eMarketer published its April 2026 forecast and put Meta at $243.46B in worldwide ad revenue for the year, against $239.54B for Google. It is the first time since the early 2010s that Google does not lead the chart. The headline number is the kind of thing finance Twitter writes pair-trade threads about, and several have. The interesting part of the release is not the four billion dollar gap. It is the slope of the lines on either side of it, and what each slope is made of.
Meta’s worldwide ad growth was 22.1% in 2025 and is forecast at 24.1% in 2026. Google’s growth has been roughly flat at 11.9% for two years running. Meta is now projected to hold about 26.8% of worldwide ad spend share in 2026. The two companies are not competing on the same terms anymore, which is the part the headline obscures. They are running two different products through two different bottlenecks, and the bottlenecks have moved in opposite directions.
What Meta actually shipped
The 2026 Meta ad business is mostly not the 2022 Meta ad business with better targeting. The product changed. Three pieces of that change matter, and they compound.
The first is Advantage+ campaign optimization, which has been gradually pulling decisions out of the marketer’s hands and into Meta’s planner. Audience selection, placement mix, budget allocation across ad sets, and creative rotation are now decisions the platform makes. The marketer’s job has shifted toward providing inputs, brand assets, conversion definitions, a budget, and letting the system run. For a comfortable share of advertisers, especially small and mid-market ones who never had a media planner in the first place, that is a strict improvement. The campaigns get better and the labor cost of running them drops at the same time.
The second is generative ad creative inside the Meta tools. The system can now produce variants of a single creative at the placement level, generate alternative copy, restage product images for different markets, and iterate on the variants that perform. The cost of producing the hundredth ad variant has fallen close to zero. That number used to be the actual constraint on creative testing for most accounts. Removing it does two things at once. It widens the funnel of ads any single account is running, which produces more learning per dollar. And it lets advertisers who would never have produced even a tenth variant participate in the testing process at all.
The third is Reels monetization catching up to Stories and feed. Meta spent two years backfilling Reels with ad inventory and conversion data. The result is that the fastest-growing surface in the app now monetizes on roughly the same terms as the rest of it, instead of being a growth-engagement loss leader. Reels’ ad eCPMs are not yet at parity with feed for every category, but the gap has closed enough that the surface is contributing to revenue at the same speed it is contributing to time-spent.
The combined effect is that Meta’s ad business compounds in 2026 the way it did in the late 2010s, except the compounding mechanism is the AI stack rather than the targeting graph. The targeting graph took several thumps from iOS privacy changes and Apple’s ATT framework. The AI stack started compounding right after, on a different axis, and it has done so faster than most analysts expected.
What Google’s chart is actually showing
The Google number is steady, not falling. Search ad revenue in absolute terms is still growing. The growth rate is the issue, and the reason the growth rate is what it is sits in two adjacent product changes.
The first is AI Mode. Google rolled it out as the default reading experience for a meaningful share of queries. The reported numbers from the AI Mode team put it at roughly 75M daily active users and over a billion queries per month. Those are not small numbers, and the per-query monetization on AI Mode answers is structurally different from the ten-blue-links page. The ranked-link layout has been Google’s monetization surface for two decades. The ad slots above and around the organic results are where the budget lands. AI Mode answers do not have ten blue links, which means they do not have the same number of high-intent ad slots. Google has been building shopping and sponsored-citation formats into the answer experience, and those formats are working, but the inventory unit is different and the unit economics are still being calibrated. A query that previously surfaced four sponsored results above the fold now sometimes surfaces zero, sometimes one, and sometimes a different format entirely.
The second is the slow erosion of the top-of-funnel query itself. A growing share of product research, comparison shopping, and how-to navigation is happening inside ChatGPT, Perplexity, and the AI assistants embedded in retail surfaces. Those queries, by volume, are the most monetizable surface Google has ever owned. They are not all migrating, but enough of them are migrating to flatten the curve. YouTube ads and Google’s retail-media business are both growing, and both are partially absorbing the headwind. They are not large enough, fast enough, to hide it on the top-line number.
The summary of those two facts is that Google is converting a smaller share of its query volume into ad inventory than it was three years ago, and the inventory it is converting is yielding less per impression than the legacy ranked-link page did. The product changed underneath the ad business, and the ad business is still being retrofitted to the new product. The retrofit will take time. It is not obvious, on a longer arc, that AI Mode and AI assistants in general can ever monetize at the same rate as the ranked-link page did, because the ranked-link page was uniquely well-suited to high-intent ad placement.
What the chart is and is not predicting
Two things often get conflated when this story is told. The first is whether Meta will keep this lead. The second is whether the underlying shift, AI compressing the marketing labor required per dollar of revenue, is durable. Those are different questions.
On the first, the lead is real but narrow. A four billion dollar gap on a base over $200B is a single quarter of execution. The 2027 chart could put Google back ahead, especially if the AI Mode monetization surface stabilizes faster than current trend lines imply. We would not bet on this being a permanent reversal. We would bet on it being the first of several reversals across the next few years, and a sign that the chart is now more competitive than it has been at any point since Google won it in the first place.
On the second, the more durable point is that AI is doing different work for the two companies. For Meta, AI is shrinking the labor cost of running an ad campaign and widening the population of advertisers who can run one well. For Google, AI is replacing the substrate that ad inventory has historically been priced on. One of those is a tailwind. The other is a substrate change. Tailwinds compound on the existing business. Substrate changes force a rebuild. That asymmetry is the actual structural fact the eMarketer chart is reporting.
The thing worth watching
Two numbers will tell more of the story over the next two years than the gross revenue line. The first is Meta’s revenue per advertiser, especially in the long-tail SMB segment. If that keeps rising, the AI stack is doing what its proponents say it is, and the marketing-labor compression is real. The second is Google’s effective revenue per query in AI Mode versus the legacy results page. If that ratio stops falling and starts climbing back toward parity, the substrate rebuild is on track. If it stays where it is, the chart Meta crossed in 2026 is not the only chart in motion.
The thing the eMarketer release quietly settles is that “advertising platform” and “search engine” are no longer synonyms in the way they have been for fifteen years. They are two different product categories now, with different growth functions and different bottlenecks, and they happen to share a leaderboard. The leaderboard is going to be more interesting as a result, and a lot less stable.