Palantir Down 33% YTD as AI Replacement Threats and Valuation Debates Continue to Weigh on Stock
N.R. Finch
Palantir has fallen roughly 33% this year, more than double the 14% drop in the iShares IGV software ETF, as markets wrestle with whether large language models can directly replace its core data-analytics business.
Down 33% — what exactly is the market punishing?
Palantir is off about 33% year-to-date; the iShares IGV software ETF — where Palantir is the third-largest holding — is down only about 14%.
This means → the market is not dumping the software sector broadly. It is singling out Palantir, pricing in a risk it sees as more concentrated here than at peers.
The core fear is singular: can large language models — AI systems like ChatGPT that understand and generate text — directly replace Palantir's specialist data workloads? If yes, its moat disappears.
The "ghost" — who says Palantir can be replicated?
Rosenblatt senior analyst John McPeake told CNBC: a "ghost" looms — whether Anthropic or OpenAI can do it all.
In plain terms = some believe you could tell a large model "build me a Palantir" and it would appear. That expectation is compressing Palantir's valuation.
UBS analyst Karl Keirstead wrote in a mid-June client note that Anthropic's latest blog post on data-analytics capabilities "doesn't sound great for Palantir and Snowflake," warning that investor fears of LLM vendors shipping direct replacement products are rising.
What do partners and customers say?
Snowflake product chief Christian Kleinerman — a Palantir partner — acknowledged at a June 2 investor day that frontier AI models and specialist data services have genuine "overlap," but stressed the relationship is "more complementary in many customer scenarios."
More telling is what customers themselves are doing. Per the UBS report, one Palantir client said publicly at AIPCon 10 in early June: "I want to explore building a parallel open-source graph database, or using OpenAI and Anthropic." The client added: "Palantir isn't open — that's the problem."
This reflects a shift: some customers have moved past worrying about replacement and are actively exploring alternatives — from "could it be replaced?" to "let me try."
How does Palantir respond?
CEO Alex Karp pushed back hard, calling the idea that LLMs can replicate Palantir's commercial products "an absolute farce."
He told CNBC in early June: large language models matter enormously, but the value is in the implementation layer — implying that writing code is easy; turning it into a production-grade enterprise product is another matter entirely.
Put simply = Karp's argument is: AI can write an analytics script, but making it run safely, stably, and compliantly inside a large enterprise — that is what Palantir actually sells.
Bulls vs. bears — who has the stronger case?
Bear camp: hedge-fund manager Michael Burry is short Palantir. On June 25 he wrote on Substack that he covered half his short at $107.15 but still holds put options.
Bull camp: Wedbush tech research head Dan Ives told CNBC on June 27 that the market is severely underestimating enterprise AI demand and monetization over the next six to nine months — benefiting everyone from Microsoft and Oracle to Palantir.
Palantir's P/E and EV/sales multiples were previously in triple-digit territory and are now trending down. This means → the debate boils down to one question: can the incremental demand from AI infrastructure build-out outrun the substitution pressure LLMs exert on Palantir's core business? The answer will emerge quarter by quarter in upcoming earnings.
Content is for reference only, not financial advice.