Snowflake AI Growth: CoCo, CoWork, Cloud Data Platform, and the $6.07 Billion Forecast

Snowflake has raised its fiscal 2027 product revenue forecast to $6.07 billion from $5.84 billion after a quarter in which product revenue reached $1.49 billion, up 37% year over year. The financial update reflects strong demand for Snowflake's cloud data platform as enterprises continue migrating legacy workloads and adding AI-driven applications.

The AI component is increasingly visible inside the product itself. Reuters highlighted Snowflake's coding assistant Cortex Code and its enterprise chatbot CoWork, while Snowflake's current documentation presents the coding-agent product line as Snowflake CoCo across Snowsight, desktop, and command-line experiences.

Snowflake is therefore moving beyond data storage and analytics toward a platform where developers, data teams, and business users can use agents directly against governed enterprise information, with permissions and security controls inherited from the Snowflake environment.

··········

SNOWFLAKE IS TURNING ITS DATA PLATFORM INTO AN AI OPERATING LAYER.

The current product direction places AI agents close to the enterprise data, permissions, and compute they need to perform useful work.

Snowflake's AI strategy is easier to understand as a stack. The underlying cloud platform stores and processes enterprise data, CoCo gives technical users an agent that can work with code and Snowflake environments, and CoWork gives business users a conversational layer for analysis and action.

........

Component	Primary role	Current use	Practical value
Snowflake CoCo	AI agent for technical work	Data engineering, analytics, machine learning, coding, and agent building	Works directly with Snowflake context, schemas, RBAC, SQL, Python, and local development tools
Snowflake CoWork	Business-facing agentic application	Natural-language analysis of structured and unstructured enterprise data	Generates insights, charts, tables, recurring automations, and actions through governed data agents
Cortex Agents	Agent orchestration layer	Routes questions to semantic data, search, and custom tools	Lets an AI agent choose tools and execute multi-step tasks while remaining inside Snowflake governance
Cloud data platform	Data and compute foundation	Storage, analytics, application development, and AI workloads	Keeps data, access policies, compute, and AI services in one governed environment

........

This architecture can reduce the amount of data movement and duplicated permission logic required when an organization connects external AI tools to internal databases. It also means the quality of an AI workflow still depends on data quality, semantic definitions, permissions, tool design, and the models selected for each task.

··········

COCO AND COWORK TARGET DIFFERENT USERS INSIDE THE SAME GOVERNED ENVIRONMENT.

CoCo is designed for builders and technical teams, while CoWork turns the same data foundation into a conversational work surface for business users.

Snowflake CoCo is an AI-driven agent integrated with the Snowflake platform and optimized for data engineering, analytics, machine learning, and agent-building tasks. Snowflake currently offers core CoCo experiences in Snowsight, a standalone desktop IDE, and a command-line interface.

In Snowsight, CoCo can generate and explain SQL or Python, use the file or notebook currently open as context, and present proposed changes for review before they are applied. The desktop and CLI versions can also work with local repositories, Git operations, Snowflake warehouses, and external tools.

Snowflake documents role-based access control, OS-level sandboxing, approval controls, and automatic risk assessment as part of the CoCo security model. Several extensions remain at different maturity levels: the CoCo Agent SDK, MCP support, ACP support, and plugins are documented as preview features, while the core Snowsight, desktop, and CLI experiences are generally available.

CoWork approaches the same platform from the business side. A user can ask a question in natural language, and Cortex Agents can route the request to structured-data analysis, search over unstructured information, or custom tools that execute functions and procedures.

The application can return explanations, tables, charts, reusable artifacts, recurring automations, and generated documents. Snowflake also describes source traceability, row-access policies, column-level security, user quotas, and resource budgets, which are important because agentic systems can consume both AI tokens and regular Snowflake compute.

··········

THE EARNINGS UPDATE SHOWS COMMERCIAL MOMENTUM, WITHOUT SEPARATING AI REVENUE.

The new guidance is significant, but the reported numbers combine Snowflake's traditional cloud data business with its newer AI workloads and applications.

........

Metric	Latest figure	Comparison	Interpretation
Fiscal 2027 product revenue forecast	$6.07 billion	Previously $5.84 billion	Snowflake increased its annual product-revenue outlook
Q2 product revenue	$1.49 billion	+37% year over year	Shows strong consumption across the platform
Q2 total revenue	$1.55 billion	$1.48 billion analyst average	Quarterly revenue exceeded the LSEG-compiled consensus
Adjusted earnings per share	$0.62	$0.45 analyst estimate	Adjusted profitability exceeded expectations for the quarter
AWS infrastructure agreement	$6 billion over five years	Signed in the previous quarter	Provides access to AWS Graviton processors and AI infrastructure for future Snowflake workloads
After-hours share move	More than 20% higher	Following the results	Reflects investor reaction to the stronger guidance and quarter

........

The strongest interpretation is therefore broader than a single AI product success. Snowflake is benefiting from legacy-system migrations, continuing demand for cloud data warehousing, and growing interest in building AI applications where enterprise data already resides.

Because Snowflake operates largely through consumption, the commercial effect of AI depends on real usage: queries executed, warehouses used, models called, agents run, and business processes repeated. A successful AI strategy should eventually increase useful platform consumption, but the current quarter does not provide a clean revenue split for CoCo, CoWork, or other Cortex products.

··········

SNOWFLAKE'S AI ADVANTAGE DEPENDS ON TURNING GOVERNED DATA INTO REPEATABLE WORK.

The practical case is strongest for organizations that already keep important data and governance controls in Snowflake and want agents to work inside that environment.

For those organizations, CoCo can shorten technical workflows around SQL, Python, pipelines, analytics, and agent development, while CoWork can expose governed data to business users through natural-language analysis, reports, automations, and actions.

The trade-off is that convenience does not remove the underlying requirements for accurate data models, carefully scoped permissions, cost controls, review procedures, and reliable tools. Agentic workflows can execute real operations, so governance has to cover both what the model can see and what it is allowed to do.

Snowflake's current position is therefore credible because the AI layer is being built on top of a large existing data and compute platform. The stronger $6.07 billion forecast supports the case that demand is expanding, while it should not be treated as independent proof that CoCo or CoWork are technically superior to competing enterprise AI systems.

For buyers, the decision is relatively concrete: Snowflake's AI stack becomes more attractive when the organization's data, security model, and operational workflows already depend on Snowflake; companies whose data and compute remain elsewhere should compare the integration effort, token and compute costs, governance model, and agent capabilities against alternatives before consolidating additional workloads on the platform.

··········

FOLLOW US FOR MORE.

·····

DATA STUDIOS

·····

[datastudios.org]