode illustrates isn’t necessarily that Burry is right or wrong about Palantir specifically. It’s that the AI investment landscape is entering a more complex phase — one where the market is beginning to draw distinctions between infrastructure plays, model providers, and application layers, and debating which of those tiers captures the most durable value.

Anthropic’s valuation recently reached $380 billion, a figure that reflects investor conviction that the model layer is where the leverage lives. Palantir’s case rests on the idea that data infrastructure and operational trust — particularly in government — represent a moat that model providers cannot easily replicate.

Both arguments have merit. The risk for investors is that at current valuations, both stocks demand a level of confidence in the future that leaves little margin for disappointment.

As always, one social media post — even a deleted one — is not a thesis. But when Michael Burry posts, it’s worth understanding exactly what he’s saying and why the market reacted the way it did.