Google pledges $185bn, merges Vertex AI and Workspace into a unified enterprise AI stack at Cloud Next
- Google used Cloud Next ’26 to launch a new Gemini Enterprise Agent Platform, folding Vertex AI’s model-building tools into a single system for building, securing and managing enterprise AI agents. - Alphabet separately told investors its 2026 capital spending will reach $175 billion to $185 billion, underscoring how much infrastructure Google is putting behind new TPUs, networking and AI services. - The pitch ties chips, cloud software and workplace tools into one stack as companies move from AI pilots to managed deployments. (blog.google)
Google used Cloud Next ’26 to pitch a single enterprise AI stack built around its new Gemini Enterprise Agent Platform. (blog.google) Google said the platform combines Vertex AI model-building and tuning with new tools for agent integration, security, DevOps and governance. The company described it as a “one-stop-shop” for autonomous agents. (blog.google) Sundar Pichai said the enterprise conversation has shifted from building one agent to managing thousands of them. Google positioned the new platform as the control layer for that larger deployment problem. (blog.google) Underneath that software pitch, Google introduced eighth-generation Tensor Processing Units, splitting them into TPU 8t for training and TPU 8i for inference. Google said both are aimed at the “agentic era,” where AI systems handle more real-time work. (blog.google) (cloud.google.com) Google’s broader Cloud Next message was vertical integration: custom chips, networking, cloud infrastructure, models and workplace software sold as one managed system. The company’s event page framed that strategy as the path to an “Agentic Enterprise.” (cloud.google.com) (blog.google) The money behind that strategy is unusually large. Alphabet told investors in its February 4, 2026 fourth-quarter earnings call that 2026 capital expenditures are expected to be in the range of $175 billion to $185 billion. (abc.xyz) That spending guidance predates Cloud Next, but it gives the conference announcements more weight. Google is not only adding agent software; it is budgeting for the data centers, chips and networks needed to run those systems at scale. (abc.xyz) (blog.google) Google also used the event to push AI deeper into workplace software. Its Cloud Next materials highlighted Workspace Intelligence and broader Gemini Enterprise features aimed at bringing agents and automation into everyday employee tools. (cloud.google.com) (www.crn.com) The immediate test is whether customers buy the bundle instead of stitching together separate chips, models and software from multiple vendors. Google’s argument at Cloud Next was that enterprises now want fewer pilots and more governed production systems. (blog.google 1) (blog.google 2)
