Ali Ghodsi

Who they are

Ali Ghodsi is Co-Founder and CEO of Databricks — did his PhD in distributed computing at KTH Royal Institute of Technology and co-founded Peerialism AB, a Stockholm peer-to-peer data transfer company, before building Databricks out of UC Berkeley's AMPLab in 2013.

Person

Ali joined Databricks in 2013 at founding — one of seven UC Berkeley academics commercializing Apache Spark, at a moment when 'data lakehouse' wasn't yet a term anyone used. His path there ran through Sweden: MSc in Computer Engineering at Mid-Sweden University (2002), an MBA in Logistics and Marketing from the same institution (2003), then a PhD in distributed computing at KTH Royal Institute of Technology (2006). He co-founded Peerialism AB, a Stockholm-based peer-to-peer data transfer company, and Hive Streaming, before landing at UC Berkeley's AMPLab as a researcher — the lab that birthed Spark. At Databricks he first ran Engineering and Product Management before becoming CEO. He holds an adjunct professorship at UC Berkeley. He writes on the Databricks blog, contributes to TechCrunch and Business Insider, and turns up everywhere from the a16z Boss Talk podcast with Ben Horowitz to Goldman Sachs Talks at GS and Ben Thompson's Stratechery — consistently on the same themes: AI in the enterprise, the data lakehouse architecture, and open-source AI. The through-line is research-to-commercialization: he built the academic foundation, then spent a decade turning it into a business at scale.

Company

In February 2026, Databricks completed a $5 billion funding round plus $2 billion in new debt capacity at a $134 billion valuation, with investors including Goldman Sachs, Morgan Stanley, and Qatar Investment Authority — the freshest signal of where the company sits. The December 2025 Series L round had already pushed it past a $5.4 billion annualized revenue run rate, with $1.4 billion of that coming from AI products specifically. In March 2026, Databricks launched Lakewatch, an AI-driven open SIEM product for cybersecurity powered by Anthropic's Claude models, with Adobe and Dropbox among its early customers — supported by acquisitions of Antimatter and SiftD. Earlier, in 2025, Databricks launched Agent Bricks and Lakebase (an operational database optimized for AI agents), and expanded partnerships with Anthropic, SAP, Microsoft, Google Cloud, and Palantir. An IPO is being discussed internally, with a valuation target cited between $165 billion and $175 billion and a potential window in 2027.

Market

Databricks competes in the enterprise data and AI platform market against Snowflake (its most direct rival in cloud data platforms), Microsoft Fabric, Cloudera, and the cloud hyperscalers — AWS, Azure, and Google Cloud — all of whom are building or bundling competing capabilities. The company serves over 10,000 organizations including many Fortune 500 firms, and positions its unified data intelligence platform as the neutral, open-source-rooted alternative to hyperscaler lock-in. Two headwinds are worth watching: geopolitical and data-sovereignty concerns are pushing some organizations — especially in Europe — toward regional alternatives, and Databricks' investor base includes sovereign wealth funds subject to US regulatory review.

Network

Matei Zaharia, CTO and co-founder of Databricks, is Ali's closest named collaborator — he promoted the Fortt Knox CNBC interview and is a frequent public partner. Ben Horowitz of Andreessen Horowitz has appeared alongside Ali on the a16z Boss Talk podcast on enterprise AI. Ben Thompson of Stratechery conducted an extended interview with Ali in 2024 on building enterprise AI.

Matei Zaharia· CTO and Co-Founder, Databricks
Ben Horowitz· General Partner, Andreessen Horowitz
Ben Thompson· Founder and Author, Stratechery

How they likely show up

Co-founder and CEO since 2013 (over a decade at the same company) → thinks in multi-year bets, not quarterly cycles; has seen Databricks through every stage from pre-revenue to a $134 billion valuation.
PhD in distributed computing, then founded two companies before Databricks → combines deep technical grounding with a builder's instinct; unlikely to be impressed by hand-waving on architecture.
Active across Goldman Sachs, a16z, Stratechery, CNBC, and the Databricks blog → comfortable at every altitude from academic research to investor narrative; adjusts register for the room.
Public writing themes are tightly scoped — AI in the enterprise, lakehouse architecture, open-source AI — → he stays in his lane and goes deep; breadth-of-interest small talk will land flat.
Possibly — the MBA alongside the PhD and the co-founding of Hive Streaming and Peerialism AB before Databricks suggests he was thinking about the commercial layer of technology early, not just the research.
Recurring keynote speaker at Data + AI Summit and high-profile external stages → used to setting the room's agenda; comes with a prepared point of view.

Conversation tips

→ Reference a specific product launch — Lakewatch, Agent Bricks, or Lakebase — and ask about the architectural decision behind it; he responds to technical specificity, not category-level questions.
→ The open-source angle is genuine and a point of pride — ask how Spark's open-source roots shaped how Databricks thinks about the SAP or Anthropic partnerships.
→ He has given the enterprise AI thesis many times (a16z, Goldman, Stratechery) — don't ask him to re-explain it; engage with a specific part of it you found interesting or want to push on.
→ The Stanford ETL talk on 'lessons from a large founding team' signals he has thought hard about co-founder dynamics at scale — that's a richer thread than generic founder-story questions.
→ He posted publicly promoting Meta's Llama 10x growth — open-source AI models are clearly a worldview, not just a talking point; you can probe where he thinks open vs. closed model competition goes.

Toolbox

Openers

Open on Lakewatch — Databricks launched an AI-driven cybersecurity SIEM in March 2026 powered by Anthropic's Claude, with Adobe and Dropbox as early customers. It's a sharp left turn from data platforms into security, and asking what made that the right move now will get a substantive answer.
Reference the Stanford ETL talk on 'lessons from a large founding team' — he gave that talk specifically about the dynamics of co-founding Databricks with six other academics. It's a named, specific piece of his public thinking most people skip.
Bring up the SAP Databricks partnership announcement — he personally announced it on LinkedIn and it signals a bet on legacy enterprise data estates, not just cloud-native customers. It's a pointed strategic choice worth interrogating.

Discovery questions

Databricks now has $1.4 billion in AI product revenue — how has the product motion for Agent Bricks and Lakebase differed from selling the core lakehouse, where the buyer and the use case were more established?
You've publicly backed Llama's open-source growth and integrated Anthropic's Claude into Lakewatch — how do you think about where open models win versus where frontier closed models are genuinely necessary for enterprise customers?
With MosaicML, Tabular, and Neon all acquired in quick succession, and Antimatter and SiftD added in 2026 — what's the integration theory? Are these being run as distinct products or collapsed into the platform?

Avoid

Don't ask a generic 'what's your AI strategy?' question — he has given that answer dozens of times on record at Goldman, a16z, and Stratechery; come with a specific angle or the conversation will feel like homework he's already graded.

Make it yours