
Talks at GS AI and the Enterprise Revolution: Databricks CEO Ali Ghodsi
Nov 18, 2025
Ali Ghodsi, Co-founder and CEO of Databricks, shares his journey from frustration with enterprise data tools to creating a company valued over $100 billion. He discusses the importance of strong partnerships in the multi-cloud and AI landscape. Ghodsi emphasizes the need for realistic AI expectations and highlights early successes in generative AI like chatbots. He warns against underestimating required talent and stresses the importance of quality and explainability in AI systems, while comparing the evolution of AI to the growth of the internet.
AI Snips
Chapters
Transcript
Episode notes
Founding Databricks From Frustration
- Ali Ghodsi described starting Databricks in 2013 out of frustration that enterprises couldn't leverage data for AI.
- He and his team decided to build the solution themselves after open-sourcing and evangelizing failed to move incumbents.
Prioritize Pragmatic Product Investment
- Invest capital into products that address real, repeatable enterprise needs rather than hype.
- Focus hiring and R&D on features customers use in production, like Agent Bricks for reliable task automation.
Benchmarks Should Mirror Real Tasks
- Agent Bricks targets mundane enterprise tasks and benchmarks agents on those specific workflows.
- Optimizing for predictable, boring tasks improves production-quality agent automation faster than chasing frontier benchmarks.

