Databricks buys two startups to build a new AI security product

Databricks has acquired two startups, Antimatter and SiftD.ai, to build its own AI security product. Both deals were funded by a record fundraising round: in late 2024 the company raised $5 billion, one of the largest rounds in the history of enterprise software.

Databricks is one of the key platforms for data and machine learning in the enterprise segment. The company built its business on the "lakehouse" concept: a single architecture that combines the capabilities of a data warehouse and a data lake. Thousands of companies worldwide use the platform, including hundreds from the Fortune 500. Databricks' annual revenue exceeds $1.6 billion, and its latest valuation was more than $60 billion, making it one of the most valuable private technology businesses in the world.

Antimatter specialized in protecting data when it is shared between organizations. The startup developed encrypted enclave technology that lets several companies work with sensitive data together, for example to train AI models, without exposing the underlying data to one another or to third parties. The problem is becoming more acute as enterprises build RAG systems and AI agents with direct access to confidential corporate knowledge bases.

SiftD.ai tackled a related problem: detecting unauthorized data use and leaks in AI systems. As language models are woven into corporate workflows, companies face new threats, from prompt injection attacks to the unintentional disclosure of confidential data through LLM interfaces. Traditional cybersecurity tools were designed before the era of generative AI and cannot cope with these specific threat vectors.

The AI security market is growing rapidly. By various estimates, it will reach several tens of billions of dollars by the end of the decade. Databricks directly manages the data of thousands of large customers, which puts it in a natural position to offer protection, not only at the storage level but also in the AI application layer built on top of it.

Buying ready-made startups instead of developing the technology in-house lets Databricks shorten its time to market and immediately gain teams with narrow expertise. It fits the company's broader M&A strategy: after the record round, it has steadily been acquiring startups in generative AI, data management and, now, security.

The deals reflect an important trend: large platform players are consolidating niche AI security specialists and preventing independent leaders from emerging in this market. For startups in the field, the window of opportunity is narrowing. The next 12–18 months will likely determine who stays independent and who joins the tech giants.