AI is reshaping how work gets done across the economy. For many software businesses, that shift is creating real pressure on traditional, seat-based or renewal-driven models. Digital advertising is fundamentally different: it is built on usage and measurable outcomes, not user counts.
That structural difference is why digital advertising is, in my view, one of the clearest and largest beneficiaries of AI. For PubMatic, the leading AI ad tech platform, AI is not an add-on productivity feature; it is embedded in how we drive revenue and deliver results. The broad adoption of AI in digital advertising is structurally expanding our addressable market and creating a powerful moment of transformation for our business.
A simple, usage-based model
PubMatic grows with transaction value, not customer headcount or seat licenses. As advertising activity increases, we process more impressions, more auctions, and more signals. Because we own and operate our infrastructure, we directly manage the unit economics of that growth. In the third quarter of 2025, we processed 24% more impressions while reducing costs by 19% over the trailing twelve months, demonstrating the operating leverage in our model and the payoff from sustained innovation.
We compete in live auctions, not on contract renewals. Buyers and publishers continuously choose the platforms that deliver better outcomes. Every impression is a fresh decision point, and performance is measured in real time. In this environment, AI raises the bar and platforms that execute faster and optimize more intelligently gain share. That dynamic plays to PubMatic’s strengths. At the same time, it’s a business model that cannot be replicated by AI-generated software alone.
A compounding data and AI advantage
Our platform is built on proprietary data accumulated over nearly twenty years: publisher inventory signals, user and contextual insights, auction dynamics, and commerce integrations. These datasets, refined across thousands of customer integrations and trillions of transactions, compound over time and strengthen our machine learning models, which in turn power our proprietary agents. AI does not erode this advantage; it amplifies it.
Digital advertising is also distinct in how quickly it can monetize AI investment. Feedback on performance is rapid and transparent. Better outcomes drive higher spend, which generates more data and further improves optimization. This feedback loop runs continuously across millions of auctions per second, creating a direct, measurable link between AI investment and revenue impact.
Built-for-AI infrastructure
We have been preparing for this moment for years. Our GPU-accelerated infrastructure, developed in partnership with NVIDIA, enables AI inference at roughly one millisecond—materially faster than traditional approaches. That performance reduces auction loss, recaptures spend, and supports the processing of more than one trillion impressions per day.
Our focus has been consistent: build the technology, infrastructure, intelligence, and partnerships that move faster and perform better. In a transaction-based model, every microsecond matters.
Agentic execution at scale
Over the past year, we have launched a series of first-to-market AI-powered agents, introduced AgenticOS to enable autonomous execution within live auctions, and delivered the industry’s first fully autonomous, end-to-end agentic advertising campaign in connected TV. AI-driven automation is already reducing campaign setup times by 87% and troubleshooting time by 70%. These capabilities are live in market today and delivering measurable results at scale.
As advertising becomes more automated, data-intensive, and outcomes-driven, platforms built for real-time execution become essential. PubMatic’s transaction-based economics, owned infrastructure, and compounding data advantage position us not just to navigate the shift to an AI-first workforce, but to benefit from it and capitalize on it as these trends accelerate.
Positioned for the next decade
In 20 years of building PubMatic, I have seen periods of technological change create short-term uncertainty. Over the long term, value accrues to platforms that see these shifts early and become more central to how markets operate. PubMatic was built with that objective, and we remain focused on executing against it.
We are not waiting to see how this plays out. We are building it, optimizing it, and scaling it.
