By Ocient Staff
Last week, Ocient hosted a breakfast briefing at The Soho Hotel in London focused on where AdTech, AI, and data infrastructure are really heading. A topic that is particularly relevant as the industry reacts to Publicis’s acquisition of Liveramp.
What stood out most was how grounded the conversation felt. There was very little of the usual “AI is going to solve everything” talk. Instead, the discussion stayed focused on the practical reality many organizations are facing right now: AI adoption is moving incredibly fast, but the infrastructure underneath most businesses is struggling to keep pace.
We had a great mix of perspectives in the room from across the ecosystem, including leaders from Annalect, Roqad, Zeotap, Yahoo DSP, and Ocient. The questions from attendees were thoughtful, direct, and honestly reflected where the market seems to be right now.
Why businesses are buying outcomes, not AI workloads
People are past the experimentation phase. They are trying to figure out how to operationalize AI in ways that are sustainable, governed, and commercially useful.
One idea that came up repeatedly throughout the morning was that businesses are not actually buying AI workloads. They are buying outcomes. While it may sound obvious this is an important distinction, and one the industry is only beginning to fully absorb. Most organizations do not care how sophisticated the model is if it cannot improve decision-making, reduce operational friction, accelerate workflows, or create measurable commercial impact. AI is increasingly being evaluated the same way any enterprise technology is evaluated: can it produce results at scale, can it be trusted, and can it operate reliably inside real-world business environments?
That shift changes the conversation considerably. Suddenly the focus moves beyond demos and experimentation toward operational readiness, governance, scalability, and infrastructure. For the past couple of years, so much of the discussion around AI has centered on models, interfaces, and capabilities. But increasingly, organizations are realizing that none of that matters if the underlying systems cannot support production-scale deployment. It is one thing to demo AI. It is another thing entirely to run it reliably across a large organization with real governance, privacy requirements, compliance expectations, and operational pressure.
A lot of the discussion centered on that gap. Many companies are still dealing with fragmented systems, disconnected datasets, governance challenges, and infrastructure environments that were never designed for the scale or complexity AI introduces. Marketing teams may be racing ahead with experimentation, but backend systems often are not moving at the same speed.
One statistic shared during the event really captured the disconnect: Gartner estimates enterprise-grade AI deployment currently sits at only 2-3% of organizations. That number surprised some people in the room, but it also reinforced what many teams are experiencing firsthand. While AI experimentation is everywhere, relatively few companies have operationalized AI in ways that are scalable, auditable, secure, and commercially sustainable.
Governance, scale, and the operational reality of AI
One of the more interesting conversations focused on what happens in an agentic AI world. As several speakers pointed out, a single AI-driven request can trigger multiple downstream queries and workflows at once. That changes the economics and operational demands of infrastructure pretty quickly. Systems built for yesterday’s reporting and analytics workloads may not be ready for the volume and concurrency AI systems create.
That point landed with a lot of people in the room because it reflects a challenge many teams are already starting to experience.
Another theme that surfaced repeatedly was governance. There is sometimes a tendency in the market to talk about AI as if it somehow changes the rules around privacy and compliance. But as one speaker put it, GDPR still exists. Organizations still need consent frameworks, auditable access controls, remediation processes, and clear data governance policies. AI may change workflows, but it does not eliminate accountability.
That becomes even more important as organizations move toward more autonomous systems and interconnected agent-to-agent environments.
Why infrastructure is becoming the competitive moat
There was also an interesting discussion around how interfaces themselves may evolve over the next few years. One prediction raised during the panel was that by 2027, roughly 40% of buyers may bypass traditional DSP interfaces entirely, instead interacting through MCP servers and AI-driven orchestration layers. Whether that timeline proves accurate or not, the broader point was compelling: the competitive advantage in AdTech may increasingly shift away from the interface and toward the intelligence layer sitting behind it.
If that happens, infrastructure becomes even more important. And that may have been the biggest takeaway from the morning: infrastructure is the moat.
The organizations that succeed over the next several years will likely be the ones that can connect infrastructure, governance, intelligence, and business outcomes together in a coherent way. Not just experiment with AI tools, but operationalize them responsibly and at scale.
That is a much harder challenge than simply adopting a new model or application. Nobody tried to oversimplify the moment the industry is in. There was excitement about the potential of AI, of course, but there was also realism about the operational complexity involved in making these systems work inside large organizations. That honesty made the conversations more valuable.
It also reinforced why events like New Digital Age’s Foresight gathering matter right now. The future of marketing and AdTech is not going to be shaped by one technology alone. It is happening through the intersection of AI, infrastructure, data governance, media platforms, privacy expectations, and changing consumer behavior all at the same time.
Those shifts are already underway. The challenge now is whether organizations can build the operational foundations needed to keep up with them. It felt less like a typical industry panel and more like a real working session on where the market is actually headed.
Get in touch today and let’s continue the conversation.