Snowflake just quietly acquired Natoma, a startup purpose-built to govern what AI agents do with enterprise data. Combined with a $6 billion AWS deal, the move reveals that Snowflake's real ambition isn't just storing data—it's controlling the actions that AI agents take on top of that data.
Snowflake, the cloud data platform giant, announced the acquisition of Natoma in late May 2026, though financial terms were not disclosed. The deal comes alongside a separate five-year, $6 billion strategic collaboration with Amazon Web Services (AWS) that will see Snowflake deepen its integration with the AWS ecosystem. Together, these moves signal a pivot: Snowflake is building a layer that governs not only data access, but the behavior of the AI agents that increasingly query, transform, and act on that data.
Why now? The rise of generative AI and autonomous agents has created a governance vacuum. Enterprises are deploying agents that write SQL, generate reports, trigger workflows, and even interact with customers. Without guardrails, those agents can misuse data, violate compliance rules, or produce erroneous outputs. Snowflake's existing data governance features—like row-level security and dynamic masking—weren't designed for agentic behavior. Natoma fills that gap.
Natoma, a small startup founded by former Splunk and Databricks engineers, specializes in runtime governance for AI agents. Its technology monitors agent actions in real time, enforces policies, and logs every decision an agent makes against data. According to sources familiar with the deal, Snowflake will integrate Natoma's runtime governance engine directly into Snowflake’s Cortex AI and Snowpark frameworks, giving customers a single pane of glass to manage both data and agent governance.
The AWS deal, valued at $6 billion over five years, cements Snowflake's infrastructure partnership with the largest cloud provider. It also paves the way for Snowflake to offer governed AI agent services on AWS, where many enterprise workloads already sit. Industry analysts view this as a direct response to Databricks, which has been aggressive in stitching together AI, data engineering, and governance through its Unity Catalog and acquisition of MosaicML.
What does this mean for the broader market? AI agent governance is emerging as one of the most critical enterprise software categories of the next decade. As more companies deploy autonomous agents—for customer support, code generation, financial analysis—the need to audit and control those agents becomes as essential as data governance itself. Snowflake is betting that its existing customer base, already trusting the platform with exabytes of data, will see agent governance as a natural extension.
Looking ahead, Snowflake plans to preview the Natoma-powered governance capabilities at its annual Snowflake Summit in June 2026. The company will also release a set of API endpoints that allow third-party agents—from Microsoft Copilot, Salesforce Agentforce, and others—to be governed under the same policy engine. If successful, Snowflake could position itself as the neutral governance layer for an increasingly multi-agent enterprise world.
The message is clear: Snowflake is no longer just a place to store data. It's becoming the sheriff of what happens when data is unleashed.
AI
→ Neutral
Snowflake Buys Natoma To Govern The Agents Acting On Its Data
Snowflake's quiet Natoma buy, alongside a $6B AWS deal, reveals its real ambition: governing what AI agents do, not just storing the data they reach for.
Janakiram MSV, Senior Contributor
Forbes
3 min read
7/10
Key Takeaways
- Snowflake acquired AI agent governance startup Natoma in May 2026; financial terms were undisclosed.
- The acquisition accompanies a separate five-year, $6 billion strategic collaboration with Amazon Web Services (AWS).
- Natoma's technology provides real-time runtime governance, monitoring agent actions against enterprise data. Founders previously worked at Splunk and Databricks.
- Snowflake will integrate Natoma's engine into Cortex AI and Snowpark, offering a unified governance plane for data and agents.
- The move directly competes with Databricks' Unity Catalog and MosaicML acquisition, as both platforms vie for the enterprise AI governance market.
Frequently Asked Questions
AI agent governance refers to the policies, tools, and runtime controls that monitor, audit, and enforce rules on the actions taken by autonomous AI agents—such as querying databases, generating content, or triggering workflows. It ensures agents behave within compliance, security, and operational boundaries.
Snowflake acquired Natoma to gain real-time governance capabilities for AI agents that act on enterprise data. As agents become more common, Snowflake needs to offer customers control over what those agents do, not just who accesses the data. Natoma's runtime enforcement fills that gap.
The five-year, $6 billion strategic collaboration with AWS deepens Snowflake's integration with the cloud giant. It provides Snowflake with infrastructure scale and allows it to offer governed AI agent services directly on AWS, where many customers already run workloads.
Natoma's technology monitors AI agents in real time as they interact with data. It applies policy-based controls, logs every action, and can block or flag violations. Snowflake will embed this runtime governance engine into its Cortex AI and Snowpark environments.
Databricks is the primary competitor with its Unity Catalog and MosaicML acquisition. Other players include Microsoft Purview for AI governance and startups like Guardrails AI. Snowflake's deep data platform integration gives it an advantage for governance within its ecosystem.
Snowflake plans to preview the AI agent governance capabilities at its Snowflake Summit in June 2026. A broader rollout with API endpoints for third-party agents is expected later in the year.
Topics
Original source
www.forbes.com
Discussion
Join the discussion
Sign in to post a comment or reply.
No comments yet. Be the first to share your thoughts!