Alphabet’s stock fell more than 12% on Thursday, putting it on track for its worst single-day decline since the company went public in 2004. The instinct is to blame Google Search. That instinct is wrong.

Search & Other revenue came in at about $63.4 billion against roughly $65 billion in consensus expectations, and YouTube Ads landed at about $11.2 billion versus the $11.4 billion the street was modeling. Those misses are real. But they do not explain a roughly 30% gap between what Alphabet reported in EPS and what analysts expected.

The actual explanation sits two lines below operating income in the filing. Other income (expense), net swung to a loss of about $8.221 billion in the quarter, and the filing attributes that drop primarily to net unrealized losses in marketable equity securities largely from SpaceX. That is not an operating problem. It is a mark-to-market swing on a private asset, and it dragged reported EPS to $2.12 against a consensus around $3.01.

Strip that out and the underlying business looks very different. Operating income increased 30% and operating margin reached 31.8%. Google Cloud delivered revenues of $30.7 billion, a 103% increase year over year, led by AI infrastructure and enterprise AI solutions. That is not a company in trouble. That is a company whose investment portfolio happened to detonate in the same quarter its headline numbers were already under a microscope.

The genuinely worrying number is free cash flow. Alphabet reported negative free cash flow in the quarter, its first such downturn since becoming a public company, as capital expenditure for AI infrastructure surged. Management has been transparent about the spending. But negative free cash flow at a company generating $127.6 billion in quarterly revenue is not something investors shrug off, regardless of the reason.

Alphabet is not alone in burning cash to build AI infrastructure. S&P Global Ratings projects all six major hyperscalers will generate negative free operating cash flow through 2027, with one notable exception.

The selloff in Alphabet also contributed to a broader pullback in technology stocks, with the Nasdaq Composite retreating 1.4% on the day. That kind of contagion is worth watching into next week.

The core question is whether Thursday’s move was price discovery or overreaction. At least one analyst described the share-price decline as an overreaction, pointing to 25% revenue growth and strong operating performance as evidence the EPS miss obscures more than it reveals. That argument has merit. A SpaceX write-down is not recurring. Negative free cash flow during a deliberate infrastructure build-out is not the same as a structural cash problem.

The pattern of strong operating results punished by a single soft line item is not unique to Alphabet this cycle. Broadcom guided to record AI revenue and still saw its stock drop roughly 15%, showing how a market priced for perfection treats any miss.

What is recurring: Search is growing more slowly than the street wants, YouTube keeps trailing estimates, and the AI capex cycle has no obvious endpoint. Those are the real debates the quarter opened. The SpaceX loss just made them louder.