Palantir is no longer arguing that enterprise artificial intelligence can become a software product. The second-quarter print shows the conversion already happening inside United States commercial accounts, where ontology-backed deployments are expanding faster than the government franchise that built the company. Management frames the moment as a sovereignty shift: institutions want models they control rather than token streams that train someone else's next model. The market has already paid a mega-cap software multiple for that story. The debate is whether domestic commercial compounding stays this steep once the comparison base doubles again.
United States commercial revenue reached $764 million. That is a 149% year-over-year increase and a 28% sequential step-up. Total revenue was $1.9 billion, and domestic mix now sits at 81% of the company. Bookings tell a louder story than the income statement. Domestic commercial total contract value hit $2.1 billion, and remaining deal value in that same book climbed to $6.2 billion. Net dollar retention, the share of prior-year customer revenue that returned and expanded, printed at 157%. Those figures describe land-and-expand that is still accelerating, not a one-quarter spike.
The counterargument is already inside the price. Shares closed near $178 on the September session before publication, with a trailing earnings multiple above one hundred fifty times and a sales multiple that still treats Palantir like a scarce growth asset rather than a mid-single-digit-billion software company. Remaining deal value assumes options get exercised and termination-for-convenience clauses stay unused. International commercial grew only 26%, and adjusted gross margin absorbed extra cloud hosting for a government customer. The next year either confirms that domestic commercial can stay above triple-digit growth while cash conversion holds near current levels, or it shows that the multiple requires a decade of near-perfect execution that the backlog language itself refuses to guarantee.