Ali Ghodsi
Last reviewed
Sources
13 citations
Review status
Source-backed
Revision
v1 · 1,553 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Sources
13 citations
Review status
Source-backed
Revision
v1 · 1,553 words
Add missing citations, update stale details, or suggest a clearer explanation.
Ali Ghodsi (born 1978) is a Swedish-American computer scientist and entrepreneur of Iranian origin who is the co-founder and chief executive officer of Databricks, one of the most valuable private companies in the world and a leading vendor of data and artificial intelligence software. He is also an adjunct professor of computer science at the University of California, Berkeley, where he was a member of the research group behind Apache Spark and co-created the cluster manager Apache Mesos. Under his leadership since 2016, Databricks has grown from a startup commercializing an open-source analytics engine into a broad data and AI platform that was valued at 134 billion dollars by the end of 2025.[1][2]
Ghodsi was born in Iran in 1978 and emigrated with his family to Sweden as a young child, growing up in a Persian household in his adopted country. He has described teaching himself to program as a boy on a Commodore 64, an early fascination that drew him toward computer engineering.[1][3]
He pursued most of his higher education in Sweden. He earned a master of science in computer engineering in 2002 and an MBA in logistics and marketing in 2003, both from Mid-Sweden University. In 2006 he completed a PhD in computer science at the KTH Royal Institute of Technology in Stockholm, working under Seif Haridi in the area of distributed computing; his dissertation addressed algorithms for distributed hash tables, a core building block of large peer-to-peer and distributed systems. He remained at KTH as a faculty member, serving as an assistant professor from roughly 2008 to 2009, and co-founded Peerialism, a Swedish peer-to-peer data transfer startup, before relocating to the United States.[1][3]
In 2009 Ghodsi joined the University of California, Berkeley as a visiting scholar, and he became a long-term member of the AMPLab, the research group whose work on large-scale data processing produced several influential open-source projects. There he collaborated with researchers including Ion Stoica, Scott Shenker, Michael Franklin, and Matei Zaharia.[1]
Ghodsi's own research focused on resource sharing and cluster management. He co-authored the original work on Apache Mesos, a system for sharing a single cluster of machines among many distributed applications, and he co-invented Dominant Resource Fairness, a widely cited method for fairly allocating several resource types at once in such clusters. The same Berkeley group created Apache Spark, the open-source engine for large-scale data processing that became one of the most popular projects in the big-data ecosystem, and Ghodsi is counted among its original creators. Spark, first built by Zaharia around 2009, offered a faster and more flexible alternative to the earlier MapReduce model, and it would become the technical foundation of the company he helped start.[1][4]
In 2013, Ghodsi and six other members of the Berkeley group, Ion Stoica, Matei Zaharia, Patrick Wendell, Reynold Xin, Andy Konwinski, and Arsalan Tavakoli-Shiraji, founded Databricks to commercialize Apache Spark. The company, headquartered in San Francisco, offered a cloud platform for data engineering, analytics, and machine learning built around Spark.[2][4]
Ghodsi initially served as vice president of engineering and product management while Stoica was chief executive. In January 2016, Stoica returned to his faculty role at Berkeley and Ghodsi became CEO. Over the following years he reoriented the company around a new architectural idea that Databricks branded the data lakehouse. Introduced in company writing around 2020 and built on the open Delta Lake storage format, the lakehouse combined the low-cost, flexible storage of a data lake with the reliability, governance, and performance traditionally associated with a data warehouse, aiming to let organizations run analytics and AI workloads on a single copy of their data. The concept became the centerpiece of Databricks' product and marketing strategy, was later rebranded as the Data Intelligence Platform, and was widely adopted across the industry as a term of art.[5][6]
As generative AI moved to the center of enterprise computing, Ghodsi repositioned Databricks as an AI company built on top of its data platform. In June 2023 the company agreed to acquire MosaicML, a generative AI startup co-founded by Naveen Rao and Hanlin Tang, for about 1.3 billion dollars, and rebranded its technology as Mosaic AI. The acquisition gave Databricks tooling for training and serving custom large language models on enterprise data.[7]
In March 2024 the company released DBRX, an open large language model built by its Mosaic team. A mixture-of-experts model with 132 billion total parameters, of which about 36 billion are active for any given input, DBRX was among the strongest openly available models at its launch, outperforming older open models and the original GPT-3.5 on standard benchmarks. Databricks positioned it less as a consumer chatbot than as proof that enterprises could build and own high-quality models trained on their own data.[8]
Ghodsi has continued to expand the AI portfolio. At the Data + AI Summit in June 2025, Databricks introduced Agent Bricks, a system for automatically building, evaluating, and tuning task-specific AI agents using a customer's own data. Around the same time the company launched Lakebase, a fully managed serverless PostgreSQL database aimed at AI applications, following its roughly 1 billion dollar acquisition of the database startup Neon. These additions extended the platform from analytics into operational databases and autonomous agents, the areas Ghodsi has identified as central to Databricks' next phase of growth.[9][10]
Databricks has stayed private under Ghodsi while raising an unusually long series of large late-stage rounds, a strategy that has supplied growth capital and employee liquidity without an initial public offering. The company's valuation climbed from 28 billion dollars in early 2021 to 134 billion dollars by the end of 2025, making it one of the most valuable venture-backed companies in the world. Ghodsi has repeatedly said the company is in no rush to go public.[2][11]
The following table summarizes selected funding rounds.
| Round | Date | Amount | Valuation | Lead investors |
|---|---|---|---|---|
| Series A | Sep 2013 | 13.9 million dollars | not disclosed | Andreessen Horowitz |
| Series G | Feb 2021 | 1 billion dollars | 28 billion dollars | Franklin Templeton |
| Series H | Aug 2021 | 1.6 billion dollars | 38 billion dollars | Counterpoint Global (Morgan Stanley) |
| Series I | Sep 2023 | over 500 million dollars | 43 billion dollars | T. Rowe Price (with NVIDIA) |
| Series J | Dec 2024 | 10 billion dollars | 62 billion dollars | Thrive Capital |
| Series K | Aug 2025 | not disclosed | over 100 billion dollars | Andreessen Horowitz, Insight Partners, MGX, Thrive Capital, WCM |
| Series L | Dec 2025 to Feb 2026 | about 5 billion dollars equity plus about 2 billion debt | 134 billion dollars | Insight Partners, Fidelity, J.P. Morgan Asset Management |
The 10 billion dollar Series J announced in December 2024 ranked among the largest venture rounds ever raised, and the Series K round in August 2025 was the first to push Databricks past a 100 billion dollar valuation. By late 2025 the company reported an annualized revenue run-rate above 4.8 billion dollars, growing more than 55 percent year over year, with its AI products alone exceeding a 1 billion dollar run-rate, and it announced positive free cash flow over the prior twelve months.[11][12] Ghodsi's personal wealth has risen with the company; various 2025 to 2026 estimates have placed his net worth in the billions of dollars, and he has publicly framed a long-term ambition for Databricks eventually to reach a one trillion dollar valuation.[13]
As of 2026, Ghodsi remains chief executive of Databricks and an adjunct professor at UC Berkeley, and he is widely regarded as one of the central figures in the commercialization of big-data and enterprise AI software.[1][2]