Google looks to plow approx $180B into datacenters this year

"Google's parent Alphabet is doubling down on generative AI in 2026. On Wednesday's earnings call, the search and advertising giant boosted its full-year capital expenditures target to between $175 and $185 billion, roughly twice what it spent last year. The search and ads giant will use that massive pile of cash to build datacenters and buy the kit to fill them, to support its own products and the infrastructure needs of partners like Apple, OpenAI, and Anthropic."

"Speaking during Alphabet's earnings call, CFO Anat Ashkenazi said roughly 60 percent of the company's 2026 capex spend, or about $105 to $111 billion, will go toward fast-depreciating assets like servers. The remaining 40 percent, or $70 to $74 billion, will support the construction and networking of new datacenter facilities. Much of the server spend will go toward the deployment of AI infrastructure, which includes both Google's own in-house tensor processing units (TPUs) and Nvidia GPUs."

"Just like Meta, Alphabet isn't just shoehorning generative AI into every user-facing product it can. The company is also using the models to enhance its online ad businesses. According to Philipp Schindler, Google's chief business officer, Gemini has helped improve the relevance of the ads it delivers alongside results from Google searches. "Gemini's understanding of intent has increased our ability to deliver ads on longer, more complex searches that were previously challenging to monetize," he said."

Alphabet raised its 2026 capital expenditure target to $175–$185 billion to support rapid AI expansion and infrastructure needs. Roughly 60 percent of capex, about $105–$111 billion, will go to fast-depreciating assets such as servers, while the remaining 40 percent, about $70–$74 billion, will fund datacenter construction and networking. Server purchases will deploy AI infrastructure including in-house TPUs and Nvidia GPUs, with compute investment split evenly between internal workloads and Google Cloud. Scaling compute capacity amid power, land, and supply-chain constraints is a persistent operational challenge. Generative models like Gemini are also being used to improve ad relevance for complex searches.

Read at Theregister