On Meta’s third-quarter earnings call Wednesday afternoon, the company’s CFO read two numbers into the record, and most of the ad industry only heard the first one.
The first: nearly 10 million advertisers used at least one of Meta’s generative AI ad tools last quarter. Advertiser adoption of video generation grew 25% in three months. The machine that writes your headlines, extends your images and slices a fifteen-second spot into nine aspect ratios is no longer a beta feature for the brave. It’s the factory floor. Everybody clocked in.
The second number came a few breaths later. Partnership ads, the format that runs through a creator’s handle with a real person’s face on it, grew more than 97% year over year. Susan Li put it plainly: the product “surpassed $15 billion in annual revenue run rate.”
Read those two lines side by side and half the strategy decks in this industry need a rewrite.
When ten million businesses have the same tool, the tool stops being an advantage. It becomes the floor. Nobody brags that their store has electricity. Nobody wins a pitch because their deck has slides. Generative creative crossed that line faster than any production technology we’ve watched arrive, and Meta just confirmed it on an earnings call, the one stage in corporate life where exaggeration carries legal risk.
So what’s actually compounding? Not the generated variant. The person.
Partnership ads nearly doubled in a year when Meta spent most of its stage time selling automation. The company now counts more than 10 million creators as discoverable in its creator marketing hub. The business that built the most aggressive ad-automation stack on earth is quietly making a second fortune from the one input it can’t generate: a human being an audience already believes.
That’s not a contradiction. It’s the business model.
Think about how a feed works for the person holding the phone. Somebody in Silver Lake is scrolling at 11 p.m. Between a friend’s engagement announcement and a clip from a comedy set she missed last weekend, an ad appears. If it looks like every other ad she’s seen this week, smooth, flawless, faintly uncanny, her thumb has moved before her brain registers the logo. If it’s a face she follows, talking the way that face always talks, she stops. Not because the production is better. Usually it’s worse. She stops because there’s a relationship there, and relationships are the only thing in the feed that can’t be reproduced on demand.
The consumer data says the same thing, louder. Gartner surveyed 1,006 U.S. consumers this summer and found 65% think brands produce too much AI-generated content, and 57% say it has made them less trusting of brand messaging overall. The kicker: 43% say they now rely more on real people for shopping advice because of AI content. A separate Clutch survey of 601 consumers in August found 53% are less likely to buy from brands they know use AI in social content, and 71% would question a human influencer’s recommendation if AI made the post.
Put that together. The supply of generated creative is going vertical. The demand for it, measured in trust, is sagging. Economics has a word for what happens to the value of a thing when supply explodes and demand falls. We’ll let you finish the sentence.
And the bill for being ignored is going up. On the same call, Meta said ad impressions rose 10% year over year while the average price per ad rose 14%. Read that from the advertiser’s chair: you are paying more for every slot, and every slot now sits next to a near-identical generated ad from a competitor using the same tools. An impression that earns a thumb-flick instead of a pause isn’t neutral. It’s a more expensive way to be forgotten. Generated sameness doesn’t just fail quietly. It fails at a premium, and the auction keeps the receipt.
Here’s where the AI-first crowd gets loud, and they’re not entirely wrong.
The machine works. Meta’s call was full of small, real gains: a generative recommender lifting Facebook feed conversions by nearly 2% in targeted segments, an AI business assistant whose adopters drove almost 5% more conversions at the same cost per conversion. Those are real dollars. Any brand ignoring them is lighting money on fire, and any agency telling clients to boycott the tools is selling nostalgia with a day rate.
We use the tools. We’d be idiots not to. Resizing, versioning, cutting one hero into twenty placements, testing which first frame holds a thumb: let the machine eat that work. It’s tedious, it’s expensive when humans do it, and nobody went to art school to export 4:5 crops.
But look at the size of those gains, then look at the other number. Optimization is buying low single digits. The format built on a person grew 97%. One is a better engine. The other is a different fuel.
This is where most brands are about to get it exactly backwards.
They’ll read “10 million advertisers” as permission to cut the creative budget, drop the photographer and let the platform generate its way to scale. They’ll end up with an ad account full of competent, interchangeable assets, running through the same tools as ten million competitors, all optimizing toward the same average. The ad gets cheaper to make. It also gets cheaper to ignore.
The smart money is moving the other way. Spend less on the production overhead the machine can absorb. Spend more on the parts it can’t touch: the creator who actually has an audience, the director who can make a neighborhood shop feel like a scene, the strategist who decides which human story deserves to be scaled in the first place. The machine scales distribution. People supply the trust. You need both, and only one of them is scarce.
There’s a quieter implication in Meta’s numbers that should make in-house teams squirm. Partnership ads aren’t just influencer posts with a boost button. They are a brand renting someone else’s credibility because its own account can’t carry the message alone. Every dollar in that $15 billion run rate is, at least partly, a confession.
Renting trust is a fine tactic. It’s a terrible strategy. Rent credibility forever and you pay rent forever, at whatever price the landlord sets next year. The brands that win the next three years will borrow trust from creators while building their own: founders on camera, staff with actual personalities, a recognizable voice that doesn’t sound like it was approved by eleven people and a prompt.
That’s the work we do, and it’s why we’ve never pitched anyone on volume. A feed full of generated variants is a lottery ticket. A face people recognize, used on purpose and backed by a machine that handles the grunt work, is an asset that appreciates.
Mark Zuckerberg spent part of the same call claiming Threads has overtaken X in monthly actives. Platforms rise and fall. Formats get renamed. Tools get commoditized in eighteen months, as this one just did. The only thing that has survived every cycle is the reason anyone looks at a screen in the first place: to see somebody they give a damn about.
Ten million advertisers have the same machine now. Congratulations to all of them.
The edge was never the machine. It was the face. Meta just put a price tag on it, and the number is still climbing.
