# Omnicom CEO: ‘The marketplace hasn’t seen what the cost of this AI is’ > Source: > Published: 2026-07-29 04:01:00+00:00 # Omnicom CEO: ‘The marketplace hasn’t seen what the cost of this AI is’ Omnicom CEO John Wren just said the quiet part out loud: nobody’s actually priced AI yet. The admission came on the holdco’s Q2 earnings call yesterday, in response to an analyst question on where AI is driving the biggest cost savings, and, more pointedly, whether those savings are flowing back into client marketing budgets or leaking out elsewhere. Wren’s answer focused less on savings and more on cost. “The marketplace hasn’t seen what the cost of this AI is,” Wren said. “That’s going to weigh into the equation as well.” It’s a notable line from the CEO of the world’s largest ad holdco, delivered on the record, on a call meant to inform investors. It also lines up with [Digiday’s own reporting](https://digiday.com/media-buying/were-starting-to-wonder-ad-industry-chases-ai-value-as-usage-outpaces-proof/) on how the big holding companies are handling this: rather than solve the pricing question directly, several are folding AI costs into bigger commercial structures like principal media deals. That workaround exists because nobody’s found a way to measure what AI actually delivers. Without that, there’s no basis for a clean price, just a bigger bucket to bury the cost in. Wren’s comment suggests even Omnicom hasn’t solved that underlying problem. In the meantime, Omnicom’s approach is to pass savings on to clients where it can. “Where there are savings, we’re sharing them with our clients for sure, but we’re still in the early stages of this,” Wren said. His caveat matters given how hard it still is to put a real number on what those savings actually are. That difficulty hasn’t stoppes Omnicom from trying to categorize where the value comes from. It does so with two buckets: efficiency, from agentic workflows driving better decisions inside the marketing and sales intelligence Omni platform , and effectiveness, from data and identity assets improving client outcomes. Neither is easy to attach a hard figure to. “I would say, in large part, any savings clients are deriving, they are in fact reinvesting immediately into the marketplace,” said Wren. “We’re also focused on measurement and constantly going back to our clients and letting them know what we achieved.” It’s a sign of how unresolved this still is — pricing AI means solving for cost, output and attribution all at once, and none of those are settled yet. Omnicom Media Group CEO Ralph Pardo, touched on the same tension in a separate [conversation with Digiday](https://digiday.com/podcasts/as-ai-automates-media-work-omnicom-media-is-rethinking-the-agency-business-model/) earlier this month. Asked how Omnicom is thinking about charging for work when AI is handling more of it, he acknowledged the commercial model itself is shifting. “There’s no doubt that there’s new elements now that need to be considered in terms of a commercial relationship,” said Pardo. Some of that relationship still comes down to paying for people, he continued, but “there are aspects now that are shifting to paying for outcomes,” alongside data and technology costs that need to be accounted for directly. “Tokens are not free. Being able to query LLMs comes at a cost,” Pardo said. “These are all elements that we are working with our client partners and procurement teams… to start to highlight. This is the cost of doing business.” He described client reception to that conversation as largely positive, in part because clients are working through the same questions internally. But he was clear that the pricing conversation depends on transparency about where those costs actually sit. “The core to that discussion has to be open and transparent about some of the decisions and some of the areas of cost, frankly, that we are all incurring as everybody adapts their business models.” Pardo’s comments put a finer point on what Wren gestured at more broadly. Token and more widely [compute costs are real](https://www.linkedin.com/posts/digiday_digiday-sunday-poll-activity-7484623656883511297-PN6X?utm_source=share&utm_medium=member_desktop&rcm=ACoAAAFN4-sBCvrmw7uTkypu1EqZTOb77A2EZDY) and growing, and Omnicom is actively trying to build that into client conversations rather than absorb it quietly. Whether that shows up as itemized technology fees, outcome-based pricing, or folded into larger arrangements like principal media deals — an area [Digiday has reported on separately](https://digiday.com/marketing/how-ai-costs-are-quietly-reshaping-principal-media-deals/) — remains an open question, and one an exchange later in the call didn’t resolve. ### More in Media Buying #### [ Media Buying Briefing: Why Omnicom merged Mediahub and Hearts & Science ](https://digiday.com/media-buying/media-buying-briefing-why-omnicom-merged-mediahub-and-hearts-science/) Merging Hearts & Science and Mediahub into one agency that will operate under a new name and brand may not be the last move the holding company makes in simplifying its go-to-market offering. #### [ Google quietly gives ground on PMax controls ](https://digiday.com/media-buying/google-quietly-gives-ground-on-pmax-controls/) The search giant’s black-box ad product has confounded media buyers for years. Now, it’s passing them the reins – sort of. #### [ The upfront market is nearly done, as sports acts as both negotiation driver and obstacle ](https://digiday.com/media-buying/the-upfront-market-is-nearly-done-as-sports-acts-as-both-negotiation-driver-and-obstacle/) All that’s left for buyers to do is to finalize packages with any traditional media firms that don’t have some digital or streaming element.