PubMatic (PUBM) Q2 2026 Earnings Call Transcript

PubMatic reported Q2 revenue of $78.6M (+11% YoY) and adjusted EBITDA of $19.6M (25% margin), alongside free cash flow of $13.7M (+47% YoY). The company increased monetization efficiency by processing 2% fewer gross impressions sequentially while raising monetized impressions by 4%, and highlighted rapid AI/agentic traction (AgenticOS: 80 campaigns vs 30 in the prior quarter; ~4,000 AI-powered transactions). For Q3, management guided revenue of $75M–$77M (+12% YoY at the midpoint) with adjusted EBITDA of $17M–$19M, and raised full-year CapEx to $20M–$25M to support GPU-centric AI workloads; it also repurchased 2.1M Class A shares for $21.5M. Overall results and outlook point to durable operating leverage and accelerating AI-driven growth, with FX and some vertical softness cited as offsetting risks.

Analysis

The real message is that PubMatic is shifting from being a commodity SSP to a control point for decisioning, which is structurally better for margins than for raw volume. That creates a relative loser in TTD: if buyer-side UI lock-in keeps eroding, the market will start paying less for generic workflow software and more for infrastructure that can attach directly to outcomes. Second-order beneficiaries include ROKU and FOXA, because premium streaming supply becomes more monetizable when the sell-side can package performance-oriented demand more efficiently; NVDA also benefits on the capex side, but that is a longer-dated and smaller pass-through.

The near-term setup is supportive, but the next 30-90 days are about proving that the current cohort of pilots becomes repeatable budget, not just case-study theater. The key falsifiers are a slowdown in repeat campaign size, a drop-off in monetized-impression mix, or evidence that incremental GPU spend fails to convert into operating leverage by the next couple of quarters. FX and softer travel/entertainment spend can muddy the print, but they do not change the structural thesis unless they broaden into a budget reset.

Best trade is a long PUBM / short TTD pair into any post-earnings volatility, with the hedge reflecting the market’s tendency to overpay for DSP narrative when the value chain is migrating. For higher-conviction expression, use a staged long in PUBM on pullbacks rather than chasing an opening gap; the better entry is after the initial sympathy move fades and sell-side estimates begin to rework higher. The contrarian view is that the market may be front-running agentic adoption too aggressively: if this remains an early-adopter phenomenon into year-end, the multiple can compress even if the business keeps compounding.

Key Decisions for Investors

• Go long PUBM on pullbacks over the next 1-3 weeks; use the post-earnings strength to avoid paying peak sentiment. Risk/reward improves if the stock consolidates while estimates drift up into the next quarter.
• Pair trade: long PUBM / short TTD for 1-3 months as a relative-value expression of sell-side share gain versus DSP multiple risk. Falsify if TTD re-accelerates bookings or PUBM’s repeat campaign sizing stalls.
• Add a small tactical long NVDA against PUBM-driven GPU capex commentary only if you want to express the infrastructure spend angle. This is lower conviction and should be treated as a secondary beneficiary trade, not the core idea.
• Watch for a 2Q/3Q read-through in ROKU and FOXA CTV monetization commentary; if ad load and pricing improve, consider them as second-order longs on open-internet monetization. The setup is months, not days.