Snowflake Summit 2024: The Data Stack Gets a Unified Foundation and Smarter Tools

At this year's Snowflake Summit, the data world saw a shift that’s been years in the making. Fivetran and dbt Labs are now one company, merging the standard for data ingestion with the standard for transformation. For thousands of organizations, these tools were already the backbone of their data pipelines. Now, the combined team is building toward a single goal: an open, portable data layer designed for the age of AI agents.

DocuSign’s Senior Director of Global Data & Analytics, Piyush Bhargava, put it plainly: "AI and agents are only as strong as the data behind them." That sentiment drove the week’s biggest product news.

First, dbt Core v2.0 is here, rebuilt on the Fusion engine. The original Python-based engine had limits—slow parse times and no native SQL comprehension. Fusion, built in Rust, delivers parse speeds up to 10x faster. The team open-sourced the Fusion runtime under Apache 2.0, ending the two-engine confusion and giving every practitioner a faster, unified foundation.

Next up: dbt State. Instead of rebuilding every table on every run, dbt State checks what’s actually changed—in metadata and model SQL—and skips what hasn’t. CarGurus saw a 9% compute reduction and 35% fewer models built. The result? Teams stop rationing runs and start refreshing data as often as the business needs.

Finally, there’s dbt Wizard, an AI agent built specifically for analytics engineering. Unlike generic coding agents, it knows your dbt project’s lineage, contracts, and tests before it writes a single line. Endress+Hauser reported a 15-20% drop in production incidents after rolling it out.

The message from Snowflake Summit is clear: the data stack is getting smarter, faster, and more unified—built for the work teams actually do.