Why is Meta's advertising booming while Google's is slowing down amidst the same AI wave? Google's Q3 earnings report revealed a major shortfall in its advertising business, Miss (already highlighted in yesterday's subscription channel). Search advertising revenue reached $63.3 billion, a 12% year-over-year increase, but below market expectations of $65 billion, causing the stock price to drop 10% in after-hours trading. Surprisingly, just a week earlier, Meta reported advertising revenue of $63.1 billion, a 26% year-over-year increase, with impressions up 10% and the average cost per impression (APM) up 14%, showing growth in both volume and price. Both companies, with similar size and leading the rapid advancements in the AI era, are experiencing such vastly different growth rates. Why is there such a significant difference? 1. The core reason: AI is impacting the two companies in opposite directions. For Meta, AI directly impacts the auction process. Better conversion predictions, Advantage+'s automated delivery, and generative creatives all make the same impression slot more valuable to advertisers, leading them to be willing to pay higher prices. This explains the 14% increase in the APM. AI didn't change Meta's product form; it merely made recommendations and ad placements more accurate, so it was purely a tailwind. For Google, AI primarily changed the product form itself. AI Overviews and AI Mode transformed ten blue links into a direct answer, reducing the number of times users clicked and decreasing the ad placements per query, naturally putting pressure on monetization. Google isn't incapable of using AI for ad optimization; Schindler discussed the improvements Gemini made to ad quality and AI Max in the conference call. However, these gains were offset by the drag from the product form shift. Simply put, Meta is upgrading its engine with AI, while Google is replacing its chassis with AI. It's normal to experience slower performance when replacing a chassis. 2. The elasticity of ad inventory is completely different. Meta sells attention; its inventory unit is impressions in the news feed and Reels. This inventory can be expanded through algorithms to increase the duration of recommended content, increase ad load, and explore new areas (Reels, Threads, WhatsApp Business Messaging). Therefore, Meta achieved a 10% increase in impressions and a 14% increase in average cost per impression this quarter, with both volume and price rising. Google Search Ads sell intent; the inventory unit is a query with commercial intent. This inventory is not created by Google itself, but determined by whether the user currently intends to buy something. Traditional search results page ad slots are already saturated; incremental growth can only come from more commercial queries. The biggest problem this quarter lies precisely here: query growth is the highest in five years, but a large portion of the incremental queries fall into AI Mode, which management admits has a lower monetization rate. In other words, Google's inventory is increasing, but the new inventory hasn't yet been put on the shelves. 3. Different Data Loop Thickness Meta's advertisers are largely e-commerce and DTC brands. Their conversion events flow back to Meta through the Pixel and Conversions APIs. Combined with Advantage+ shopping ads completing the loop from impression to purchase directly within the app, Meta's observation of whether an ad generates an order is becoming increasingly direct. Google Search's signal is the query term itself. This signal is strong but one-off. What happens after a user clicks is less complete for Google than its meta search. While AI answers absorb some clicks, the traffic Google sends to advertisers' landing pages continues to decrease, further weakening its ability to demonstrate ROI to advertisers. 4. AI Mode's monetization is still in its early stages, but its potential may be higher. Traditional search offers only one advertising opportunity per query; users leave after viewing the results page. AI Mode involves multi-turn dialogues. From "want to buy a laptop" to "budget under $1000" to "which one has the best battery life," each turn represents a new expression of intent. Multi-turn advertising essentially breaks down a single conversation into multiple auctions. Google says AI Mode queries are on average three times longer than traditional searches, one-fifth of queries include voice, images, or documents, and most AI Mode queries are incremental. These queries carry much richer intent information than keywords, theoretically meaning the advertising value per query should be higher, not lower. AI Mode's dialogue naturally pushes users from the upper funnel to the lower funnel. Traditional search excels at catching up with those who have already decided to buy, while AI Mode can catch up with those still comparing and debating. This decision-making process used to belong to platforms like Xiaohongshu, Reddit, and YouTube reviews, and Google now has the opportunity to reclaim it. Of course, the obstacles are also obvious. A generated answer can naturally have fewer ad placements than ten blue links per page, and user trust is Google's weaker point compared to Meta. People are used to ads in news feeds, but inserting ads into an AI answer that appears to be neutral advice damages the credibility of the search engine itself. There's also the issue of self-cannibalization. The process of AI Mode catching up in monetization is simultaneously a process of traditional search being replaced; the net increase depends on whether the former's growth rate can outpace the latter.