Palantir's U.S. Commercial Revenue Just Passed Government. That Changes Its Risk Profile, Not Its 54x Sales Multiple

The most important number in Palantir's third-quarter filing is not the one that crossed over. U.S. government revenue of $953 million was 41.2% of total revenue, almost exactly the 41.2% it represented a year earlier, when total revenue was $1.181 billion. Federal revenue and total revenue have grown at the same roughly 96% pace for four straight quarters, so the crossover changed which line leads, not how heavy the federal book is. Total government revenue of $1.146 billion was still 49.5% of the company.

What the crossover does change is the asymmetry of a bad outcome. Last quarter the commercial line added $587 million of incremental revenue while the federal line added $467 million. A 30% cut to U.S. government revenue would remove $286 million, a sum that one quarter of commercial growth now more than covers. Annualized, the $4.67 billion commercial run-rate already exceeds the $3.81 billion annualized U.S. federal run-rate, so Palantir can grow through a federal air pocket for the first time in its history.

Palantir's enterprise value of $391.6 billion against trailing revenue of $7.29 billion is 53.7x sales. The comparable mark at the end of 2025 was 93.8x on $4.48 billion of revenue, so almost the entire compression came from the top line: trailing revenue rose 63% while market value fell 6%. The stock closed at $171.70 on Nov 5, 2026, 19% below its 52-week high of $212.94, and still commands about 46x the midpoint of its own fiscal 2026 guidance. Palantir is more expensive on sales than CrowdStrike, Snowflake, ServiceNow, Alphabet, Microsoft and NVIDIA individually, and far above the federal IT contractors it used to be lumped with — no software discount exists to close.

Congress changed hands the day after the print. The NBC News Decision Desk projected Democrats to win the House with an estimated 259 seats to 176, an 83-seat majority, and Democrats took the Senate. The market's response was to sell Palantir: shares fell 1.65% on Nov 3 and another 2.1% on Nov 5 to close at $171.70. Analysts were explicit about the reason — they pointed to the slowdown in U.S. commercial contract value, up 11% year over year in the third quarter against 153% in the second, and to remaining deal value that rose only 4% sequentially to $6.517 billion. Investors were not trading a political overlay; they were trading bookings, not ballots.

The crossover rewires where the marginal dollar of demand arrives. NVIDIA is the clearest supplier: its Oct 28, 2025 collaboration with Palantir integrates Ontology with Blackwell-accelerated data processing, Nemotron models for reasoning and NeMo Retriever for agents, and includes a reference design for a government AI factory. Alphabet extended that stack in June 2026, making Palantir available on Google Cloud Marketplace with two-way BigQuery-to-Foundry integrations and direct Gemini-to-AIP connectivity. Microsoft, Amazon and Oracle all carry the same deployment relationships, with Oracle hosting Foundry, Gotham and AIP across its distributed and sovereign cloud regions. The marginal dollar of demand now arrives through cloud and compute budgets, not through federal procurement.

The same shift creates a new conflict rather than removing an old one. The companies that supply Palantir's compute and distribution — NVIDIA, Alphabet, Microsoft, Amazon, Oracle — now sell AI platforms aimed at the same buyer. On the federal side the direction of travel runs the other way: OneGov's OpenAI arrangement buys 27 months of consumption-based token pricing at half the standard rate, so agencies are contracting for model output rather than for platforms. The risk migrates from who controls the budget to who controls the layer above the models, and Palantir's own argument — that an application layer over open models beats dependence on closed ones — is the only durable answer to it.

The near-term test is not political. Palantir's own fiscal 2026 guidance implies fourth-quarter U.S. commercial revenue above $1.216 billion, more than 23% sequential growth, and total revenue of $2.649–$2.653 billion, which is 14.6% sequential and 88% year over year against 96% delivered in the third quarter. The gap between those two questions is where the next re-rating sits: the December quarter decides whether the bookings slowdown was a duration artifact or a structural one.

Over one to three years the debate shifts from mix to durability, and three things decide it: whether contract durations lengthen as management expects; whether stock-based compensation, running at 13.8% of revenue in the third quarter, keeps eating a 62% adjusted operating margin; and whether the hyperscalers' own AI platforms take the enterprise seat Palantir now occupies. The federal half is not the problem — it grew 81% and the procurement rules are moving its way. The problem is that a 62% margin business growing 91% a year is priced at about 46x forward sales, so duration and dilution decide the outcome.