Google’s AI spend push

Google is treating artificial intelligence as a spending problem on the scale of a utility bill, not a software feature list. Alphabet told investors on February 4 that 2026 capital spending could reach $175 billion to $185 billion. (abc.xyz) That money is aimed mostly at technical infrastructure: servers, data centers, and networking. Alphabet Chief Financial Officer Anat Ashkenazi said on March 3 that “the vast majority” of the 2026 budget is for that infrastructure rather than side bets. (abc.xyz) Google Cloud added a second piece on April 22 at Cloud Next in Las Vegas: a $750 million fund for partners building and deploying what Google calls agentic AI. The company said the fund covers consulting firms, systems integrators, software vendors, and channel partners across its 120,000-member ecosystem. (googlecloudpresscorner.com) Agentic AI is Google’s term for software that does more than answer prompts. In its partner blog, Google described funding for companies to build AI agents into products, get early access to Google models, and work with Google engineers on customer deployments. (cloud.google.com) The spending tells you where the bottleneck is. Running large AI systems requires chips, electricity, networking gear, and buildings full of machines before a customer ever sees a chatbot, search box, or automated workflow. (abc.xyz) It also shows Google is trying to win through enterprise plumbing as much as through consumer products. At Cloud Next, the company paired the partner fund with a Gemini Enterprise Agent Platform and new Tensor Processing Unit hardware, tying software, chips, and resellers into one sales push. (crn.com) Google is not starting from a small base. Alphabet had already told investors in February 2025 that capital expenditures for 2025 would be about $75 billion, with the largest component going to servers and the rest largely to data centers and networking. (abc.xyz) By July 2025, that 2025 spending target had risen to $85 billion as Google said cloud demand and artificial intelligence workloads were outstripping available capacity. CNBC reported the increase came as the company worked through a backlog of cloud customers seeking infrastructure. (cnbc.com) The partner fund addresses a different constraint: distribution. Google said partners influence how joint customers adopt agentic AI, while outside reporting on the announcement noted firms such as Accenture, Deloitte, KPMG, and PwC are central to deciding which cloud platform big companies standardize on. (googlecloudpresscorner.com) (thenextweb.com) So the story is not just that Google wants more AI products in market. It is spending at two levels at once — hundreds of billions for the machines underneath, and hundreds of millions for the firms that will persuade customers to use them. (abc.xyz) (googlecloudpresscorner.com)