Advertising holding companies are racing to integrate generative artificial intelligence into their media planning, buying, and creative workflows. But behind the promises of automation and efficiency lies a substantial, unquantified financial liability: the massive infrastructure costs of running these advanced computing systems. As agencies begin to absorb these technical overheads, the financial pressure is highly likely to trickle down to their media partners. For digital publishers, this shift threatens to squeeze margins on their most lucrative offerings—branded content and co-produced video campaigns.
The warning signs of this impending cost shift emerged directly from the top of the agency ecosystem. Speaking at the Cannes Lions International Festival of Creativity, Omnicom CEO John Wren raised alarms about the underlying economics of the technology. Wren pointed out that while software companies and enterprise partners are eager to sell AI tools, the broader marketplace has not yet reckoned with the actual operational costs of executing AI at scale.
To preserve their own profitability, agency holding companies are structured to pass backend operational costs along the supply chain. When agencies face margin pressure, their traditional lever is to squeeze vendor fees, production markups, and media distribution margins. Publisher-led branded video studios, which rely heavily on agency-negotiated budgets, represent a prime target for these cost-recovery efforts.
In a typical branded content deal, a publisher sells a bundled package that combines creative concepting, physical production, talent sourcing, and guaranteed media distribution across its digital properties. These deals carry much higher margins than programmatic display advertising, helping to fund independent newsrooms and specialized editorial operations.
However, agencies are increasingly deploying their own proprietary AI platforms to perform work that was historically outsourced to publishers or production houses. For example, Publicis Groupe’s massive €300 million investment in its CoreAI platform and WPP’s collaboration with Nvidia to build AI-driven content engines demonstrate how agencies are consolidating production capabilities. These platforms utilize AI for rapid storyboarding, copy variations, and localized video iteration. By moving these creative steps in-house—and justifying the massive capital expenditure to their shareholders—agencies are positioned to demand lower creative fees from publishers. The publisher is then left with the lower-margin components of the deal: physical execution and distribution.
Furthermore, agencies are likely to demand deeper discounts on the distribution portion of branded video campaigns to offset their internal technology overhead. If an agency must absorb high computing costs to license proprietary large language models and clean room data architectures, that capital must be clawed back. The easiest path to doing so is insisting on lower effective CPMs (cost per thousand impressions) and reduced production fees from publisher partners. This pressure is compounded by the fact that global ad spend growth remains modest, meaning agencies must find internal efficiencies or vendor-side discounts to maintain their historically stable operating margins, which typically hover around 11% to 15% for major holding groups.
This shift presents a serious challenge to publisher infrastructure and planning timelines. Building a branded video studio requires significant upfront investment in physical equipment, specialized production staff, and distribution tech stacks. Unlike programmatic ad setups, these creative operations cannot easily pivot when budgets contract. A publisher that has scaled its internal studio based on historical production margins may suddenly find those margins unsustainable if agencies insist on clawing back fees to cover their AI infrastructure investments.
To insulate themselves from this looming margin squeeze, sophisticated media operators must re-evaluate how they package and price their branded content. Rather than offering easily commoditized creative services that agencies can replicate with internal AI tools, publishers need to double down on what cannot be automated: direct audience access, first-party data targeting, and proprietary talent networks.
Publishers should also demand greater transparency regarding how agencies evaluate the efficiency of AI-assisted campaigns. If an agency insists on reducing a publisher’s production fee because the agency utilized AI for the initial campaign strategy, the publisher must ensure that the performance metrics and distribution guarantees of the campaign are adjusted accordingly.
Ultimately, the infrastructure costs of the AI transition will not be borne solely by the technology providers or the agencies themselves. As holding companies look to balance their books in the face of rising cloud computing and software licensing fees, the pressure will inevitably move downward. Publishers who rely on high-margin branded video deals must prepare now for a more adversarial negotiating environment, ensuring their operational costs are lean enough to withstand the agency squeeze.
This article was generated with the help of AI.
