Alphabet Q3 Ad Revenue Hits $127.6B But Earnings Call Ignores Ads

Google booked $127.6 billion in ad revenue last quarter, up 25%, and the earnings call barely mentioned it. Sundar Pichai spent his time on Gemini hitting a billion users and 32 billion tokens a minute flowing through Google's AI APIs. Search grew from $56.6B to $63.4B. YouTube hit $11.2B. The money machine got a paragraph; the AI story got the room.

The cynical read is narrative management. The ad business is mature and needs an AI wrapper to defend the multiple. CFO Anat Ashkenazi's pivot to the AI story landed at exactly the moment Search faces its first real threat from answer engines. "AI Mode click-through on par with traditional search" is the claim doing all the work, and Google runs the measurement and reports the result. Thirty-two billion tokens a minute tells you nothing about whether that inference is making money or burning it.

The more charitable read is that the token volume is the real fact, and it only works because Google owns the whole stack: custom TPU chips, custom networking, custom serving. Their cost per token at that volume sits structurally below anything a provider renting NVIDIA hardware can match. More important, Google spreads that inference cost across a Search product that already reaches billions of users at near-zero marginal cost. No rival hyperscaler has that front door, and that math compounds as volume grows.

The empirical result buried in Chief Business Officer Philipp Schindler's comments is worth taking seriously. AI Mode queries run three times longer than standard search queries and still produce click-through rates on par with regular search ads. Longer, messier queries usually mean fuzzier intent and worse targeting. Google is saying the opposite held at scale. If that replicates, the shift to conversational search does not wreck ad economics the way the bears assumed. The caveat: we only have Google's number, and "on par" is carrying the whole claim.

For advertisers, the global rollout of AI Mode ads is the immediate operational fact. Google generated 1.5 billion creative assets with Gemini last quarter and is folding that into existing ad workflows. The problem is control. When Google generates your creative and decides where the ad lands inside an AI-written answer, you lose visibility into placement and you take Google's word on performance. Brand-safety and indemnification questions on AI-generated placements are unanswered, and they will stay unanswered until something goes wrong publicly.

Everything hinges on one question: does conversational AI search monetize as well as the ten blue links it replaces? Google just said yes, on its own numbers. If true, the answer-engine threat is smaller than feared. If "on par" turns out to mean "on par on a hand-picked slice," then this was a maturing ad business buying time with a story. No one outside Google can settle it from this release.

The call: Google Search advertising revenue for Q4 2026 comes in above $63.4B, reported on the late-January 2027 earnings call. Confidence: medium. AI Mode ads are rolling out globally into the biggest ad-spend quarter of the year. The only scenario where Search revenue falls quarter-over-quarter into the holidays is if AI Mode cannibalizes paid clicks faster than seasonal demand adds them, and nothing in this release suggests that is happening yet.

Google booked $127.6 billion in ad revenue last quarter, up 25%, and spent almost none of the earnings call talking about it. Sundar Pichai talked about Gemini hitting a billion users and 32 billion tokens a minute flowing through Google's AI APIs. Search ads grew from $56.6B to $63.4B, YouTube hit $11.2B, and the money machine got a paragraph while the AI story got the whole room.

This is a briefing, not a decision you make. Nothing here is yours to undo. What's actually being decided is whose version of "what Google is" the market believes going into 2027, and who has to adjust their own plans because of where Google is pointing its inference capacity. The thing that sets any clock is the next earnings call and the global rollout of AI Mode ads, which Google is pushing live now.

The Skeptic. Twenty-five percent growth gets one paragraph and Gemini gets the call. That is narrative management. The ad business is mature and needs an AI coat of paint to defend the multiple. "AI Mode click-through on par with traditional search" is a soft claim doing heavy work. On par with what baseline? Measured how? Over what window? Google runs the measurement and Google reports the result. And 32 billion tokens a minute tells you nothing about whether that inference is making money or burning it. Anat Ashkenazi pivoted to the AI story exactly when Search faces its first real threat from answer engines. That timing is the point.

The Compute Pragmatist. Thirty-two billion tokens a minute, sustained, is the real fact here, and it only works because Google owns the whole stack. Custom chips (TPUs), custom networking, custom serving. Their cost per token at that volume sits structurally below anything an outside provider renting NVIDIA chips can match, and the gap widens as volume climbs. The move that makes AI Mode affordable: Google spreads that inference cost across a Search product that already reaches billions at near-zero cost to add one more user. No rival hyperscaler and no inference startup has a billion-user front door to spread the cost over. Whoever you buy inference from is competing against that math and losing on it.

The Researcher. Ignore the revenue and look at the empirical result buried in Philipp Schindler's comments. AI Mode queries run three times longer than normal search queries and still produce click-through rates on par with regular search ads. That is genuinely interesting. Longer, messier queries usually mean fuzzier intent and worse targeting. Google is saying the opposite held at scale: people typing more got ads just as relevant. If that replicates, it means the shift to conversational search does not wreck ad economics the way the doomers assumed. The caveat is the same one the Skeptic raised. We only have Google's number, and "on par" is carrying the whole claim.

The Enterprise Buyer. For anyone buying Search ads, the global rollout of AI Mode ads is the thing that touches your account. Google generated 1.5 billion creative assets with Gemini last quarter and is folding that straight into existing ad workflows at no extra friction. Attractive if you run lean on creative. The worry is control and measurement. When Google both generates your creative and decides where the ad lands inside an AI-written answer, you lose visibility into placement and you take Google's word on performance. The indemnification and brand-safety questions on AI-generated placements are not answered, and they won't be until something goes wrong publicly.

The Safety Lens. A billion and a half AI-generated creative assets in one quarter is past any human review pipeline. Google is implicitly betting its automated filters hold at that volume. Longer AI Mode queries also widen the opening for someone to manipulate what ads get retrieved by stuffing the query. None of this came up on the call, which at the moment of maximum deployment is itself the signal. Regulators tracking the EU AI Act should recognize that Google's AI search is now at a scale where one filter failure becomes a systemic event across billions of impressions.

Where they disagree. The Compute Pragmatist and the Skeptic are looking at the same 32 billion tokens and drawing opposite conclusions. The Pragmatist sees a cost advantage nobody can touch. The Skeptic sees a number designed to impress precisely because it hides the margin. Both can be right: the per-token cost can be the lowest on the planet and the AI Mode product can still be dilutive to margins today, because volume at low margin still drags blended profitability. The second split is Researcher versus Buyer. The Researcher is excited that longer queries kept ad relevance intact. The Buyer only cares whether that holds on their account with their budget, and Google controls the only scoreboard that would tell them.

What it hinges on. One belief carries most of the weight: does conversational AI search monetize as well as the ten blue links it replaces? Google just claimed yes, on its own numbers. If that is true, the answer-engine threat to Google's core is smaller than feared and the AI pivot is real. If "on par" turns out to mean "on par on a hand-picked slice," then the Skeptic wins and this was a maturing ad business buying time with a story. Nobody outside Google can settle it from this release. Whether Search ad revenue keeps growing after AI Mode ads go global is the only measurement Google cannot narrate around.

Prediction: Alphabet's Q4 2026 Google Search and other advertising revenue, reported on the late-January 2027 earnings call, will come in higher than the $63.4B it posted in Q3 2026.

Confidence: Medium. AI Mode ads are going global into the seasonal Q4 ad peak.

Why: Google just told us two things that point the same way: AI Mode queries match regular search on click-through, and the AI Mode ad format is rolling out worldwide right now. Q4 is the biggest ad-spend quarter of the year because of holiday retail, so Search ad revenue almost always steps up from Q3 regardless. Layering a globally expanding, already-monetizing AI ad surface on top of that seasonal lift makes a sequential increase the strong base case. The opposite outcome, Search ad revenue falling quarter over quarter into the holidays, would require AI Mode to actively cannibalize paid clicks faster than seasonal demand adds them, and nothing in this release suggests that is happening yet. If it did fall, the Skeptic's read would be confirmed in the only number Google cannot talk its way around.

Revisit by 2027-02-15: We're right if Alphabet reports Google Search and other advertising revenue above $63.4B for Q4 2026. We're wrong if that line comes in at or below $63.4B.