What's wrong with Palantir? How investors see the stock

Palantir has fallen nearly 33% year to date, underperforming the iShares IGV software ETF, which is down about 14%, as investors worry AI models from Anthropic and OpenAI could replicate parts of its data-heavy workflow. Analysts cited rising concern that frontier AI features may be productized as substitutes for Palantir and Snowflake, while CEO Alex Karp dismissed the threat as a "farce." The article also notes hedge fund manager Michael Burry is short the stock, even as Wedbush's Dan Ives argues AI infrastructure buildout could support monetization in coming quarters.

Analysis

The market is treating frontier AI as a substitution threat before it is a monetization threat. That matters because software multiples compress fastest when investors believe implementation moats can be abstracted away by a model layer; if that thesis broadens, the damage is not confined to PLTR and SNOW but spills into any workflow vendor whose value proposition is “we sit between data and action.” The likely second-order winner is the model platforms themselves, followed by horizontal infrastructure names that sell picks-and-shovels to the AI buildout, while vertical application software faces a higher bar for retention and pricing power.

For PLTR specifically, the key risk is not immediate revenue attrition but a slower re-rating as buyers delay expansion deals and procurement teams use model-native alternatives as negotiating leverage. That creates a classic multiple/estimate gap: even if bookings hold, the stock can de-rate for several quarters if investors lose confidence in long-duration margin expansion. SNOW has a similar but more defensive setup because it is more likely to be embedded as plumbing, yet it is still exposed to a narrative where data prep and mapping are increasingly bundled into frontier-model workflows.

The near-term catalyst set is binary and time-sensitive: any credible productization by Anthropic/OpenAI of analytics, graphing, or agentic workflow building would hit sentiment again over the next 1-3 months, while stronger enterprise AI monetization from MSFT/ORCL could stabilize the group by reframing AI as additive spend rather than substitution. The contrarian view is that investors may be overestimating how quickly generic models can replicate regulated, secure, domain-specific deployments; the true moat may sit in integration, compliance, and deployment muscle, not the model output itself. That means the current selloff may be more severe in the public tape than in actual customer behavior, especially if CIOs continue to prefer vendors that can own outcomes end-to-end.

Key Decisions for Investors

• Short PLTR vs long MSFT or ORCL on a 1-3 month horizon: express the view that platform AI monetization will outperform application-software re-ratings; target 1.5-2.0x relative downside in PLTR if frontier-model substitution headlines continue.
• Buy PLTR put spreads 2-4 months out rather than outright puts: downside can continue, but implied vol likely stays elevated; structure for a defined-risk move if de-rating accelerates on another productization headline.
• Watch SNOW as the cleaner, lower-beta hedge against PLTR weakness; initiate a tactical long only if management can show that AI-native workflows are increasing data usage per customer, otherwise the stock remains vulnerable to the same substitution narrative.
• If PLTR falls another 10-15% without a fundamental cut to guidance, consider covering into weakness: the market may have priced in more AI displacement than near-term procurement cycles can justify.
• For longs, prefer MSFT/ORCL over software applications into the next quarter: they have the best chance to convert AI demand into measurable revenue, making them the better relative beneficiaries if enterprise spend remains intact.