Alphabet reported Q4 revenue of $113.8B, up 18% year-over-year and above the $111.4B consensus. Cloud revenue hit $17.7B, up 48%, versus $16.2B expected. Net income rose 30% to $34.5B.
The number that matters: Google signaled capital expenditure could reach $185B in 2026. That's nearly double analyst estimates and represents a massive bet on AI infrastructure. For context, the entire company spent roughly $100B on capex in 2025.
What's driving the spend: GPU procurement, data center expansion, and the infrastructure needed to support both Google's own AI products and cloud customers' workloads. The company cited supply constraints as a limiting factor, suggesting they'd spend more if they could get the hardware.
The Apple situation got awkward. Multiple analysts pressed executives about the rumored Gemini partnership with Apple's Siri redesign. Management declined to provide details, even basic ones like whether the partnership exists or where inference runs. Bloomberg's Mark Gurman reports Google execs suggested processing would happen on Google servers, contradicting Apple's usual approach.
YouTube subscriptions (Premium and TV) are now generating roughly $20B annually. Search revenue hit $63B for the quarter. Google also detailed testing of ads in AI Overview mode, though didn't break out revenue attribution.
The market didn't love the capex guidance. Alphabet shares dropped in after-hours trading despite beating on revenue and earnings. Investors are weighing strong current performance against unprecedented spending commitments.
History suggests caution here. When hyperscalers announce dramatic capex increases, the question isn't whether AI is real (it is), but whether returns will justify the scale of investment. Google's betting they will. We'll know in 18 months whether data center capacity translates to revenue growth that supports the spend.
Notably absent: specific GenAI revenue attribution, concrete AI product monetization timelines, or clarity on competitive positioning versus Microsoft/OpenAI in enterprise.
The trade-off is clear: Google's buying every GPU it can find, betting that owning the infrastructure wins the AI era. The risk: building data centers faster than they can fill them with paying customers.