Inside Snowflake Summit 2026: Agentic AI, a $6 Billion AWS Bet, and a New Identity

Snowflake CoCo (formerly Cortex Code), which was already the company’s internal nickname for the product, was formally rebranded as the builder layer for developers and data engineers . The tool functions as an AI coding agent that understands enterprise data context, and at Summit, Snowflake announced added support for data streaming, broadening CoCo’s role as a development platform for building enterprise applications directly on Snowflake .

The rebranding effort draws a clean line between the business teams consuming AI insights and the technical teams building the pipelines that power them.

Cortex Sense: From Reactive Dashboards to Proactive Intelligence

Embedded within CoWork, Cortex Sense represents a philosophical shift in how Snowflake thinks about analytics. Instead of requiring users to query dashboards, Cortex Sense is described as a proactive insight engine that delivers “the right insights and impactful actions in real time” . The move acknowledges that the agentic enterprise isn't just about building AI assistants—it’s about creating systems that anticipate needs and surface actions before a human asks for them.

Acquiring Natoma to Solve the Agent Permission Problem

Perhaps the most strategically revealing announcement at Summit 2026 was Snowflake’s intent to acquire Natoma, an enterprise Model Context Protocol (MCP) platform . If the rebrands define who uses AI at Snowflake, the Natoma acquisition defines how securely they’ll use it.

Natoma built a centralized MCP gateway that manages AI agent permissions across enterprise applications. The technology controls what agents are allowed to do—such as sending emails, summarizing Slack conversations, or opening Jira tickets—and enforces security policies at the tool-call level . Snowflake CEO Sridhar Ramaswamy framed the capability as essential for a world where agents act autonomously: “With Natoma, users can do things like send emails, summarize Slack conversations, check calendars, and open Jira tickets without ever leaving Snowflake Intelligence or Coco” .

Once integrated, Snowflake customers will be able to securely connect CoWork and CoCo to enterprise systems through a verified library of MCP servers, with centralized governance provided by the MCP Gateway . This gives Snowflake a native identity and permissions layer for multi-agent workflows—solving one of the most pressing enterprise concerns about agentic AI: how to prevent agents from going rogue while still giving them meaningful access to tools .

Financial terms of the acquisition were not disclosed .

The $6 Billion AWS Infrastructure Commitment

Days before Summit, on May 27, 2026, Snowflake signed a multi-year Strategic Collaboration Agreement (SCA) with AWS, committing $6 billion over five years to AWS compute and AI infrastructure . The spend will focus heavily on AWS Graviton processors and GPU infrastructure for AI training and inference—an allocation that underscores the shift from experimental AI workloads to persistent, production-grade operations .

This is Snowflake’s largest AWS commitment to date. Its five-year spending agreements with AWS have grown from $1.2 billion at the time of its IPO in 2020 to $2.5 billion in 2023, and now $6 billion . Snowflake has now sold over $7 billion in services through the AWS Marketplace since its founding .

The expanded deal also deepens product integrations around generative and agentic AI and broadens joint go-to-market efforts. The message is clear: running a fleet of AI agents across thousands of enterprise customers requires infrastructure certainty, and Snowflake is locking that in with its primary cloud partner.

Deepening the Anthropic Partnership

Anthropic featured prominently at the conference. Co-Founder and President Daniela Amodei delivered a keynote alongside Snowflake CEO Sridhar Ramaswamy, and the two companies announced a significant expansion of their strategic partnership .

Claude models now power Snowflake CoWork and CoCo, making Anthropic the foundational AI provider for Snowflake’s agentic platform . The partnership also introduced a Claude Marketplace and security-focused development workflows, further embedding governance into the AI development lifecycle .

Additionally, Snowflake added Gemini 3.5 Flash to Cortex AI Functions for multimodal video and audio analysis, and launched Cortex AI Function Studio in public preview for creating, evaluating, and optimizing AI functions .

The Strategic Picture: Governance as a Product

Taken together, the Summit 2026 announcements are less a collection of features and more a unified architecture for the agentic enterprise. The Natoma acquisition delivers the security and permissions layer that enterprises demand before they’ll let AI agents act on their behalf. The CoWork/CoCo rebrands segment the platform for two very different audiences—business users and developers—while Cortex Sense reimagines how those users interact with data. The AWS deal provides the compute foundation required to run agentic workloads at scale. And the Anthropic partnership ensures the models at the core of the system are enterprise-grade.

Snowflake is betting that the winning enterprise AI platform won’t be the one with the most powerful model, but the one that provides the most trusted environment for models to act. Summit 2026 was the moment that bet became unmistakably public.