Topic: SPO landscape, Trade Desk Kokai impact, and COGS leverage
Key points:
PubMatic is a platform for direct inventory monetization; reselling is not its business.
Trade Desk is not in yield optimization, which is core to PubMatic’s value for publishers.
COGS: unit cost reductions have been consistent for a decade, often double-digit; Q3 gross impressions processed grew >20% without proportional CapEx increase, driven by software and AI.
Mgmt stance: Neutral on SPO noise (industry definitions are clear, PubMatic’s direct model is differentiated); bullish on COGS leverage (AI and owned infrastructure will continue to drive gross margin expansion as revenue reaccelerates).
Q2 — Matthew John Swanson
Topic: CTV growth drivers and AI/agentic AI competitive advantage
Key points:
CTV: >90% coverage of top 30 global streamers; added Tubi, Future Today, Local Now; advertiser mix expanding from hundreds to tens/hundreds of thousands.
AI right to win: (1) own infrastructure (e.g., NVIDIA partnership), (2) scale (~1 trillion daily ad impressions, ~2,000 publishers), (3) demonstrated innovation — 17 AI agents launched in publisher platform vs. competitor planning first agent.
Mgmt stance: Bullish on CTV (long runway via Activate, curation, ADCP launch); bullish on AI leadership (1–2 year execution advantage over competition).
Q3 — Shweta R. Khajuria
Topic: (Follow-up on Trade Desk Kokai impact and SPO)
Key points:
PubMatic is a platform for direct inventory; no more efficient way for buyers to access premium publisher inventory.
Kokai evaluates media differently; two steps taken in Q3: (1) revised ML algorithms to optimize inventory mix to Trade Desk, (2) worked with SPO partners to configure DSP seats — both largely completed.
Trade Desk spend stabilized in August and September; ~85% of Trade Desk clients now live on Kokai; bulk of movement likely behind.
Mid-market DSPs grew >25% YoY in Q3 as part of DSP diversification.
Topic: AI search impact on publisher traffic and NVIDIA partnership benefits
Key points:
AI search impact limited: if zero search traffic to publishers, high-end impact is single-digit % of revenue; ~60% of impressions are CTV/mobile app (unaffected); browser-based search referral traffic ~15% (likely lower for top publishers).
Offensive opportunity: growing AI search experiences (e.g., Conde Nast archive search) create new ad inventory.
NVIDIA partnership: (1) GPUs for real-time ad decisioning in low-latency CTV/live sports — 5x faster bid response reduces timeouts, increases bids and win rates; (2) Triton inference servers for traffic shaping (trillions of impressions/day, decisions in milliseconds); (3) software accelerator for Apache Spark to speed data processing and optimization.
Mgmt stance: Neutral on AI search risk (limited single-digit % potential impact, mitigated by CTV/mobile app growth); bullish on NVIDIA partnership (structural differentiation driving sustainable growth in 2026).
Q5 — Robert Coolbrith
Topic: 5x faster bid response and demand-side optimization capabilities
PubMatic has a lead in accelerated computing; counterparties on demand side can take advantage, but PubMatic may need to take a bigger role in optimizing or hosting demand-side logic to fully capture capabilities.
Mgmt stance: Bullish (lead in accelerated computing enables new optimization strategies; potential to expand role in demand-side optimization).