40-Year Data Vet Kahan Explains Palantir's LLM-Agnostic AI Platform

For AI and machine learning teams, the Stagwell–Palantir interview is a case study in why platform-agnostic architecture is winning enterprise AI budgets. Kahan's 40-year data career and his insistence on not being tied to any one model or cloud provider speak to the real question facing enterprises: how to adopt AI without betting on a single vendor. The answer, according to Palantir, is an abstraction layer that can swap in the best capabilities at any time.

The development is a maturation of Stagwell's November 2025 agreement with Palantir to build an AI-powered data and audience platform, which initially raised eyebrows because Palantir is best known as a defense and surveillance contractor. Nearly a year later, Stagwell's chief AI officer, John Kahan, is explaining why Palantir's architecture is the right foundation for a challenger holding company. Kahan, a 40-year data and analytics veteran who held senior roles at IBM and Microsoft and chaired Stagwell's board until last year, was persuaded by CEO Mark Penn to delay retirement and run the Palantir collaboration as interim CEO of Stagwell Marketing Cloud. Elspeth Rollert has since taken over The Marketing Cloud after returning from maternity leave, but Kahan remains the executive between the two CEOs, spending most of his time liaising between the companies.

“

Nearly a year later, Stagwell's chief AI officer, John Kahan, is explaining why Palantir's architecture is the right foundation for a challenger holding company.

Kahan's core argument is that Palantir offers a platform-agnostic, AI-agnostic set of capabilities. He does not want Stagwell tied to any one model, any one platform, or any one capability. Palantir's platform can sit on top of any large language model or cloud provider, allowing Stagwell to deliver the best available capabilities to customers at any point in time. In a media and advertising ecosystem defined by walled gardens, cookie deprecation, and rapid AI model commoditization, that neutrality is strategically valuable. Instead of betting on a single model provider whose pricing, terms, or performance may change, Stagwell can swap components in and out. For advertisers, this could mean a data and audience engine that is not captive to Google, Meta, Amazon, or any single AI vendor.

The privacy dimension is equally important. The sources indicate Palantir is innovating on the privacy side of the data business, including differential privacy techniques that add calibrated noise to datasets to allow aggregate audience intelligence without exposing individual-level information. Kahan's presence at the UN General Assembly as a Novartis Foundation board member underscores his exposure to health data governance, where privacy constraints are extreme. If Palantir can bring health-grade privacy methods to advertising data, it could differentiate Stagwell at a time when regulators are tightening cross-border data transfers and targeting restrictions. This is not merely a compliance feature; it can become a commercial advantage in a market where brands are increasingly nervous about data ethics and legal risk.