Analyst(s): Nick Patience
Publication Date: June 4, 2026
What is Covered in This Article:
- Snowflake’s rebranding of Cortex Code to CoCo and Snowflake Intelligence to CoWork, and what the distinction means for enterprise AI adoption
- The deepened Anthropic partnership and Snowflake’s role as a Claude Marketplace launch partner
- How Snowflake is positioning governed, trusted AI as its core differentiator against hyperscalers and model providers
- The strategic rationale for Snowflake building its own agentic interfaces rather than ceding that layer to third parties
- What the Summit announcements mean for enterprise AI architecture decisions in the second half of 2026
The Event—Major Themes & Vendor Moves: Snowflake Summit 2026 took place in San Francisco in early June, drawing 20,000 customers, partners, and mere analysts to the company’s annual flagship event. The event was bolstered by a powerful financial performance, with Q4 FY2026 product revenue hitting $1.23 billion, a 30% year-over-year increase. This fiscal momentum, highlighted by a single-day 37% stock surge on May 27 and a 60% climb from April lows, infused the summit with a level of energy that surpassed the previous year’s gathering.
The dominant theme across keynotes, sessions, and meetings was the agentic enterprise: how organizations move from querying data to acting on it, and how Snowflake positions itself as the platform that makes that transition safe, governed, and scalable. Three product threads ran through most of the Summit’s major announcements: CoCo, CoWork, and a significantly expanded relationship with Anthropic, alongside supporting announcements around data streaming, semantic context, security, and interoperability that we have covered in a separate report.
At Snowflake Summit, the ‘Snowmentum’ Was Palpable
Analyst Take: Snowflake Summit 2026 marked a meaningful shift in how the company describes itself and its ambitions. The formal rebranding of Cortex Code to CoCo and Snowflake Intelligence to CoWork is more than a naming exercise; it reflects a deliberate effort to establish two distinct product identities serving two distinct audiences within the enterprise, under a single strategic frame.
CoCo is the developer- and data-engineer-facing product: a Snowflake-native AI coding agent that reached more than 7,100 customer accounts in roughly six months since launch, making it the company’s fastest-growing product. Newly available as a standalone desktop app, with a VS Code extension and an Excel plugin, CoCo signals that Snowflake is serious about meeting technical users where they already work rather than requiring them to live inside the Snowsight web interface.
CoWork addresses a fundamentally different user: the knowledge worker who may never have interacted with Snowflake directly. Positioned as a personal agent that connects enterprise data, external business systems via MCP connectors, and individual workflow context, CoWork’s new capabilities include AI-native Artifacts – shareable, interactive dashboards queryable in natural language – and Deep Research, a multi-agent capability that reasons across structured and unstructured data. CoWork grew 2x quarter-over-quarter in the period leading into Summit, which is a meaningful early signal given the scale of the addressable audience.
The Governance Differentiator
Snowflake’s central differentiation claim at Summit was governed, trusted AI, the argument being that moving AI into production at enterprise scale requires the kind of security, compliance, and data governance that a platform like Snowflake can provide, and that most AI tools in the market are not delivering this reliably. New AI security announcements — including data exfiltration policies, an AI security package in Trust Center, and multi-party authorization for sensitive operations – were presented as evidence of that commitment in practice rather than in positioning alone.
However, governance is now a standard claim across the enterprise AI vendor landscape, and Snowflake is not the only platform making it. The more durable question is whether Snowflake’s governance capabilities are architecturally embedded – genuinely hard to replicate – or whether they represent a current lead that hyperscalers and competitors will close. CEO Sridhar Ramaswamy acknowledged the competitive reality directly, noting that any company operating in a market where model providers are approaching trillion-dollar valuations needs to be clear-eyed about its right to exist. That kind of candor – and discussions we had at the event with executives – suggests the governance message is being built on product substance rather than marketing positioning alone.
The Anthropic Partnership — More Than Model Access
The Snowflake-Anthropic relationship, announced in expanded form at Summit, was reinforced by the presence of Anthropic co-founder and president Daniela Amoedi chatting to Ramaswamy in the day 1 keynote (last year he did the same with OpenAI’s Sam Altman). Claude models, including Claude Opus 4.8, made available same-day on Snowflake Cortex AI, power both CoWork and CoCo within Snowflake’s governed environment. Snowflake is one of six launch partners in Anthropic’s Claude Marketplace, a commercial structure that allows customers to apply existing Anthropic commitments toward Snowflake AI consumption.
The significance here is not simply model availability, as most enterprise AI platforms offer multi-model choice, but the depth of co-innovation across both the developer and knowledge worker products and the commercial alignment that the Marketplace structure creates. For enterprise buyers, the combination of Anthropic’s frontier model capability and Snowflake’s data governance layer addresses a gap that has slowed production AI deployments: confident answers grounded in verified enterprise data, with auditability and compliance built in.
Why Snowflake Is Building Its Own Agents
One of the more strategically significant exchanges at Summit came in response to a direct question about why Snowflake needs its own agentic interfaces rather than allowing customers to use whatever agents they prefer. Ramaswamy’s answer was direct: any company that sits out the agentic interface layer is placing itself in “strategic peril”. Beyond the competitive logic, building CoCo and CoWork gives Snowflake direct telemetry on how customers are actually using the platform: where they succeed, where they struggle, and where adoption has not yet reached workflows that would benefit from it. That instrumentation is, as Ramaswamy put it, invaluable, and it represents a compounding advantage that pure infrastructure vendors cannot easily replicate.
Agentic Control Plane
During a press and analyst briefing, EVP of Product Christian Kleinerman made a deliberately opaque reference to a forthcoming MCP gateway, saying only “stay tuned.” That forward pointer was presumably a reference to the acquisition of Natoma, an enterprise MCP platform, on the same day as its Q1 FY2027 earnings. Natoma brings a governed gateway with around 100 MCP servers out of the box, along with delegated permissions, brokered access, and centralized authorization across AI ecosystems. It could extend Snowflake’s agentic control plane beyond data workflows into the everyday applications, e.g., CRM, file storage, communications, etc., where enterprise work actually happens.
What to Watch:
- AWS, Azure, and Google Cloud are all building comparable agentic and governance layers natively. The pace at which they close the gap on Snowflake’s current lead in governed AI – particularly for customers already deeply committed to a single cloud – will be a meaningful test of Snowflake’s durability in this positioning.
- CoWork’s 2x quarter-over-quarter growth is an early indicator, but the knowledge worker audience is a fundamentally different GTM motion than Snowflake’s traditional technical buyer. Whether that growth rate holds as the product moves beyond early adopters will be worth tracking closely.
- The integration timeline and how Natoma’s capabilities surface inside CoWork and CoCo will be worth watching closely over the next two quarters.
- As CoCo and CoWork scale, the cost of frontier model inference against large enterprise data sets becomes a real consideration. Snowflake’s approach to model tiering – using smaller models for routine tasks, reserving frontier models for complex reasoning – will be worth watching as it matures into a productized capability.
- With Databricks also making significant moves in the agentic and governance space – and holding its event two weeks after Snowflake, the degree to which customers see these as genuinely differentiated platforms versus increasingly overlapping alternatives will shape both companies’ enterprise narratives through the rest of 2026.
You can read all the relevant press releases on the Snowflake website.
Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of Futurum as a whole.
Other Insights From Futurum:
Snowflake Q4 FY 2026 Results Highlight AI-Led Consumption and Platform Expansion
Snowflake Summit ’25: Accelerating AI with Unified Data & Compute
Author Information
Nick Patience is VP and Practice Lead for AI Platforms at The Futurum Group. Nick is a thought leader on AI development, deployment, and adoption - an area he has researched for 25 years. Before Futurum, Nick was a Managing Analyst with S&P Global Market Intelligence, responsible for 451 Research’s coverage of Data, AI, Analytics, Information Security, and Risk. Nick became part of S&P Global through its 2019 acquisition of 451 Research, a pioneering analyst firm that Nick co-founded in 1999. He is a sought-after speaker and advisor, known for his expertise in the drivers of AI adoption, industry use cases, and the infrastructure behind its development and deployment. Nick also spent three years as a product marketing lead at Recommind (now part of OpenText), a machine learning-driven eDiscovery software company. Nick is based in London.
