“Change is neither good nor bad. It simply is.”— Don Draper, Mad Men (2009)
MrBeast gets more views in a day than the Super Bowl gets in a year. A twenty-seven-year-old with a laptop reaches more humans than the most expensive advertising event in history. Cost of a Super Bowl ad: $7 million for thirty seconds. Cost of MrBeast’s upload: zero. He doesn’t need agencies, media buyers, or approval chains. Just an idea and an internet connection. Now imagine that power multiplied by AI, infinite MrBeasts, each one personalized to your exact sense of humor, your exact attention span, your exact moment of need. That’s not the future. That’s next Tuesday. And every agency in Manhattan is pretending it’s not happening.
Here’s the only math that matters: Creating an ad now costs $0. Making someone see it now costs everything.
The violence of this asymmetry breaks capitalism as Madison Avenue knows it. Google’s CPCs hit $4.66, some verticals seeing $40+ per click. Meta’s ad prices jumped 10% while organic reach hit 0.01%. TikTok’s CPMs doubled to $10 while throttling anything that smells like marketing. Meanwhile, OpenAI just dropped Sora - text-to-video that generates Super Bowl-quality commercials in 60 seconds for $200/month. Read that again: SUPER BOWL-QUALITY VIDEO FOR $200/MONTH. What used to cost Ridley Scott $10 million now costs less than your Netflix subscription. Claude writes better copy than your creative director. Midjourney creates award-winning visuals faster than you can describe them. Sora makes finished commercials while agencies are still arguing about the storyboard.
Last week, a solo founder generated 100,000 personalized video ads, each one unique to the viewer’s browsing history, deployed across every platform, evolving in real-time based on engagement. His tools: Sora for video ($200), Claude for copy ($20), Midjourney for stills ($30), and Motion for performance optimization ($250). Total monthly cost: $500. Total views: 200 million. His competitor, a Fortune 500 brand, spent $50 million with BBDO on a campaign that reached 10 million people. The founder is twenty-two. Never worked in advertising. Never even used After Effects. He just typed “luxury perfume ad, golden hour, Sophia Loren energy, but make it Gen Z” into Sora and got something better than what David Fincher directed for $15 million. The founder iterated 10,000 times before breakfast. Fincher’s still in pre-production.
The Panic
Watch the holding companies panic. $30 billion mergers disguised as “AI transformation.” 70,000 employees billing $500/hour for work that happens in milliseconds. Acquiring AI startups that do what ChatGPT does for free. Legacy agencies are merging to survive, mistaking consolidation for innovation, confusing press releases for progress. Yesterday, OpenAI released Sora to the public. Today, every production house in LA is questioning their existence. Tomorrow, they’ll be listing their RED cameras on eBay.
They think they’re integrating AI. They’re actually performing their own autopsy. Their entire business model, billable hours, creative scarcity, campaign cycles, assumes humans are the bottleneck. But humans aren’t the bottleneck anymore. Attention is. And you can’t bill hourly for infinite supply meeting infinite demand.
I’ve seen this movie. At POPSUGAR, we scaled to 100 million monthly readers, then watched Meta throttle our organic reach from 20% to 3.8% overnight. One algorithm update. Eight figures in revenue gone. We learned the brutal truth: you either own the distribution or you ARE the distribution. Agencies own neither. They’re middlemen in a world that just eliminated the middle.
The Precedent
This is Kodak, but faster. When digital photography emerged, Kodak owned the patents, they invented the digital camera in 1975. But their business model depended on selling film, so they buried the technology. They thought they were in the film business when they were actually in the memory business. The agencies think they’re in the creative business when they’re actually middlemen selling attention. And the middleman just became obsolete.
Remember when travel agents insisted the internet was “just another booking channel”? They’re all dead now. Agencies are making the exact same bet.
The winners never debate whether the change is real. They debate how fast they can ship.
The Anti-Agency
Here’s what replaces them: The Anti-Agency. Not a company but a living system. A swarm of autonomous agents generating millions of creative variants, deploys them across every channel, measures engagement in microseconds, evolves the winners, kills the losers, and repeats. Forever. No humans in the loop. No meetings. No pitches. Just an algorithm that gets smarter with every impression.
Once creative becomes computation, the agency becomes an API.
Picture this: Sora generating a thousand video variants, different actors, different settings, different emotions, all from a single prompt. Each one A/B tested in real-time. The winners breed with each other, creating hybrid variants. The losers die. By lunch, you’ve evolved through more generations than Darwin could imagine. By dinner, you’re outperforming every hand-crafted campaign on the internet. No director. No DP. No post-production house charging $10,000/day for color correction. Just prompts evolving at the speed of compute.
Brands won’t hire agencies. They’ll subscribe to software that generates 10,000 ads before your morning standup. Instead of quarterly campaigns, they’ll deploy creative that evolves every second, killing what fails, scaling what works. The winners won’t be those with the best ideas but those with the fastest evolution velocity. How quickly can your message adapt? How fast can your creative mutate? How many generations can you cycle through while your competitor is still in their kickoff meeting?
The CMO of 2026 won’t manage agencies. They’ll manage mutation rates and orchestrate agent armies. Brand guidelines won’t be PDFs; they’ll be training parameters. Creative briefs won’t be documents; they’ll be prompts. And the entire $700 billion agency industry? It’ll be a GitHub repo.
The Investment Thesis
At Sugar Capital, we’re funding the assassins, not the ambulances. We’re not investing in tools that help agencies “integrate AI” or “streamline workflows.” That’s hospice care for a dying industry. We’re funding what replaces them entirely, systems that think faster, platforms that turn every brand into its own self-improving creative organism. The agency of tomorrow won’t have employees or offices or pitch decks. It’ll be a server farm running thousands of AI agents, each one generating and killing campaigns every second, learning what works, forgetting what doesn’t. By the time you refresh your feed, it’s already evolved.
Your new creative director doesn’t sleep. It doesn’t take meetings. It doesn’t need awards shows to validate its existence. It just needs compute and data.
The martini lunch is over.
The algorithm already ate Don Draper for breakfast.
And it’s still hungry.
Great writing, sir! As for the investment thesis...there is some (a lot of truth) but reality is slightly more nuanced. Slop can be automated/AI-fied and since slop is 95% of what's out there, well, the business case is easy. There is still a 5%, high-margin, niche where creativity and taste (advertising holdcos have neither, so they are dead anyway) matters.
The other - and far more important - comment is that your case is built for a world in which communication works predominantly through videos/images, propagated through paid adv. Paid adv is a joke (you write as much yourself, mentioning costs) and adv is a dying form, automated or not. Influencer marketing, events (to use very pop and all-encompassing labels) are where it's at. So maybe the market for slop is not the only market, not even a growing one...although I agree that, within that market, AI will probably kill agencies (unless we all discover that the cost of running data centers makes the whole business case untenable).
Outstanding writing once again B. Bravo.