Alongside its Q4 2025 results on January 28, 2026, Meta published a newsroom report titled “2026: AI Drives Performance,” laying out what AI is actually doing to ads and engagement. The hardest number in it: the combined revenue run-rate of its advertiser-facing video generation tools hit $10 billion in Q4, with quarter-over-quarter growth running nearly three times faster than overall ads revenue. Against persistent questions about Meta’s AI capex, the message is blunt — AI has stopped being a cost story and become a revenue engine, and the part of the business closest to cash, ad delivery, is where the payback shows up first.
A $10 Billion Run-Rate for Video Generation
- The video generation tools advertisers use to create video creative and ad variants reached a combined $10 billion revenue run-rate in Q4 2025 — a figure confirmed on the earnings call itself.
- Quarter-over-quarter growth is running nearly three times faster than overall ads revenue: advertiser adoption of generative creative is still accelerating, not plateauing.
Against mid-single-digit growth for the ads business as a whole, a $10 billion line expanding at three times the speed is Meta’s strongest argument that its AI spending pays for itself. It also marks how fast the shift happened: generated video creative went from experiment to default workflow for mainstream advertisers in roughly two years.
Ranking Models Pay: GEM, Lattice, and a New Instagram Model
- The GEM ads ranking model, after its training GPU count doubled, delivered a 3.5% ad click lift on Facebook and more than a 1% conversion gain on Instagram.
- A new Instagram run-time model added a 3% conversion rate increase, while changes to the Meta Lattice architecture drove a 12% improvement in ads quality.
- These percentages, multiplied across Meta’s ads base, are billion-dollar annualized effects.
Worth noting is the method: the gains come not from swapping in a bigger model, but from engineering iteration on ranking infrastructure — doubling compute, tuning architecture, optimizing where models run. That is an unglamorous, compounding bet on infrastructure rather than a launch event, and it is the part competitors will find hardest to copy quickly.
Attribution and Business Messaging: The Other Billion-Dollar Lines
- Incremental attribution — versus standard attribution — drove a 24% increase in incremental conversions and reached a multi-billion-dollar annual run-rate seven months after launch.
- US click-to-message ads revenue grew more than 50% year over year, boosted by the Website to Message ad format, and paid WhatsApp messaging crossed a $2 billion annual run-rate.
- Business AIs in Mexico and the Philippines are handling more than 1 million conversations per week — AI sales agents carrying real traffic, not a pilot.
The shared logic across attribution and messaging: AI’s value is not only generating content but proving that ads actually work — and that proof matters most for retaining small advertisers.
The Engagement Side of the AI Dividend
- Facebook ranking changes drove a 7% lift in views of organic feed and video posts, and its surfaces now deliver over 25% more same-day Reels than in Q3 2025; Threads saw a 20% lift in time spent; 75% of Instagram recommendations in the US now come from original posts.
- Meta AI daily actives generating media tripled year over year, and AI dubbing now covers nine languages.
Engagement gains feed back into the ads side: more watched time and better recommendations directly grow the inventory of monetizable impressions.
How to Read This Report
Three cautions. First, a run-rate is an annualized extrapolation, not booked revenue — and Meta is skilled at packaging algorithmic iteration as an “AI story,” so trust the direction and discount the magnitude. Second, these gains come from a combination of ranking, attribution, and creative generation, not a single AI product; for most companies the copyable part is the workflow, not the model. Third, the advertiser takeaway is direct: generated video creative and variant testing are now standard equipment in mainstream buying, and lagging is a cost. That tracks what we expected in our 2026 opening outlook: AI payback is moving from chips and cloud toward the links closest to cash — ads and business messaging.
Sources
- 2026: AI Drives Performance — Meta Newsroom
- META Q4 2025 Earnings Call Transcript — Meta Investor Relations
- 5 biggest takeaways from Meta’s Q4 2025 earnings call — Business Insider
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.