Ali Ghodsi

Who they are

Ali Ghodsi is Co-founder and CEO of Databricks — started the company in 2013 out of UC Berkeley's AMPLab and recently open-sourced Omnigent, an AI agent framework on GitHub.

Person

Ali Ghodsi co-founded Databricks in 2013 when it was a seed-stage startup spun out of the UC Berkeley AMPLab by the 'Apache Spark Seven' — a small engineering team with a managed Spark service and modest early revenue. The academic foundation is deep: MSc in Computer Engineering and an MBA from Mid-Sweden University, then a PhD in distributed computing from KTH Royal Institute of Technology in 2006, followed by post-doctoral research at UC Berkeley. From KTH he co-founded Peerialism AB, a Stockholm-based peer-to-peer data transfer startup, and was involved with Hive Streaming before moving to Berkeley where the Databricks idea took shape. At Databricks he served as VP of Engineering and Product Management before stepping into the CEO role, growing the company from that early Spark service into a platform used by over 20,000 customers including more than 60% of the Fortune 500. Alongside the main company he recently open-sourced Omnigent, an AI agent framework with contextual policies and session-based features. He writes on the Databricks blog, contributes to TechCrunch, and has been a recurring voice on a16z's Boss Talk, Stratechery, Goldman Sachs Talks at GS, and the Data + AI Summit keynote stage — consistently on enterprise AI, lakehouse architecture, open-source, and AI governance. He holds an adjunct professorship at UC Berkeley, keeping one foot in the research world.

Company

The most recent development: in summer 2026, Databricks secured a new funding round valuing the company at approximately $188 billion — roughly a 40% step up from the $134 billion valuation it reached in December 2025 when it raised over $4 billion in a Series L round. That Series L came alongside a $5 billion total raise (including $2 billion in new debt capacity) in February 2026, with investors including Goldman Sachs, Morgan Stanley, and Qatar Investment Authority. The company has crossed a $5.4 billion annualized revenue run rate with $1.4 billion coming from AI products, and is operating with more than $7 billion in combined equity and debt. In March 2026 it launched Lakewatch, an AI-driven open SIEM cybersecurity product powered by Anthropic's Claude models, with Adobe and Dropbox among early customers, backed by the acquisitions of Antimatter and SiftD. An IPO is anticipated, with a valuation target reported between $165 billion and $175 billion, potentially in 2027.

Market

Databricks competes directly with Snowflake and Microsoft Fabric in cloud data platforms, and faces pressure from hyperscalers — AWS, Google Cloud, and Azure — who bundle competing data and AI services with existing enterprise relationships. The broader market is moving fast toward unified data and AI infrastructure, which has driven Databricks' growth but also intensified rivalry; geopolitical complexity adds another layer, as investments from sovereign wealth funds like Qatar Investment Authority are subject to US regulatory scrutiny, and data-sovereignty concerns in Europe are pushing some organizations toward regional alternatives.

Network

Ghodsi's closest professional orbit is the Databricks leadership team — Matei Zaharia, CTO and fellow co-founder, is his most prominent long-standing collaborator. He has amplified Jonathan Frankle, Databricks' Chief AI Scientist, on LinkedIn, signaling internal elevation of the AI research function. On the investor side, Ben Horowitz of Andreessen Horowitz is a recurring public interlocutor, including a joint a16z podcast appearance on enterprise AI.

Matei Zaharia· CTO and Co-founder, Databricks
Jonathan Frankle· Chief AI Scientist, Databricks
Ben Horowitz· Co-founder and General Partner, Andreessen Horowitz

How they likely show up

Long tenure at Databricks (co-founder since 2013, through seed stage to $188B valuation) → thinks in decade-long arcs; unlikely to be impressed by short-horizon thinking.
Moved from VP Engineering & Product to CEO within the same company → operator first, not a career CEO; still close to the product and technical decisions.
PhD in distributed computing + post-doc at Berkeley + adjunct professorship → frames problems in systems terms; respects intellectual rigor over buzzwords.
Recurring appearances on Goldman Sachs Talks at GS, Stratechery, a16z, and CNBC → comfortable in high-stakes public venues; used to holding his own with financially and technically sophisticated audiences.
Publicly amplified Omnigent open-source project on LinkedIn → still personally engaged with hands-on technical work outside the CEO role; not purely a figurehead.
Co-founded Peerialism AB and Hive Streaming before Databricks → has a founder's tolerance for early-stage ambiguity, not just a scaled-company operator mindset.

Conversation tips

→ Reference specific product moves — Lakewatch, Lakebase, or Agent Bricks — rather than speaking about Databricks generically; he'll engage more on the specifics of what they're building.
→ The open-source angle (Apache Spark, Omnigent) is a genuine conviction, not just marketing — asking about the tension between open-source and enterprise monetization will get a real answer.
→ He's done the a16z podcast, Goldman Sachs stage, and Stratechery interview — he's heard the macro AI narrative many times. Come with a specific, grounded question rather than a broad 'where is AI going?' opener.
→ He bridges academic rigor and commercial scale — referencing the AMPLab origins or the KTH distributed systems background shows you know the story and will land better than pure business framing.
→ Ask about the IPO timing tension — $188B valuation, $5.4B ARR, clear path — and what's actually driving the decision to wait; it's the most live strategic question on his plate right now.

Toolbox

Openers

Open on Omnigent — he recently open-sourced an AI agent framework on GitHub with contextual policies and session-based features, which is a pointed side bet from someone running a $188B company; ask what problem he was solving that Databricks itself wasn't.
Lead with Lakewatch — Databricks launched an AI-driven open SIEM in March 2026 powered by Anthropic's Claude, with Adobe and Dropbox as early customers; it's a new category move that signals where he thinks the platform is going beyond data.
Reference the Stanford Technology Ventures talk 'Lessons from a Large Founding Team' — Databricks launched with seven co-founders from Berkeley, an unusually large group, and how he navigated that founding dynamic is a specific and revealing story.

Discovery questions

Lakewatch moves Databricks into cybersecurity — how do you decide which new categories the Lakehouse platform should expand into versus where you stay a platform and let partners build?
You've been building Databricks for over a decade from a seed-stage Spark service to a $188 billion company — at what point did the job of CEO stop feeling like an extension of the technical founder role and start feeling like a different thing entirely?
With the $5.4 billion annualized revenue run rate and $1.4 billion already coming from AI products, what's the real reason the IPO is targeting 2027 rather than now — is it market conditions, internal readiness, or something else?

Avoid

Don't pitch him on generic 'AI transformation' framing — his public writing and talks are consistently about specific infrastructure choices, open-source tradeoffs, and enterprise data architecture, not broad AI enthusiasm.

Make it yours