Frequently asked questions
Is PubMatic a buy at current prices?
DalalBytes rates PubMatic Positive at 76/100, because Q2 2026 delivered the inflection: revenue up 10.5% after three flat years, 25% adjusted EBITDA margin, 67% gross margin, and AgenticOS traction (80+ fully autonomous campaigns, 4,000+ AI-powered deals). At $19.18, about 2.5x EV-to-trailing-sales ex-cash, the market prices a no-growth ad-tech utility. The core risk: DSP buyer concentration has hurt the company twice.
What are PubMatic's 2029 price targets?
DalalBytes' FY2029 targets are bear $11.00, base $39.00, and bull $65.00, from a $19.18 reference price. Base (+103%): double-digit growth sustains at ~13% CAGR to $430M, EBITDA margins return toward 30%, and the market re-rates a profitable independent SSP to 4x sales. The bear case (-43%) is DSP concentration striking a third time. The bull case (+239%) is AgenticOS becoming the open web's buying standard.
What is the verdict on PubMatic?
PubMatic runs the auction house for the open internet's advertising: publishers bring ad slots, buyers bid in milliseconds, PubMatic takes a cut. After three years of flat revenue and a bruising 2025, Q2 2026 delivered the inflection: revenue up 10.5%, adjusted EBITDA margin 25%, gross margin 67%, and the industry's boldest AI-agent bet in AgenticOS. The balance sheet is a fortress for its size: $137.5M of cash, zero debt, free cash flow positive in all eight quarters we reviewed, and $211.4M of buybacks retiring 15.5M shares. At $19.18, about 3x trailing sales, the market still prices it as the turnaround that might stall. Score: 76/100.
How does the PubMatic business engine work?
PubMatic is a supply-side platform (SSP): software that lets publishers auction their ad inventory to the highest bidder in real time. Founded in 2006 by brothers Rajeev Goel (CEO since 2008) and Amar Goel, IPO'd December 2020. Around the core SSP sit four products moving it up the value chain: OpenWrap (header-bidding wrapper, now on CTV and integrated with Unity LevelPlay, covering 90%+ of in-app SDK inventory), Activate (buyers purchase directly through the SSP, skipping DSPs; ad spend more than doubled year over year in Q2 2026), Convert (self-serve commerce media for retailers and transaction businesses), and Connect (data and identity: Walmart Connect Select for shopper audiences on CTV, PayPal Ads ID with 25 billion transactions across 400 million verified accounts).
What is PubMatic's owned-infrastructure edge?
Most ad-tech rents cloud computing. PubMatic owns its data centers, and the compounding shows: cost of revenue per million impressions processed fell 20% on a trailing-twelve-month basis in Q2 2026, the same 20% improvement printed in Q1 2025 and Q2 2025, while the platform processed nearly 92 trillion impressions, up 18%. More volume at lower unit cost, every quarter: that is operating leverage. The multi-year NVIDIA collaboration adds a second gear: 5x faster bid responses, 85% fewer auction timeouts, 3x more ad requests per server, five racks consolidated into one.
Is AI a tailwind or headwind? (AgenticOS)
AgenticOS, launched January 2026, is an operating system of 20+ AI agents for autonomous ad buying and selling, built on open protocols (AdCP, co-founded with Yahoo and Scope3; MCP) and transacting through Activate. Traction is early but measurable: 80+ fully autonomous campaigns (from 30+ in Q1), 4,000+ AI-powered deals (from 1,000+), all five global agency holding companies participating, and Decision Fabric letting buyers run their own models inside PubMatic's infrastructure. Our call: net tailwind near term. Agentic buying needs neutral, low-latency rails on the open web, and PubMatic is shipping product while competitors announce frameworks. The structural risk is walled-garden capture: if Google and The Trade Desk keep agentic budgets inside their walls, the open web gets bypassed. That is the thesis-breaker to watch.
What do the financials show? (the dip and the turn)
Eight quarters tell the story: Q3 2024 $71.8M (+12.7%), Q4 2024 $85.5M (+1.1%), then the DSP disruption dragged Q3 2025 ($68.0M, -5.3%), Q4 2025 ($80.1M, -6.4%), and Q1 2026 ($62.6M, -1.9%) down, before Q2 2026 snapped back to $78.6M (+10.5%). Through it all, operating cash flow and free cash flow stayed positive every single quarter. TTM revenue is $289.2M. Stock-based compensation, the quiet tax, is falling: $8.3M in Q2 2026, down 15% year over year. Management's guidance record is a credibility asset: seven revenue beats in eight quarters, with Q2 2026 beating consensus revenue by 13.7% and consensus EBITDA by 116%.
What happened in the DSP episode?
In August 2025 PubMatic guided Q3 revenue down 11% at the midpoint, citing reduced spend from one top DSP buyer. The stock fell 21% to $8.34 in a day; a law firm announced an investigation; Jefferies downgraded. It was the second DSP scare in a year. The response is why this report is positive: management diversified the demand mix (mid-market DSP spend up 25%+ in Q2 2026), the underlying business excluding the legacy DSP grew 13% in Q1 2026, and the company has beaten its own guidance ever since. Concentration remains the core risk: a single large buyer has now hurt this company twice.
What does the technical setup say? (strong trend near the highs)
From the $19.18 reference price the chart is a powerful uptrend, not a base: price holds above the 20-day ($17.24), 50-day ($16.03), and 200-day ($11.07) averages, trading at essentially the 52-week high of $19.19 after a 132% one-year run. Momentum confirms (MACD positive), but RSI at 69.9 is near overbought: strong trends can stay there, but entries need patience. Support sits at the 20-day near $17.24, then $16; resistance is the $19.19 high, then the round $20 level. Q3 earnings on November 10, 2026 is the next checkpoint.
What is PubMatic worth? (valuation and three-year targets)
At $19.18 the market cap is about $874M; ex-cash enterprise value is about $736M, or 2.5x trailing sales. The market prices PubMatic as a no-growth ad-tech utility, but Q2 suggests that framing is stale. Our FY2026 estimate is about $299M (first-half $141.2M, Q3 guided $75-77M, seasonally strong Q4). The scenario map to FY2029: bear $11.00 (-43%), base $39.00 (+103%), bull $65.00 (+239%). The formula is (FY2029 revenue times EV/Sales, plus net cash) divided by diluted shares. Base-case logic: double-digit growth sustains at ~13% CAGR to $430M, EBITDA margins return toward 30%, and the market re-rates a profitable independent SSP to 4x sales. The bear case is DSP concentration striking a third time. The bull case is AgenticOS becoming the open web's buying standard.
What are the risks and kill criteria?
Principal risks: DSP buyer concentration (high, and historical); walled-garden AI capture bypassing the open web (medium-high); ad-market cyclicality (medium); the 132% run-up into the November 10 print (medium); SBC offsetting buybacks (medium, but SBC is falling); the CFO transition (low, orderly); the 2025 class-action overhang (low, headline only). The kill criteria are clear: if the DSP mix re-concentrates, or if agentic budgets visibly bypass the open web, the growth story breaks and this thesis gets re-underwritten from scratch.
Download the full report (PDF)
Research and opinion, not investment advice. Do your own due diligence before investing.