Alphabet reported a solid Q4 earnings beat on February 4, then watched its shares fall anyway. Revenue hit $113.83 billion versus an expected $111.43 billion, and EPS came in at $2.82 against a $2.59 consensus. Google Cloud particularly outperformed at $17.66 billion revenue versus a $16.18 billion estimate.
The problem: The company announced capex spending of $175-185 billion for 2026, more than double the 2025 figure. Most of that spending targets AI compute capacity for Google DeepMind. Operating margin compressed 50 basis points as infrastructure costs rose.
This is the classic "sell the news" pattern that hits tech companies after earnings beats. Good results get priced in during the run-up. The stock closed up nearly 2% the day before earnings. What investors wanted to hear about in the call was margin expansion. What they got was a massive spending commitment with unclear ROI timelines.
Barclays noted that "Infrastructure, DeepMind, and Waymo costs weighed on overall Alphabet profitability" and will continue pressuring margins through 2026. Deutsche Bank called the capex plan stunning but questioned whether that's positive given current market uncertainty around tech spending.
The broader context matters here: Enterprise software stocks sold off hard this week on fears that AI tools could replace traditional software products. Alphabet's aggressive AI spending signals they're taking that threat seriously, but it also validates investor concerns about margin pressure across the sector.
Google Cloud's growth and Search acceleration remain strong. The company is shipping AI integrations like Gemini across products. But for a $4 trillion market cap company, sustaining the historical 25% CAGR that got them here looks increasingly difficult when you're doubling infrastructure spending.
New York State Common Retirement Fund cut its Alphabet position by 3.7% in Q3 2025. That move looks prescient now. The question for enterprise tech leaders watching this: If Google needs to spend $180 billion to stay competitive in AI, what does that signal about the economics of the AI race for everyone else?