Salesforce expands Google Cloud AI partnership as enterprise buyers seek stability

At Salesforce's 2026 Dreamforce conference, discussions between CEO Marc Benioff and leaders from Nvidia, OpenAI, and Anthropic highlighted a divide over the pace of AI development. [1] While Nvidia CEO Jensen Huang urged frontier labs to "run as fast as you can," some of the 50,000 attendees expressed difficulty taking advantage of existing models. [1]

Alec Bronston, a senior Salesforce director at retail data company Spins, stated that a potential slowdown in model releases would provide an opportunity to catch up. [1] Days before the event, an Anthropic researcher resigned amid this broader debate over rapid advancement. [1]

During the event, Salesforce and Google Cloud expanded a strategic partnership to connect data and AI agents across their platforms. [3] Salesforce is integrating its headless architecture with Google Cloud's Gemini Enterprise and will run its Hyperforce infrastructure natively on Google Cloud. [3]

Salesforce plans to begin migrating select United States customers in the fourth quarter of 2026, with an expansion to Germany scheduled for 2027. [3] According to Google Cloud executive Kevin Ichhpurani, this integration uses the open Model Context Protocol to help customers build AI agents that access workflows in sales and finance. [3]

The contrast between Nvidia's push for faster AI development and the enterprise struggle to deploy existing models highlights a growing disconnect between frontier labs and their enterprise customers. While infrastructure providers continue scaling, buyers are seeking stability to integrate AI into core workflows. The expanded partnership between Salesforce and Google Cloud attempts to bridge this gap by focusing on native interoperability and practical agent deployments across sales and finance. By running Hyperforce natively on Google Cloud and utilizing the Model Context Protocol, the companies are prioritizing unified data access over raw model capability. What the sources don't address: whether other major enterprise software providers will similarly adopt the Model Context Protocol to standardize AI agent interoperability.