Ali Ghodsi

Ali Ghodsi is the co-founder and chief executive of Databricks, the San Francisco-based data and artificial intelligence company he helped build out of a UC Berkeley research lab. Born in Iran and raised largely in Sweden, Ghodsi trained as a computer scientist, earning a doctorate from the KTH Royal Institute of Technology in Stockholm before crossing to California as a visiting scholar in Berkeley's AMPLab, where a small team was rethinking how organisations process huge volumes of data.

That lab produced Apache Spark, the open-source processing engine that superseded Hadoop's MapReduce as the default way to crunch big data, and Apache Mesos, an early cluster-management system. Ghodsi, working alongside Matei Zaharia and others, co-authored foundational papers on both projects and, in 2013, co-founded Databricks to commercialise Spark. He became chief executive in 2016 and has run the company through a decade of exponential growth, culminating in a 2026 valuation north of $100 billion, among the highest ever reached by a private technology company.

Under Ghodsi, Databricks built the 'lakehouse' architecture that merges the flexibility of data lakes with the reliability of data warehouses, a positioning that put it in direct competition with Snowflake and drew it deep into enterprise AI infrastructure as large language models made proprietary data pipelines strategically critical. He steered acquisitions such as Tabular and MosaicML to extend the platform into open table formats and model training, while resisting a public listing for years, betting that private capital markets would fund faster expansion than Wall Street's quarterly scrutiny.

Ghodsi has become one of the more outspoken chief executives on the direction of enterprise AI, arguing publicly that generative systems are already delivering economically meaningful gains and pushing back on what he calls wasteful 'tokenmaxxing' by rival AI labs. He sits on Databricks' board alongside a roster of Silicon Valley investors and continues to frame the company's mission around keeping enterprise data and AI development open and interoperable rather than locked into a single cloud vendor.

Databricks remains privately held as of 2026, with Ghodsi having repeatedly signalled openness to an eventual initial public offering without committing to a date. His continued dual role as both a working computer scientist and a chief executive scaling a company serving thousands of enterprise customers, including a majority of the Fortune 500, has made him one of the defining figures of the modern data-platform era.