Google deepens enterprise ties

Google is pushing deeper into corporate AI by tying its models, software and chips more tightly to big enterprise customers and partners. (blog.google) At Google Cloud Next in Las Vegas on April 22, Google said Gemini Enterprise and a new Gemini Enterprise Agent Platform would be central to what it calls the “agentic enterprise,” with new infrastructure including eighth-generation Tensor Processing Units and broader agent tools. (blog.google; cloud.google.com) Salesforce and Google Cloud said April 22 that they are expanding their partnership so artificial intelligence agents can run end-to-end workflows across both companies’ systems, using shared context from Salesforce data and Google Cloud services. (salesforce.com) SAP and Google Cloud said the same day that SAP’s Joule agents in customer-experience software will connect with Gemini Enterprise, which SAP described as a central hub for data integration and multi-agent coordination across SAP and Google Cloud platforms. (news.sap.com) Merck said April 22 that it will invest up to $1 billion over multiple years with Google Cloud to build what the companies called an agentic platform across research, manufacturing and corporate functions, with Gemini Enterprise available to Merck’s 75,000 employees. (merck.com; googlecloudpresscorner.com) Those deals land as companies move from chatbot pilots to systems that can search records, draft work, trigger software actions and hand tasks between specialized tools. Google is selling the cloud, the model and the orchestration layer together. (cloud.google.com; blog.google) The infrastructure side is getting bigger, too. TechCrunch reported on April 22 that Mira Murati’s Thinking Machines Lab signed a multibillion-dollar agreement with Google Cloud for Nvidia GB300-powered capacity, giving the startup access to some of Nvidia’s latest systems through Google’s cloud. (techcrunch.com) Google’s cloud pitch now spans both ends of the market: enterprise software companies that want agents to work inside existing business systems, and frontier-model startups that need huge clusters of graphics processors to train and run new models. (salesforce.com; news.sap.com; techcrunch.com) That makes cloud choice more consequential for corporate buyers. Once data connections, workplace assistants, software agents and compute contracts are all tied to one provider, switching clouds gets harder and more expensive. (news.sap.com; salesforce.com; merck.com) Google is not alone in that race, but this week’s announcements show how it wants enterprise AI bought: as a bundled stack of agents, data plumbing and scarce computing power, all rented from the same cloud. (blog.google; cloud.google.com; techcrunch.com)