Keynote Analysis
summary
Databricks announces Unity Catalog open source as the new metadata source of truth (repo public now, Polaris in 90 days), with unified governance, masking, metrics and lineage that aim to federate other systems and make catalogs control access instead of databases. Tabular acquisition and a stated $2 billion push prioritize Delta interoperability over Snowflake compatibility. Demos claim a 73% Databricks SQL performance lift, show LakeFlow Connect pipelines, Mosaic for multi‑model app building, and a model strategy favoring many small LMs rather than one giant frontier model.
AI-written summary · about
chapters (7)
- 0:00Interoperability and Data
- 4:23Small Language Models
- 7:05Industry Heat and Competition
- 9:20Unity Catalog Deep Dive
- 12:24Data Pipelines and Costs
- 14:14Serverless and BI Strategy
- 17:11Future Outlook on AI Budgets
related talks
20mKeynote Review
02h 52mKeynote Day 1 - Full
14mKeynote Analysis Day 2
24mMatei Zaharia, Databricks | Databricks Data + AI Summit 2023
52mElevate your data and AI governance with Databricks Data Intelligence Platform
05mDatabricks Product Announcements
14mTony Baer, dbinsight & Doug Henschen, Constellation Research | Databricks Data + AI Summit 2023
02h 16mKeynote Day 2 - Full
