Palantir (PLTR) Q2 2026 Earnings Call Transcript

Palantir reported Q2 revenue of $1.935B, up 93% YoY, with U.S. revenue rising 115% YoY to $1.573B. The company also posted record cash generation with $1.22B in adjusted free cash flow (up 115% YoY, 63% margin) and raised full-year 2026 revenue guidance to ~$8.154B at the midpoint (+82% YoY). Additionally, it increased U.S. commercial FY26 guidance to in excess of $3.424B (+134% YoY) and lifted adjusted FCF guidance to $4.5B–$4.7B, alongside $4.9B in remaining performance obligations (+103% YoY). Deal momentum was strong with 220 deals >=$1M (record) and $13.1B total remaining deal value (+83% YoY).

Analysis

This print reinforces a shift from “AI experimentation” to budgeted, sticky enterprise infrastructure, which should keep taking share from legacy implementation-heavy vendors. The first-order winner is still PLTR, but the second-order beneficiary is NVDA: sovereign, customer-controlled stacks imply more local inference, fine-tuning, and refresh demand, even if model vendors lose relative control. The losers are the generic AI layer and services stack — hyperscaler-led orchestration, systems integrators, and federal IT contractors — because a platform that owns workflow, security, and benchmarks compresses billable hours and reduces model-switching friction.

The key risk is not a demand cliff; it’s re-rating risk. At this scale, even a small deceleration in U.S. commercial bookings or any sign that gross margin is being traded for hosted deployments can cut the multiple before fundamentals roll over. Near term, the stock can keep running for days to weeks on estimate revisions, but the 1-3 month test is whether the next quarter shows broad-based conversion rather than a few oversized deals; over 6-18 months the question is whether sovereign AI becomes a repeatable procurement standard or remains a high-touch sale.

Contrarian view: consensus may be underestimating how much of the upside is already in the “operating leverage” story, while underestimating the fragility of sentiment if growth merely normalizes. The market will likely forgive SBC and concentration only while top-line acceleration stays extreme; if not, this becomes a valuation compression candidate despite excellent fundamentals. Missing data to watch: cohort retention excluding the top 20, conversion rate from pilots to multi-year contracts, and whether cloud-hosting costs keep creeping into margins as deployments scale.

Key Decisions for Investors

• Maintain a tactical long PLTR, but only add on 3-8% post-earnings pullbacks or after intraday weakness; use 90-180 day call spreads rather than outright upside chase if implied vol is elevated. Falsifier: any sequential slowdown in U.S. commercial bookings or guidance that implies growth falling materially below the current run-rate.
• Enter a pair trade: long PLTR / short a basket of legacy federal IT services (SAIC, BAH, CACI) over the next 1-3 months. Thesis: platform adoption disintermediates labor-heavy integration work. Cover if those names begin winning Palantir-adjacent contracts or if government buying shifts back toward services-led implementations.
• Add NVDA on weakness as a secondary beneficiary of sovereign AI deployment, but size it below PLTR; 6-12 month horizon. The upside is from more deployed compute and fine-tuning, not from model-branded hype. Falsifier: if enterprise deployments remain mostly low-usage pilots instead of persistent inference workloads.
• Put PLTR on a watchlist for valuation-driven de-risking if the next earnings cycle shows margin compression from hosted infrastructure without a commensurate acceleration in new logo wins. If that happens, trim into strength rather than waiting for a drawdown.