September 2026 Summaries

At dbt Summit 2026, dbt Labs and Fivetran presented a product strategy centered on the idea that the governed data foundations used for trustworthy analytics are also essential for reliable AI. Key releases include dbt v2, now generally available as a unified Rust-based engine with improved SQL comprehension and performance, and dbt State, which selectively rebuilds changed models to reduce compute use and simplify freshness management. The companies also introduced Lake Compute in private beta, allowing individual dbt models to run against Apache Iceberg tables through a DuckDB-based engine while others remain on a warehouse. To support AI applications, dbt expanded integrations for its Semantic Layer, MCP Server, and Agents Schema, while Fivetran announced its Context Layer for converting structured and unstructured business information into maintained AI context. dbt Wizard, available across the platform, CLI, and desktop environments at varying preview stages, provides a project-aware agent for analytics engineering and conversational exploration, while dbt Charts in public beta aims to bring dashboards into version-controlled, YAML-based development workflows alongside data models.

Fivetran and dbt Labs announced several products at dbt Summit 2026 aimed at helping enterprises prepare governed data for AI agents while maintaining flexibility across data platforms. The releases include the general availability of dbt v2, a Rust-based engine that reportedly parses large projects up to 10 times faster than dbt v1, and dbt State, which uses warehouse metadata and SQL changes to avoid unnecessary builds and reduce compute costs. The companies also introduced an Open Data Infrastructure vision centered on customer-owned data stored in open formats such as Apache Iceberg, supported by the existing Managed Data Lake Service and the private-beta Lake Compute SQL engine. Fivetran Context Layer, in private beta, combines structured dbt metadata with unstructured business information through the open-source Agents Schema standard, making context available to AI tools including Anthropic integrations and ChatGPT. New dbt Wizard agent experiences for the platform, command line, desktop, and conversational analytics are designed to help users create and validate SQL using project lineage, tests, and governance context, while dbt Charts offers YAML-defined, version-controlled business intelligence alongside dbt models.

dbt State is now generally available across dbt environments, including self-managed orchestrators, Snowflake dbt Projects, and warehouses such as Snowflake, BigQuery, Databricks, and Redshift. It evaluates model code and warehouse metadata on each run to determine whether models need rebuilding, allowing unchanged nodes to be skipped, cloned, reused, or deferred without custom selectors, manifest management, or orchestration logic. The company reports that early users have achieved average compute savings of 15–30%, while some organizations cite reduced job costs, faster development cycles, and simplified deployments. dbt State shifts freshness management from job schedules to model-level configuration through controls such as lag tolerance, enabling teams to consolidate jobs and apply consistent rules across large DAGs. In development, it can automatically reuse fresh production or upstream assets, reducing the need for manual cloning, deferral setup, and costly upstream rebuilds. The product also provides explanations for build decisions, concurrency management, cost insights on the dbt platform, and consumption-based pricing tied to distinct target tables that are reused, skipped, cloned, or deferred each day.

dbt Labs announced five 2026 Partner of the Year winners at Partner Day during dbt Summit, recognizing organizations for customer delivery, regional expertise, technology collaboration, and emerging growth. phData was named overall Partner of the Year for a fourth consecutive year, cited for trusted dbt implementations, strong certifications, and support for AI and agentic-data initiatives. Snowflake received Technology Partner of the Year for helping joint customers deploy production-grade pipelines more quickly, while Cívica and Datum Studio were recognized as EMEA and APJ Partners of the Year, respectively, for certified delivery capabilities and extensive regional customer work. 66degrees earned Emerging Partner of the Year after rapidly beginning customer engagements, contributing Google Cloud expertise, and participating in joint events. dbt Labs said the awards reflect multiple forms of partner value, including technical depth, quality of delivery, regional presence, and investment in new customer solutions.