Google Stock Fell Four Months on a Cost Problem, Not a Search Problem

Four months ago Google's stock was worth about $408. Since then it has shed roughly a sixth of its value, grinding through its longest monthly losing streak in more than a decade, and even as sellers let up in early September it still traded near $338 — about 17% below that May high. Search the headlines for why and one word comes back: AI. The story most investors carry is that generative assistants are quietly eating the business Google was built on — that asking a chatbot is replacing the habit of typing a query, and that someday the advertising attached to that query disappears with it.

That is the fear everyone assumes the market is pricing. It is not the fear doing the work.

The signal that's meant to reassure

Start with the business being threatened, because the evidence runs against the story. In the quarter ended in June, Google's search advertising revenue grew 17% to $63.3 billion. Total revenue rose 24% to $119.8 billion, ahead of the roughly $117 billion analysts had penciled in. The disruptive tool the market worried about has not shown up in the financials as a substitute. It has shown up as an accelerant: AI-powered summaries and an "AI Mode" route more queries into the ad engine rather than away from it, and AI-driven ad campaigns — adopted, by the company's count, by about half a million advertisers — lift conversion rates.

This is the reassurance. Bank of America, which rates Alphabet a Buy with a $430 price target, has been arguing that investors are missing exactly this point: the core search machine is proving resilient at precisely the moment the "AI kills Google" narrative said it shouldn't. A $430 target sits about 27% above the stock's current price, built on applying a roughly 28x multiple to the bank's estimate of core Google earnings for 2027.

If the search engine is growing fine, the obvious question is why a stock with 24% revenue growth kept falling for four straight months.

What actually brought it down

Because the market was not selling search demand. It was selling the bill for the buildout — a cost problem wearing the same scare label.

In July, Alphabet raised its planned 2026 capital spending to roughly $195 billion to $205 billion, up from an earlier range around $180 billion to $190 billion, and told investors that spending would likely climb again in 2027. Adjusting to that outlay, free cash flow fell about a fifth from a year earlier and turned negative for the June quarter — the first time it had printed red in recent memory. The stock dropped about 5% the day it heard that.

So there are two stories wrapped in one stock, and they pull in opposite directions.

One is a demand story: does AI poison Google's search economics? The reported evidence so far says no — the core got bigger, not smaller.

The other is a cost story: does roughly $200 billion a year of capital turn into future profit, or does it just pile up as data centers consuming cash faster than they earn it? That second question is the one actually moving the price, and unlike the first, it has no reassuring answer yet. BofA's signal addresses the fear that never killed the stock, which is worth knowing before you mistake one reassurance for the other.

The bet hiding inside the cost

In effect, Google is trying to expand the toll road while it is still collecting the tolls. The counterweight to the nervousness is Google Cloud, the segment where all that capital is supposed to land. In the most recent quarter Cloud revenue grew 82% from a year earlier, and it carries a backlog of committed business that Bank of America has put above $460 billion. That is committed demand — orders, not adjectives. The bull case for the whole stock rests on that conversion: that today's red ink becomes tomorrow's high-margin recurring cloud revenue.

This is where a beginner's mental model has to be honest about what is and isn't proven. Real spending is booked. The business is real and growing. But "search still works" and "cloud is growing fast" are two different claims, and only the first is settled. The second is the live wager: whether the biggest capital program in the company's history pays for itself, and how long investors have to wait for the cash to come back out. An implicit price-to-earnings near 33x on forward estimates already assumes a long runway of success; it does not reward a failed or merely slow conversion.

The reassuring signal, then, doesn't erase the risk. It sharpens it. The market has already stopped believing the AI-cannibalizes-search story — search grew 17% while the scare was loudest. What investors are actually standing in front of is the capex-to-profit question, and that is decided by charting whether a ~$200 billion annual build turns into operating cash flow. Berkshire Hathaway's vote of confidence, and the DOJ's antitrust rulings that stopped short of forcing a breakup — a judge rejected demands to sell the company's ad exchange — reduce the regulatory fog around that bet, but they don't pay the capex bill.

One metric carries the confirmation, and one signal ends the story. Confirmation is free cash flow recovering as cloud margins scale — the moment the spending starts coming back out of the other end. The signal that turns the reassuring narrative ordinary is a stall: if search growth slows despite AI, or if cloud's growth rate visibly decelerates while the buying continues, then Google is paying first-class prices for a buildout that has become a money furnace, and the $430 target starts to look like it's measuring a different company. As long as the search core holds and the cloud backlog converts, the four-month selloff looks less like a red flag than like the market finally pricing a very large, very expensive bet. That bet is the whole question. Watch the cash, not the fear.