Alphabet plans to spend $175-185 billion on capital expenditure in 2026, more than double the $91.4 billion it spent in 2025. The company disclosed the figure on its Q4 2025 earnings call Wednesday.
The increase reflects the reality of the AI infrastructure arms race. CEO Sundar Pichai cited "relentless innovation" and ongoing constraints in power and chip supply. Around 60% of the spend goes to servers and networking, 40% to data centers.
The number exceeded Street estimates of $115-119 billion. For context, Microsoft plans up to $120 billion in capex for its fiscal year ending June 2026. Only 59 S&P 500 companies have market caps smaller than $180 billion.
Google Cloud accelerates
Google Cloud revenue hit $17.7 billion in Q4, up 48% year-over-year. That's the first time Google has outpaced Azure's growth rate in years. Cloud backlog doubled to $240 billion.
Gemini now has 750 million monthly users, up from 650 million in October 2025. The momentum in enterprise AI appears to be paying off, though the scale of spending required to maintain it is substantial.
Market reaction
Alphabet shares dropped 3% in extended trading before recovering to close down 0.39%. Some investors are unsettled by the scale of investment relative to near-term returns, despite strong underlying results.
Broadcom, which manufactures custom chips for Alphabet, rose 6% after hours. The company stands to benefit directly from the infrastructure buildout.
What this means in practice
The spend signals Alphabet's commitment to competing in enterprise AI infrastructure. The fine print matters here: supply constraints mean not all planned spending may convert to deployed capacity in 2026.
For enterprise buyers evaluating Google Cloud for AI workloads, the investment in TPU infrastructure and data center capacity should translate to improved availability and pricing stability. The backlog figure suggests demand isn't the constraint - supply is.
History suggests these infrastructure races are expensive in the short term. Whether the returns justify the spend becomes clear 18-24 months later when utilization data emerges.