Alphabet has set a new benchmark for AI infrastructure spending, announcing it expects capital expenditures in 2026 to range between $175 billion and $185 billion—more than double its 2025 spending. Despite beating Wall Street expectations on revenue, earnings per share, and cloud performance, the company’s shares declined in extended trading, highlighting investor caution around escalating AI-related costs. The massive capex projection surpasses earlier guidance from October, when Alphabet only hinted at a “significant increase” in spending. It now outpaces forecasts from peers: Meta expects 2026 capex between $115 billion and $135 billion, up from $72.2 billion in 2025, while Microsoft did not provide a specific 2026 forecast but noted a sequential decline in spending after reporting $37.5 billion in the latest quarter. Amazon, set to report results Thursday, is expected to spend about $146.6 billion in 2026, a rise of nearly 18% from its 2025 figure of $124.5 billion. Wall Street’s nervousness stems from broader concerns that AI investments may not yield proportional returns, especially as software stocks have lost 30% of their value over the past three months. Even Alphabet, a top performer in 2025, is now facing scrutiny over its aggressive spending plans. Alphabet’s cloud unit, which hosts its AI services, saw its backlog surge 55% quarter-over-quarter and more than double year-over-year, reaching $240 billion by the end of the fourth quarter. Cloud revenue rose nearly 48% compared to the same period last year. Finance chief Anat Ashkenazi explained that the 2026 capex will fund AI compute capacity for Google DeepMind, meet soaring cloud demand, and support strategic investments across Alphabet’s other ventures. She also emphasized that the spending will enhance user experience and improve advertiser return on investment across Google’s core services. In 2025, 60% of Alphabet’s capex went toward servers, with 40% allocated to data centers and networking equipment—highlighting the focus on physical infrastructure. The company recently acquired data center firm Intersect for $4.75 billion in cash and debt, signaling its commitment to expanding capacity. Executives highlighted key AI milestones, including Gemini’s monthly active user base growing to 750 million—up from 650 million in the prior quarter. They also reiterated the strategic partnership with Apple to upgrade Siri using Gemini AI models, reinforcing Google’s role as Apple’s preferred cloud provider. When asked about top concerns, CEO Sundar Pichai cited compute capacity, citing challenges with power, land availability, and supply chains. Google’s AI infrastructure head, Amin Vahdat, told employees in November that the company must double its serving capacity every six months to keep pace with demand, underscoring the intensity of the infrastructure race. As Vahdat put it, “The competition in AI infrastructure is the most critical and also the most expensive part of the AI race.”