AI Gave Every Marketer the Same Content Engine. Brand Is the Only Moat Left.

In April 2025, Ahrefs pointed its crawler at 900,000 newly published web pages and found AI-generated text on 74.2% of them. One page per domain, 900,000 domains, and three of every four already carried machine-written copy.

That number is the whole strategy problem in one line. The thing most marketing teams spent a decade getting good at, producing more content faster, is now something a 20-person team and a 2,000-person team can both do before lunch. The skill that used to separate a good content operation from a mediocre one has a subscription price and a flat learning curve. Content has stopped telling anyone apart.

Execution went to zero. The advantage went somewhere else.

The demand side confirms it. In July 2025, Pew Research studied the real browsing of 900 U.S. adults and found that when a Google result carried an AI summary, people clicked a traditional link in just 8% of those searches, down from 15% when no summary appeared. Clicks on the links inside the summary ran at 1%. The ranking you fought for still exists. The visit it used to earn does not. AI search is draining the organic traffic that funded a generation of content teams.

Paid is not the escape hatch either. Benchmarkit’s 2025 survey of SaaS companies put the median cost to acquire $1.00 of new customer ARR at $2.00, a 14% jump in a single year. Bottom-quartile companies pay $2.82 for the same dollar. When everyone owns the identical content engine and rents the identical ad platforms, they bid against each other for a shrinking pool of attention, and that auction moves one direction.

Here the strongest objection arrives. In May 2026, McKinsey published a widely shared argument that the durable AI moat is proprietary data that improves with use, and that advantage is moving away from brand toward data loops rivals cannot copy. McKinsey is right about Amazon. Amazon sees every search, purchase, and return, and that flywheel compounds into a genuine moat.

Most marketing teams do not have Amazon’s flywheel. A 25-person B2B software company does not sit on a closed data loop that gets smarter each quarter. What it can own is older and narrower than a data moat, and it happens to be the two things a commodity content engine cannot manufacture: a brand people recognize, and an audience it can reach without paying a platform for permission.

Why do those two survive the commodity crush? Because both are memory, and memory does not scale with compute. A brand is the set of associations already sitting in a buyer’s head before any search happens, which lets it skip the auction entirely. An owned audience, an email list, a subscriber base, a community, is a distribution channel you are not renting, so no algorithm change and no AI summary can revoke it overnight. AI can write a million versions of your blog post. It cannot make a stranger trust you, and it cannot hand your competitor your subscriber list.

The practical move is a reallocation, not a retreat. The commodity layer still has to run, because table stakes are still stakes, and a team that stops publishing disappears faster than one that publishes sameness. Every extra hour spent out-producing rivals on content that reads exactly like theirs is an hour bidding in a market that resets to zero every morning. The teams pulling ahead are pushing effort up the stack, into the parts of marketing that were always hard and are now the only parts still scarce: a point of view, a voice a reader can place blindfolded, direct relationships that no intermediary controls. Owned media is one version of that bet, which is why some companies now build publications as marketing infrastructure.

So here is the verdict, and the three moves under it.

Stop grading your content team on volume. Volume is the commodity now, and the only metric that matters is whether a reader could tell it was you before they saw the logo.

Fund the audience you own before the audience you rent. A subscriber costs more than a click this year and outlasts every algorithm change next year.

And put brand on the performance line, not the soft one. In a market where execution is free and attention is not, the recognizable name wins the auction it never has to enter.

The moat did not vanish. It moved to the one place AI cannot follow, the gap between your name and what a buyer already believes about it.