Palantir Technologies shares have fallen about 40% from their November 2025 peak, and critics now dismiss the company as little more than a wrapper around other firms' AI models. The underlying business tells a different story: Q1 2026 revenue grew 85% year over year and net dollar retention reached 150%.
Palantir Technologies' shares have dropped about 40% from their peak. Yet Q1 2026 revenue grew 85% year over year. Net dollar retention, meanwhile, hit 150%. That gap between a falling stock and an accelerating business has fed a narrative recasting Palantir as just a wrapper for other companies' large language models.
Revenue Kept Climbing Through the Stock's Round Trip
Palantir stock surged roughly 135% in 2025, peaking near $207 a share in November. At that peak, the company carried a valuation approaching $500 billion and a price-to-sales ratio near 110. Shares have since fallen to trade near $122 to $123, valuing Palantir around $300 billion.
But the business kept accelerating. Full-year 2025 revenue grew 56% to $4.48 billion. Q1 2026 revenue rose further, up 85% to $1.63 billion. Guidance for 2026 was first raised to a midpoint of $7.19 billion alongside the full-year results. It was raised again to a midpoint of $7.65 billion after Q1 2026. Gross margin held near 84% in both periods, and U.S. commercial revenue growth accelerated from 109% in 2025 to 133% in Q1 2026.
The Ontology, Not the Model, Is the Point
Palantir's AI Platform, or AIP, is model-agnostic: it plugs into OpenAI, Anthropic, and open-source models rather than a proprietary frontier large language model. That single layer draws the "wrapper" label, but the company's differentiator is its Ontology — a framework tying enterprise data, business logic, operational rules, security controls, and real-world actions into one governed system.
After meeting with management and customers at AIPCon, UBS analysts pushed back on the idea that Palantir's platform is easily copied, pointing to its data integration and governance capabilities. Still, Databricks, Snowflake, hyperscalers, and AI labs are all building toward similar enterprise tools, and cheaper, faster semantic-modeling tools could narrow that head start over time.
Commoditized AI May Not Hurt Palantir's Case
As foundation models get cheaper and more interchangeable, value could shift toward whoever can securely connect them to messy enterprise systems, compliance needs, and real-world workflows — the specialty Palantir has built for two decades. That doesn't make Palantir's valuation cheap: shares still trade around 130-140 times trailing earnings, with a price-to-sales ratio of 56.
Source: 24/7 Wall St.
Trading involves risk.
Risk Disclaimer
