Laily UPN
The annual Cannes Lions International Festival of Creativity is traditionally a theater of optimism. It is where the industry’s brightest minds gather to project the future of marketing, draped in the prestige of the Croisette. This year, the buzzwords were unmistakable: AI, Efficiency, Agentic, and Orchestration. These terms were delivered from stages with practiced confidence, painting a picture of a seamless, automated, and hyper-productive future.
However, move beyond the polished stage panels and into the quieter, more candid spaces of the festival—the candlelit dinners and the late-night terrace conversations—and a markedly different narrative emerges. The sheen of the "AI revolution" is beginning to dull, replaced by the sober, grinding reality of unit economics and the messy friction of business transformation.
The Cost of Innovation: Moving from Hype to Line Items
For the past twelve months, the industry has been intoxicated by the promise of generative AI. The initial pitch was seductive: infinite scale, lean teams, and unprecedented speed. But as the experimental phase concludes, companies are waking up to a fiscal hangover.
"No one really admits openly as to how much they’ve been burning to develop," says Ian Maxwell, CEO of ad tech firm Converge Digital. Maxwell’s critique cuts through the industry’s performative enthusiasm. He points to a critical flaw in the current "AI-first" fervor: the astronomical cost of compute and token consumption.
"We know in our own use case if you throw a whole code base at it, the costs become absolutely astronomical," Maxwell notes. "With the rise in token costs, it’s now a case where it is vastly more costly than simply having engineers."
This shift in sentiment marks a pivot from theoretical potential to "felt" cost. Last year, the conversation focused on the upside. Today, the arithmetic has changed. Businesses are realizing that the projections offered by early-adopter hype did not account for the long-term, sustained cost of running complex AI models at scale.
Chronology of a Shift: From Hype to Procurement
The evolution of the AI narrative over the last 18 months can be broken down into three distinct phases:
- The Phase of Discovery (Early 2023): AI is treated as a magical productivity multiplier. "Ten times the output" is the common refrain. Exploration is rampant, and budgets are treated as R&D slush funds.
- The Phase of Proliferation (Late 2023 – Early 2024): AI-generated content floods the ecosystem. The focus shifts to quantity and speed. Agencies begin building proprietary tools, and the "agentic" workflow enters the lexicon.
- The Phase of Procurement (Current): The "bill-shock" period. Clients demand transparency, and agencies struggle to justify the ROI of their AI stacks. The conversation turns from "What can AI do?" to "What does it cost to keep it running?"
As Jess Dervyn, an analyst in Gartner’s marketing practice, observes: "Clients are expecting a lot of transparency. They want to understand where AI is being used. They have to be honest and say, ‘this is where we’re using AI.’ But they’re not as transparent about how much it costs. It’s changing everything."
The "Invisible" Variable: Sustaining Scale
The industry is currently caught in a cycle of "build first, ask later." Because the technology is deemed transformational, the perceived risk of not experimenting outweighs the risk of overspending. However, as projects move from pilot to production, the "token cost" ceases to be a theoretical abstraction. It becomes a fixed line item, sitting alongside rent, power, and payroll.
Peter Mears, Global CEO of Havas Media Network, identifies the tension: "The introduction of new technologies accelerates what is possible, but also accelerates the expectations required to want more for less." This creates a paradox where efficiency gains are immediately cannibalized by the need to maintain, fine-tune, and license the underlying models.
AI Isn’t Controversial; It’s Just Early
One of the most persistent myths at Cannes this year was that the industry is deeply divided on the "ethics" of AI. The reality, according to many insiders, is much more mundane: the industry is simply in a period of lag.
"I don’t think going into these kinds of discussions thinking anything is sacrosanct is a mistake," says Jim Mollica, CMO of Bose. He notes that while AI is excellent for the "routine, fundamental stuff," it remains remarkably poor at nonlinear, human-centric creativity.
The industry is currently performing a massive, uncoordinated experiment to determine where the line between "efficient" and "soulless" lies. As Abby Laursen, VP of Product Marketing at Snap, explains, the narrative has shifted from "endless production" to "brand fit." The question is no longer "Can we make this with AI?" but "Should we?"
The Disclosure Debate
A recurring sub-theme at the festival was the issue of disclosure. Should brands be required to flag AI-generated content? Debra Aho Williamson of Sonata Insights suggests that the debate may be moot before it even reaches a consensus.
"Does disclosure matter?" she asks. "Or is it something that becomes second nature to how you create advertising—like CGI, or graphic design, or any Photoshop?" Just as we stopped disclosing the use of digital retouching in the 1990s, we are likely heading toward a point where AI usage is the default state of creation, rendering the "AI-generated" label an antique.
Voices from the Croisette: The Operational Model
The most sophisticated discourse at Cannes wasn’t about the content AI produces, but how it forces a restructuring of the agency-client relationship. Mark Singer, CMO at Deloitte Digital, emphasizes that "orchestration" is the true topic du jour.
"The concept of orchestrating is not just doing work with agents, but actually and fundamentally working differently than the way you have before with your agencies and partners," Singer says. He challenges the necessity of traditional processes, such as the formal briefing, suggesting that the "operational model" of marketing is being forced into a state of flux.
Implications for the Future of Marketing
The transition of AI from a "transformation story" to a "procurement story" has several critical implications for the industry:
- The Death of the "10x Productivity" Myth: Agencies that sold AI based on hyper-inflated productivity claims will face a reckoning. Realistic, supervised, and managed AI workflows—which might only offer 3x efficiency—will prove more sustainable than "black box" solutions that hide massive compute costs.
- Procurement as the New Creative Director: As AI costs hit the balance sheet, finance and procurement departments will exert more control over creative production. If a campaign costs more to generate via AI than via traditional methods—when accounting for token costs, latency, and quality control—it will be rejected.
- The Rise of "Composable" Workflows: The winners in the next two years will not be the agencies with the most expensive proprietary models, but those with the most "composable" architectures. Being able to swap out models, manage token consumption, and integrate AI into existing human workflows will become the primary competitive advantage.
- Brand Identity and the "Lag": Because every brand is operating on a different clock, there will be no industry-wide standard for AI usage. Some brands will lean into "all-AI" creative, while others will build their value proposition around the "human-made" label. This diversity will define the next cycle of creative competition.
Conclusion: The Bar Remains
Whether it is termed "agentic," "orchestrated," or "efficient," every AI initiative must eventually clear the same hurdle: it must be financially viable at scale. Cannes, for all its glitz and hyperbole, is a microcosm of the wider market. It is currently in a state of pre-rationalization, where the excitement of the "what" outweighs the reality of the "how much."
But the tide is turning. As the festival concludes, the quiet conversations in the backrooms are beginning to dominate the agenda. The industry is realizing that while AI is undoubtedly a revolutionary tool, it is not a free one. The next phase of the AI story will be written not in the language of Silicon Valley hype, but in the language of balance sheets, procurement contracts, and disciplined unit economics. The industry hasn’t reached that point of maturity yet—but it will.
Lina Hope
Asro
