Picture a polar bear gazing, misty-eyed, at a convoy of red trucks rolling through a snowy town. You know this ad. You've known it since 1995. Except in the 2025 version, the bear's fur ripples like wet plastic, the trucks warp at the edges, and the crowd of woodland creatures flickers between photoreal and cartoon in a way that makes your eye itch. Coca-Cola generated the whole thing with AI — roughly 100 people and five AI specialists spinning up over 70,000 clips to homage its own beloved "Holidays Are Coming." The internet called it soulless, creepy, a reason to switch to Pepsi. Pratik Thakar, Coke's head of generative AI, shrugged it off: "The genie is out of the bottle, and you're not going to put it back in."
He's right, annoyingly. The genie is out. Eighteen months ago, the debate inside agencies was whether generative AI could really do the work. That debate is over. It can — the ad above proves a machine can now impersonate one of the most nostalgic pieces of film in advertising history, badly but cheaply, at industrial scale. The interesting question in 2026 isn't whether AI makes the creative. It's what happens to a hundred-year-old business model when the creative is suddenly close to free.
The short version: it breaks. And the thing left standing in the rubble is the one thing the machines still can't do — decide what's worth making in the first place.
The Timesheet Is Dying, and Everyone Knows It
The creative industry has, for a century, sold time. You brief an agency, a battalion of people spend weeks on it, and you pay for those weeks — the billable hour, the retainer, the headcount. Generative AI takes a dynamite stick to that entire arrangement, because it compresses the weeks into hours. A 2025 survey of more than 180 agencies by Productive found AI cutting deliverable timelines by three to four times at some creative shops. Same output, a fraction of the billable hours. If you charge for time and the time collapses, so does your revenue — even when the work is identical.
The holding companies have stopped pretending otherwise. In February 2026, WPP's new CEO Cindy Rose unveiled Elevate28, a plan to strip out £500 million in annual costs by 2028 and fold the sprawling empire into four AI-backed units all plugged into WPP Open, its AI platform. Rose's framing was the tell: WPP, she said, is no longer a "holdco" at all. This from the company that more or less invented the modern holding-company model.
The human cost is already on the books. WPP shed 7,000 roles in the back half of 2025 — 700 from Ogilvy alone. Dentsu committed to cutting 3,400 jobs. Interpublic laid off 3,200 as it was swallowed by Omnicom, a merger expected to find $750 million in savings — and labor, as always, leads the balance sheet. Forrester's Jay Pattisall reckons agency headcount fell 8% in 2025, and his team now forecasts a 15% cut to the U.S. advertising workforce by the end of 2026. "This generation of AI," Pattisall notes, is arriving faster than his own models predicted, because the 2023 forecasts "had not factored for agentic" tools — the AI that doesn't just draft but executes.
Layoffs were "the new normal," one laid-off Jellyfish exec told Digiday, describing a job market where a decade ago you could "walk out of a job and walk into five interviews the following day. Now, I can't get the interview." Restructuring and soft client spending did most of this year's damage. But as Pattisall puts it, generative AI is "the ghost at the leaving party."
The Micro-Agency Math That Terrifies the Giants
Here's what should really keep a holding-company CFO up at night: the economics now favor the small. Way small.
A traditional 50-person agency runs on 15–20% margins and something like $150,000 of revenue per employee, dragged down by $300,000–$500,000 in annual overhead. An AI-native micro-agency — a handful of senior people with AI doing 70–90% of the production — operates on overhead measured in single-digit thousands, posts margins of 50–80%, and can clear $500,000 to over $1 million in revenue per head. The senior people aren't cheaper; there are just radically fewer of them, and each one is worth three of the old model.
This isn't a niche curiosity. It's the same physics playing out across the whole software-adjacent economy — the AI code editor Cursor reportedly crossed $1 billion in recurring revenue with a team you could fit in a mid-size conference room. The lesson bleeds directly into creative services: the winning structure is a small, senior team where humans act as interpreters between the client's intent and the machine's output, work is priced by outcome rather than hour, and the deliverable is bought, not built.
For the boutiques and freelancers the 2025 version of this story worried were being squeezed out, that's a genuine reprieve. The tools that used to be the exclusive kit of a big agency — the render farm, the edit suite, the army of junior designers — now cost a subscription. The gap between what a three-person shop and a three-hundred-person shop can physically produce has nearly closed. What hasn't closed is the gap in what they can think up. Hold that thought.
Mediocrity at Scale
If everyone has the same near-free generation engine, everyone starts making the same stuff. The industry even coined a term for it, and by June 2026 the term had a stage.
At Cannes Lions 2026, the world's biggest advertising festival, the loudest conversation wasn't celebratory — it was a reckoning over quality. Cannes Lions CEO Simon Cook stood on the main stage and warned that the industry now lives in "a world defined by mediocrity at scale," where AI risks "diluting distinctiveness, emotional relevance, and stories that cut through." The phrase circling every panel was "AI slop." Ogilvy literally put "Slop Age" in its session titles.
The most pointed shot came from Publicis CEO Arthur Sadoun, who released a 145-second satirical mock-documentary called "The Wrong Promises" just before the festival — anonymized clients recounting agency pitches where they were offered a $5-million bonus to adopt an AI platform, or told they wouldn't pay "until we win you a gold Lion." Sadoun's read: "The compound effect of over-promising on AI and unsustainable commercial offers in pitches to generate headlines is leading to massive job cuts in our industry. Collectively, we have to stop this race to the bottom." To underline it, Publicis sent 40% fewer staff to Cannes and cut its own award entries. When the critics start giving up headcount and trophies, the critique is worth taking seriously.
The audience punishment is real and immediate. When Samsung used AI-generated footage to tease the Galaxy S26 in February 2026 — for a phone whose entire pitch is its camera — viewers clocked the fakery on sight and roasted it. The pattern is consistent: audiences notice when AI is used to fill a gap the idea should have filled.
Cannes wrote the backlash straight into its rulebook. Entries dropped 25.5% — from 26,900 to 20,050 — after the festival introduced tougher rules demanding proof of impact and requiring every entry to be personally endorsed by the submitting agency's CEO and CMO. Tie the work to a named executive and AI fakery becomes a personal risk, not a faceless one. There's even a new AI Craft subcategory in the Film Craft Lions, which asks entrants to prove AI actually changed what the execution could do, rather than just cheaply faking what a camera would have caught anyway.
The Humans Strike Back (and Charge More)
Now the twist. In the same week the industry was flagellating itself over slop, the campaigns actually winning juries were conspicuously, defiantly human. Nestlé's KitKat "The Heist" — a caper about a vending machine and a break room. Lay's roping football stars into a WhatsApp group chat. Ideas that work on their own terms, where the AI in the production, if any, is beside the point.
This is the counterintuitive lesson of 2026: the more AI commoditizes execution, the more the un-automatable stuff — taste, judgment, cultural intuition, a genuinely surprising idea — commands a premium. When a machine can produce infinite competent mush, competent stops being worth anything and distinctive becomes the whole game. A McKinsey report released at Cannes put a number on the gap: 90% of CMOs are experimenting with AI, but fewer than 10% have scaled it or captured real value. Everyone has the engine. Almost nobody has figured out how to point it somewhere interesting.
The agency leadership class, to its credit, has largely landed in the same place — pro-AI, anti-autopilot. Across the 4A's Look Ahead 2026 interviews, Dentsu Creative's Abbey Klaassen warned about brands over-optimizing for algorithms and drowning in a "sea of sameness." WPP's Michael Frohlich cautioned that agencies risk losing their "heart and soul" if the tech becomes the point rather than the enabler. Barkley OKRP's Jeff King, a self-described technology optimist, said he still "bets on insights and creativity." The consensus: let AI take the friction out of the work so people spend more time thinking — not less.
But thinking has to be sold differently, because you can no longer bill it by the hour. Industry sage Rishad Tobaccowala told the 4A's the survival formula is to "focus like crazy and partner like mad" — specialize hard, automate the obvious, and stop forcing juniors to "pay their dues" doing production a machine now does better. CMO Vineet Mehra was blunter: agencies "need to stop selling time and start selling outcomes." WPP has already moved a chunk of its revenue to performance-linked fees. The retainer isn't dead, but the logic underneath it — you pay us for the hours we pour in — is finished. You pay now for the judgment about where to pour.
The Brief Is the Only Moat
So here's where the business of creativity actually landed in 2026. The machines can make the ad — they proved it on Coca-Cola's own Christmas trucks. Production, the thing agencies sold by the hour for a hundred years, is racing toward zero. Headcount is following it down. The safest place to stand is no longer inside a big building full of billable people; it's a small, senior team that knows exactly what to ask the machine for.
Because the tool is now available to literally everyone, and the tool is not the edge. As DesignRush put it, surveying the wreckage of the slop debate: everyone has the same generator, "but the brief is not" the same. The brief — the taste to know what's worth making, the judgment to kill the mediocre version, the cultural read that turns a competent execution into one people actually feel something about — is the last thing that hasn't been commoditized. It might be the last thing that can't be.
The uncomfortable part, if you make brands and creative work for a living, is that this cuts both ways. The floor has dropped out of "competent," which is terrifying if competent was your whole offer. But the ceiling on "distinctive" has never been higher, because you can now get any idea into the world for pocket change — if it's a good enough idea to be worth the electricity. The genie really is out of the bottle. The question the industry spent all of 2026 learning to answer isn't what the genie can build. It's whether you had anything worth wishing for.