Google’s enterprise AI push

Google used its Cloud Next conference in Las Vegas on April 22 to make AI agents the center of its enterprise pitch. (usnews.com) Alphabet executives told customers and investors that agents — software that can carry out multi-step tasks with less human input — are now a main way Google plans to make money from artificial intelligence in business software. Reuters reported the push alongside new infrastructure and software announcements at the three-day event. (usnews.com, cloud.google.com) Google paired that message with two new eighth-generation Tensor Processing Units, or TPUs, its in-house artificial intelligence chips: TPU 8t for training models and TPU 8i for running them after they are built. Google said TPU 8t is designed for large training jobs on a single shared memory pool, while TPU 8i is built for inference, the step where a model answers prompts and performs tasks. (blog.google) The software layer is the Gemini Enterprise Agent Platform, which Google described as a system to build, scale, govern and optimize agents. The company said the platform folds together Vertex AI model and agent tools with new features for integration, orchestration, DevOps and security. (cloud.google.com) Google is trying to turn a crowded market for large language models into recurring enterprise spending on tools that sit inside daily work. Chief Executive Sundar Pichai said customers now process more than 16 billion tokens per minute through Google’s direct application programming interfaces, up from 10 billion last quarter. (blog.google) That sales pitch depends on getting Gemini into the systems employees already use. Google said the Gemini Enterprise app can connect to enterprise data, including third-party systems, and Slack’s marketplace lists a Google integration that lets users pull Slack messages, channels, files and threads into Gemini without leaving the app. (cloud.google.com, slack.com) Google also used customer deals to show that the strategy is moving beyond demos. Merck and Google Cloud announced a multi-year partnership on April 22 valued at up to $1 billion to deploy Gemini Enterprise across research and development, manufacturing, commercial and corporate work. (merck.com, googlecloudpresscorner.com) The backdrop is a sharper contest with Microsoft, Amazon, OpenAI and Anthropic for corporate artificial intelligence budgets. Google’s own Cloud Next materials framed this year’s event around the “agentic enterprise,” and partner announcements tied Gemini more tightly to Salesforce, SAP, ServiceNow and Palantir. (cloud.google.com, cloud.google.com) Google has been building toward this for more than a year. Agentspace, a 2025 enterprise product for creating and orchestrating agents, was folded into Gemini Enterprise in October 2025, and Google now presents the new platform as the front door for workplace AI. (blog.google, cloud.google.com) The immediate test is whether customers buy more than models and chatbots. Google left Las Vegas arguing that the money will be in agents that can be built on Google chips, managed in Google software and embedded in the tools workers already open all day. (usnews.com, blog.google)