Zeta Global and Palantir announced a strategic partnership on June 23, 2026, at Cannes Lions, with one concrete architectural commitment at its center: Zeta’s Data Cloud will be rearchitected on Palantir Foundry. The headline number (Zeta projects the arrangement could drive more than $100 million in annual revenue in the coming years) is a forward-looking target, not guaranteed. What matters more for analysts is where in the stack the bet actually lands.
Foundry contributes three things to the new architecture: an ontology layer, governance and compliance infrastructure, and a containerized operational model. Those are not marketing features. They are the preconditions for any AI system operating on enterprise data to produce decisions that can be traced, audited, and attributed. That positioning puts this partnership at the opposite end of the martech pipe from activation tools that push audiences to ad platforms. This is the foundation underneath.
Athena Moves from AI Layer to Governed Operating System
Zeta describes Athena as “the operating system for agentic marketing — the intelligence layer where opportunities, decisions, and outcomes converge.” With the Data Cloud rebuilt on Foundry, Athena will draw on what the companies describe as richer, more expansive enterprise data and act on it in real time. Palantir co-founder and CEO Alex Karp framed it precisely: “Palantir and Zeta are using Ontology to create a next gen marketing environment.” Zeta’s co-founder and CEO David A. Steinberg added that the goal is “bringing AI-powered marketing” to enterprise organizations at scale.
The shift matters because “AI layer” and “governed operating system” are not the same thing. An AI layer can produce outputs. A governed operating system on an ontology-backed data platform can produce outputs that are auditable: you can ask what data the agent acted on, what relationships it assumed, and what outcome was claimed. That auditability is exactly what enterprise buyers need before they allow an autonomous system to make decisions with budget consequences. The partnership’s framing of “measurable outcomes at enterprise scale” is the companies’ stated goal, not a proven capability, but the infrastructure choice moves in the right direction.
What Is a Data Ontology, and Why Does It Matter for Marketing Measurement?
A data ontology is a formal model that defines what entities exist in a company’s data (customers, campaigns, events, products) and the relationships between them. Fixing those definitions before any agent acts is what makes autonomous decisions auditable: if the system knows precisely what a “conversion” is and how it relates to a “campaign touchpoint,” an analyst can trace an AI-generated outcome back to its cause and attribute it correctly.
This is the gap that has made agentic marketing hard to trust. Agents operating on loosely defined or siloed data produce outputs that cannot be reliably connected back to causes. An ontology solves that at the schema level, before any model runs. Foundry’s ontology and governance layer, the same infrastructure the partnership puts under Zeta’s Data Cloud, is built for environments where auditability is a compliance requirement. Applying it to Zeta’s marketing data infrastructure puts a measurement-grade foundation under Athena’s decision loop.
A Category Signal Worth Watching
The Zeta-Palantir structure illustrates a split that is becoming clearer across martech: the stack is separating into an activation layer and a governance layer. The activation end covers agents that read and write to ad platforms, as we tracked with Meta Ads MCP connectors giving AI models read-write access to campaigns. The governance end is what this partnership addresses: the ontology, lineage, and attribution infrastructure that has to exist before activation decisions can be trusted.
Durable measurement lives in the governance layer. Brands that build on a well-defined ontology can answer “did this campaign work and why” with data that holds up to scrutiny. Brands that skip the governance foundation and go straight to agentic activation will produce outcomes, but they will struggle to attribute them cleanly or defend them in a budget review. The Foundry rearchitecture makes that governance layer explicit in Zeta’s product offering, rather than implicit in engineering choices a customer cannot inspect.
For context on measuring outcomes inside AI surfaces specifically, DISQO’s approach to incrementality measurement for LLM advertising covers the signal side of that problem. Governance and incrementality methodology are complementary concerns: one fixes what you measure, the other tests whether the intervention worked.
Both Zeta (NYSE: ZETA) and Palantir (NASDAQ: PLTR) are public companies. The $100 million revenue projection is a multi-year forward-looking estimate tied to Palantir’s eligible enterprise customer base, per trade coverage of the partnership terms. No customer deployments or rollout dates have been announced.