After Cannes, AI is no longer a creative experiment. It is becoming the operating layer of modern marketing.
Cannes used to be where the advertising industry exuded confidence. Agencies arrived with case studies, brand films, award campaigns, client parties, and enough creativity-related vocabulary to keep several generations of marketers employed. Technology was always present, but it usually appeared as a tool, a trend, or a threat to be politely absorbed into the language of brand storytelling.
This year, the conversation changed. Artificial intelligence no longer sat at the edge of the advertising business as a speculative creative toy. It moved into the center of the operating model.
According to The Wall Street Journal, agencies and brands are rapidly embedding AI into ad creation, campaign targeting, campaign monitoring, and the operational structure of the ad business itself. Cannes became less a showcase of AI experiments than a signal that Madison Avenue has moved from curiosity to dependency.
That shift is easy to underestimate because advertising has always been good at adopting new language before it changes its own behavior. Every cycle brings a new vocabulary: digital transformation, personalization, programmatic, performance marketing, retail media, creator economy, social-first, full-funnel, commerce media. The industry knows how to rename old services in new clothes.
AI is different because it does not merely add a new channel or a new production technique. It changes how work moves through the agency system. It affects who produces, who reviews, who optimizes, who measures, and who controls the client relationship. It compresses timelines, automates variation, shifts value toward data and workflow infrastructure, and exposes how much of the old agency model depended on labor-intensive coordination.
For holdcos, this is not a question of whether AI can make better ads. That is too narrow. The more serious question is whether the agency group can defend its role as the operating architecture between brands, platforms, data, media, creative production, and commercial outcomes. That is where the pressure is now.
Cannes Became a Signal, Not a Sideshow
The advertising industry has spent the past few years treating AI as a mixture of creative novelty and reputational risk. It could generate images, write variations, mock up campaigns, produce scripts, imitate styles, and create material quickly enough to make traditional production workflows look slow. At the same time, it brought obvious concerns about originality, client approval, disclosure, training data, synthetic case studies, and the credibility of award entries.
The Cannes conversation now appears to have moved beyond that early phase. Campaign Live reported that Cannes Lions chair Phil Thomas said 40% of entries used AI this year, double the share reported for 2025. That number is not just an awards statistic. It is an industry adoption signal.
AI has stopped being the scandal at Cannes. It is becoming part of the entry form.
That does not remove the integrity problem. In some ways, it sharpens it. Once AI becomes common in the creative process, the hard question is no longer whether a campaign used AI. The hard question is how AI was used, who made the critical decisions, what was actually original, what was client-approved, and what the submitted work proves.
Awards culture will now test human judgment as much as creative output. A campaign produced with AI can still be strategically brilliant, culturally precise, and commercially effective. It can also be a polished simulation built on weak insight, borrowed aesthetics, and exaggerated claims. The difference will not always be visible in the final asset.
That is a governance issue hiding inside a creativity discussion.
Cannes has always rewarded the finished case study. AI forces the industry to care more about the process behind it. The provenance of the idea, the proof of execution, the client’s role, the human contribution, and the measurable outcome all become more important when production itself becomes easier to automate.
The same problem applies outside awards. Brands will not only ask whether an agency can produce more content faster. They will ask whether the agency can produce defensible work under conditions where speed, scale, and synthetic output can easily outrun quality control.
AI Is Moving Inside the Marketing Machine
The WSJ story captures the central commercial shift: AI is being embedded across advertising functions, not reserved for a few experimental creative teams. Campaign creation, targeting, monitoring, production, planning, and operational restructuring are starting to converge around AI-enabled workflows.
That convergence changes the economics of agency work.
Traditional agencies did not only sell ideas. They sold process. They sold meetings, research, drafts, decks, adaptations, media plans, localization, reporting, and layers of coordination. The client paid not only for brilliance, but for the machinery required to get from brief to campaign.
AI attacks the machinery.
It can produce first drafts, generate campaign variants, summarize research, resize and adapt assets, propose audience segments, monitor performance, surface anomalies, create reporting narratives, and support optimization. None of this eliminates the need for human expertise. It does reduce the amount of billable effort required for many tasks that agencies used to monetize.
That creates a business-model problem before it creates a creativity problem.
If AI makes large parts of production and coordination faster, the agency must either pass efficiency back to the client, reinvest it into better work, or defend fees through a new value proposition.
The old logic of staffing-heavy delivery becomes harder to justify when clients can see that many intermediate steps are being compressed.
This is especially difficult for holdcos because they operate at a scale where workflow economics are not theoretical. When AI changes utilization, margins, headcount assumptions, production hubs, and service mix, the effect is structural. The agency group has to decide what becomes automated, what remains premium, what gets centralized, what stays close to the client, and what should be built as proprietary infrastructure.
The winners will not simply be the groups with the best AI demos. The winners will be the groups that can redesign how work travels across creative, media, data, technology, legal, procurement, and measurement without turning the entire organization into a prompt factory.
The Platform Advantage Is Getting Stronger
The uncomfortable part for agencies is that the largest AI advantage may not sit inside the agency business at all.
Google, Meta, Amazon, TikTok, Pinterest, OpenAI, and other platform players are expanding AI capabilities directly inside advertising systems and user interfaces. Their advantage is not just model capability. It is proximity to data, distribution, identity signals, commerce behavior, ad inventory, measurement loops, and user attention.
That is the platform problem for agencies. A platform can use AI to reduce friction between creative generation, audience targeting, placement, measurement, and optimization.
It can make the buying interface easier. It can encourage brands to generate and test more assets directly inside the platform environment. It can push advertisers toward automated campaign systems that make agency mediation feel slower or less necessary.
Agencies still have important advantages. They understand brand context, organizational politics, market positioning, creative direction, category nuance, and cross-channel strategy. They can help clients decide what should be done, not only what can be automated. They can also protect brands from becoming too dependent on platform-defined optimization. But that defense only works if agencies control a meaningful layer of orchestration.
If AI ad infrastructure becomes platform-native, the agency risks being pulled toward execution support while the platform controls the operating environment.
That is a dangerous position. It leaves agencies advising on brand ambition while platforms shape the actual delivery logic.
The future agency role will depend on whether agencies can build or command systems that connect strategy, creative intelligence, data governance, channel execution, performance measurement, and commercial learning across platforms. The agency cannot become merely the human wrapper around someone else’s automation stack.
Agentic Advertising Raises the Stakes
Adweek’s recent Cannes coverage points to the next phase: agentic advertising infrastructure. The story described major players unveiling systems that allow AI agents to connect with one another and perform marketing tasks that humans previously handled, including planning, buying, targeting, and execution.
That development should get holdco attention.
Agentic advertising changes the unit of competition. The industry moves from tools that help humans produce marketing assets toward systems that can coordinate parts of the marketing process.
The question becomes less about whether an AI can write a headline and more about whether AI systems can plan audiences, transact media, adjust campaigns, and interact with other systems in ways that reduce human handling.
This is where the agency business becomes vulnerable and valuable at the same time.
It becomes vulnerable because more workflow can be automated. It becomes valuable because clients will need someone to govern the automation, align it with brand strategy, protect it from platform capture, and decide when machine optimization conflicts with business judgment.
Agentic systems require clear operating rules. They require permissions, escalation paths, brand constraints, legal boundaries, performance thresholds, audit trails, and a disciplined view of what decisions can be delegated. Agencies that understand this can become strategic control partners. Agencies that treat agents as another production shortcut will miss the deeper shift.
The new agency product may not be a campaign. It may be an operating system for how campaigns are conceived, generated, tested, governed, adapted, and learned from. That is a very different business.
Creativity Is Still Scarce, But Its Context Has Changed
There is a predictable argument that arises whenever AI enters a creative industry: if machines can produce endless content, human creativity becomes more valuable.
That is partly true, but it is too comfortable.
Human creativity becomes more valuable only when it is attached to judgment, taste, strategic clarity, cultural understanding, and accountability. A vague appeal to “human creativity” will not protect agencies if clients believe the agency is using expensive human structures to supervise work that could be automated more cleanly elsewhere.
The real scarcity is not creativity in the abstract. It is decision quality.
When everyone can generate hundreds of versions, the scarce skill is knowing which version deserves to exist. When everyone can create synthetic visuals, the scarce skill is knowing what visual language the brand can own. When everyone can personalize, the scarce skill is knowing where personalization becomes noise. When everyone can optimize, the scarce skill is knowing what should not be optimized away.
AI does not make originality irrelevant. It makes weak originality easier to disguise.
That is why Cannes is such a useful signal. The festival sits at the intersection of reputation, craft, commerce, and self-mythology. If 40% of entries used AI, the industry is already living inside the new creative condition. The question is whether agencies can still prove that their contribution is more than acceleration.
For premium agencies, the opportunity is clear. They can use AI to remove low-value friction, expand exploration, improve versioning, and accelerate execution while protecting the human decisions that create market difference. For weaker agencies, AI will expose how much of their work was process theater.
Holdcos Need Operating Architecture, Not AI Theater
The holdco response cannot be a louder AI pitch. That is exactly the trap. The market is already full of exaggerated claims about AI transformation. Clients have heard enough about proprietary platforms, automated workflows, generative studios, AI-powered insights, and end-to-end marketing intelligence. The next phase will be less forgiving. Clients will want proof that AI changes cost, speed, quality, measurement, and accountability in ways that help the business.
That requires operating architecture.
Holdcos need to show how AI is embedded across the agency group without creating chaos. They need to clarify when AI is used, how outputs are reviewed, how client data is protected, how brand consistency is maintained, how rights issues are handled, how performance is measured, and how learning flows back into the system. They need to decide which capabilities are centralized, which remain agency-specific, and which are exposed to clients as strategic infrastructure.
They also need to defend against internal fragmentation.
Large agency groups already struggle with overlapping brands, competing platforms, inconsistent workflows, and duplicated capabilities. AI can make that worse if every unit builds its own toolset, trains its own teams, and sells its own version of AI transformation.
A holdco that cannot organize its own AI operating model will struggle to sell AI-enabled orchestration to clients.
The serious move is not to claim that AI is everywhere. The serious move is to make the operating model legible. Clients should understand where the agency group creates value, where automation reduces cost, where human judgment enters the process, and where governance protects the brand.
That is the difference between an AI story and a business story.
The New Competition Is Control
The agency business has always lived between brands and markets. It translated business objectives into messages, media choices, creative systems, and cultural presence. AI does not remove that role, but it changes the control points.
Control used to sit in talent, client relationships, media buying power, creative reputation, and production capacity. Those still matter. But new control points are emerging around data access, workflow infrastructure, model integration, agent permissions, measurement systems, and platform interoperability.
The agency that controls only the idea may lose ground to the platform that controls the loop.
That loop runs from audience signal to creative generation, from media placement to performance feedback, from commerce data to campaign adjustment, from user behavior to the next recommendation. AI strengthens whoever owns the loop because the system improves through use. The more work flows through the platform, the more the platform learns. The more the platform learns, the more advertisers depend on it.
Agencies have to prevent themselves from being reduced to creative garnish on platform intelligence.
The better path is to become the layer that helps clients maintain strategic independence. That means building cross-platform intelligence, preserving brand memory, enforcing governance, negotiating the boundaries of automation, and translating AI capability into business advantage without surrendering the client’s operating logic to any single platform.
This is where holdcos can still matter. Their scale, client access, talent base, data partnerships, and global delivery capacity give them a potential advantage. But the advantage has to be organized. Size alone will not be enough.
After Cannes, the Real Work Begins
Cannes is not the transformation. Cannes is the signal. The transformation happens inside operating models, procurement discussions, client contracts, platform partnerships, creative review processes, production teams, media workflows, measurement systems, and governance frameworks. It happens when AI changes who does the work, how long it takes, what clients pay for, and where accountability sits when automated systems influence campaign outcomes.
Madison Avenue is going all in on AI because it has little choice. Clients want speed, cost reduction, personalization, measurable outcomes, and more content across more channels. Platforms are embedding AI into the ad stack. Competitors are promising new efficiencies. Award culture is normalizing AI-assisted work. Agentic infrastructure is beginning to automate more of the marketing workflow.
The agency business is no longer debating AI from the outside. It is being restructured from within.
For holdcos, the next question is not whether they have AI capabilities. Everyone will claim that. The question is whether they can turn those capabilities into a defensible operating architecture.
That architecture has to protect strategy from automation, creativity from sameness, clients from platform dependence, and brands from uncontrolled synthetic scale. It has to produce faster work without making the work cheaper in every sense of the word. It has to make AI useful without letting AI define the agency’s value.
After Cannes, the industry can no longer treat AI as an innovation theme. It is now part of the business model.