Snowflake Reports 37% Product-Revenue Growth as CoCo Expands Enterprise AI Use

Snowflake said product revenue rose 37% year over year to $1.49 billion in its second quarter of fiscal 2027, as adoption of its AI products, including the CoCo coding agent, accelerated. The quarter ended July 31, 2026, and marked the company’s third consecutive quarter of product-revenue growth acceleration, according to Snowflake’s earnings release.\n\nSnowflake positions CoCo as a data-native coding agent for data engineering, analytics, machine learning and AI application development. The product can work with customers’ Snowflake environments, permissions and metadata, and is available through interfaces including Snowsight, command-line tools and desktop applications. Snowflake has also expanded CoCo across developer workflows, including VS Code, Excel and Claude Code integrations, according to the company’s product announcement.\n\nA central part of the offering is model choice. Customers can select models from Anthropic and OpenAI, among others, balancing quality, latency and cost. Snowflake’s documentation lists models including Claude and OpenAI GPT offerings, while its broader Cortex platform provides access to models from multiple providers. The approach reflects a growing enterprise priority: using different models for different workloads instead of routing every request to the most expensive frontier model.\n\nSnowflake reported total quarterly revenue of $1.55 billion, up 35% year over year, a 126% net revenue retention rate and $9.00 billion in remaining performance obligations, up 30%. It added 692 net new customers during the quarter, including 14 Forbes Global 2000 companies. For the following quarter, Snowflake guided to product revenue of $1.588 billion to $1.593 billion, representing expected year-over-year growth of 37% to 38%.\n\nThe company’s AI strategy is designed to keep enterprise data, access controls and model usage within a governed platform. CoCo is available to commercial and certain regulated-account customers with the required cross-region inference settings, and usage is billed according to token consumption for customers with a Snowflake account. Administrators can apply budgets and usage limits, which gives organizations more visibility into AI spending but also makes cost management an operational requirement.\n\nDevelopers tend to value CoCo’s awareness of Snowflake schemas, roles and data workflows, especially for tasks that require querying, transformation and deployment inside Snowflake. The main limitation is scope: its strongest advantage comes from Snowflake-specific context, so teams working primarily outside that environment may find general-purpose coding agents more suitable. Users also need to weigh token-based costs, model availability by region and the trade-off between premium models and cheaper alternatives.\n\nSnowflake’s results suggest that AI is becoming a material consumption driver for its data platform, but the earnings figures do not isolate CoCo’s contribution from other AI products. The next test will be whether model choice, governance and workflow integration can convert early agent adoption into durable workloads and revenue without allowing inference costs to erode the platform’s economics. Snowflake’s CoCo documentation describes the product’s agent framework, model access and account-level controls.