Google stock drops 12% after earnings miss — search slowdown and AI monetization gap spook investors

Google Stock Drops 12% After Earnings Miss — Search Slowdown and AI Monetization Gap Spook Investors

Google's stock cratered 12.5% after a rare earnings miss that exposed two structural vulnerabilities: search advertising growth is decelerating faster than expected, and the company hasn't solved how to monetize AI-generated responses as effectively as traditional ads.

The revenue miss was marginal, but EPS collapsed. Google reported $127.6 billion in revenue against a $128 billion consensus—a negligible 0.3% miss. Earnings per share, however, came in at $2.12 versus the expected $3.01, a 30% shortfall. Even accounting for an $8.7 billion write-down on equity investments (primarily SpaceX holdings, which cost 55¢ per share), adjusted EPS would have been $2.67, still below expectations.

Cloud is propping up the business. Google Cloud delivered $30.7 billion in revenue and $11.1 billion in operating profit, with operating margins at 36.2%—a business so valuable that independent analysts suggest it could be worth hundreds of billions or even a trillion dollars on its own. The cloud backlog stands at $561 billion, with 55% expected to convert to revenue within two years—roughly $275 billion flowing in over the next 24 months. That velocity matters for a segment doing $30.7 billion annually.

Search growth is the real problem. Year-over-year advertising growth slowed to 12% this quarter from 17% last quarter. The culprit is AI search mode, which is accelerating query volume but monetizing poorly. Traditional Google Search has become saturated with ads—the first page often shows ads at the top, followed by three or four sponsored brand-keyword results, then review sites purchased by mattress companies and other advertisers, leaving minimal space for organic results. AI mode doesn't follow that model. It might carry some ads, but not every sentence is a sponsored link. The path forward requires figuring out how to monetize every pixel of an AI-generated response—a harder problem than optimizing a ten blue links page.

Capex discipline is a mixed signal. Google guided 2026 capex to $195–205 billion, staying in the $200 billion range it's already spending. Q3 capex alone was $51.4 billion, suggesting the company is tracking its full-year guidance. Google signaled it expects "significant increases" in 2027, but actual spending fell short of estimates ($51.36 billion actual vs. $55.7 billion estimated). The miss wasn't reluctance—it was delays. Supply constraints on power, permitting, and GPU racking pushed capital deployment into later quarters. Microsoft experienced the same timing gap, with capex coming in at $38.37 billion versus a $45.9 billion estimate.

Frontier model velocity is real. Gemini crossed 1 billion monthly active users. The API is processing 32 billion tokens per minute, up 40% quarter-over-quarter. Google announced Gemini 2 Carbon days before earnings—a signal to investors that it remains on the frontier and can serve its own infrastructure instead of licensing externally. Internal frontier intelligence at costs 90% below market prices has strategic value for YouTube, AI search, and internal workloads.

The structural debate centers on whether search is genuinely saturated. Max Anderson, a prolific Google Ads spender, tweeted in July that search revenue growth is "artificial and extremely unhealthy" because organic discovery has collapsed. That claim has been contested, but the pattern holds: certain queries yield almost no organic results. Satya Nadella referenced the critique yesterday. Stickiness keeps users flowing through the platform despite degraded experience, which explains why the slowdown is a 5-point deceleration rather than a cliff.

Investor anxiety is also tied to a broader philosophical shift in AI training. Internal reasoning improvements at Google and other labs are increasingly attributed to reinforcement learning in games and simulated environments rather than scale alone. Ashley Vance argues this mirrors Demis Hassabis's original AGI thesis—that games are the path to frontier reasoning. Google has invested heavily in Genie, its world model, while Anthropic focused on coding and OpenAI pursued side quests like video generation and computer use. Whether Genie becomes a valuable general-purpose training asset or an expensive diversion remains open. That uncertainty filters into valuation, especially when combined with search deceleration and the capex-heavy path ahead.