Databricks Announces AI Security Product 'Lakewatch'—Enterprise Strategy Revealed Through Two Acquisitions and Claude Integration
In 2026, Databricks, known for its cloud data analytics platform, announced a new AI security product called 'Lakewatch'.Alongside this, the acquisition of two startups was also revealed. Following the $5 billion funding round completed last month, and for a company like Databricks with multi-billion dollar annual revenue, this move signals the beginning of an aggressive M&A strategy.
1. What is Lakewatch—Integrating Data Infrastructure and SIEM
1-1. Why Databricks is entering the security SIEM market
SIEM stands for 'Security Information and Event Management,' a security foundation that collects and analyzes massive amounts of log and event data within an enterprise in real-time to detect and investigate threats. Traditional major players have included Splunk, Microsoft Sentinel, and IBM QRadar.
The defining feature of Lakewatch is that it repurposes Databricks' inherent 'large-scale data storage and processing capabilities' for security purposes, and is designed with AI agents powered by Anthropic's Claude to handle threat detection and investigation. The name 'Lakewatch' is fitting, signifying the fusion of a data lakehouse and a SIEM.
1-2. Why Anthropic's Claude?
Databricks' choice of Claude is related to the context of OpenAI's expansion into the AWS government sector, which I discussed last week. In the enterprise AI market, there is a growing preference for Claude due to its superiority in 'code understanding, long-context windows, and agentic reasoning,' making it well-suited for the massive document processing required for security log analysis.
As shown in RAMP data, with the adoption rate of Anthropic in the enterprise sector rising rapidly, Databricks' selection of Claude is one example that reinforces this trend.
2. Two Acquisitions—Big Technical Assets from Small Teams
2-1. Antimatter—Secure Data Protection Technology for Agents
Antimatter is a startup founded in 2022 by security researcher Andrew Krioukov, which raised approximately $12 million led by New Enterprise Associates. According to LinkedIn, the company has fewer than 50 employees.
The company was developing technology called a 'data control plane,' a mechanism that allows enterprises to safely deploy AI agents while protecting sensitive data. They demonstrated this on stage at the 2024 RSA Innovation Sandbox Contest, gaining industry recognition. Krioukov has been at Databricks for several months and currently leads the Lakewatch team.
2-2. SiftD.ai—A Very Young Startup Released in November 2025
SiftD.ai was a smaller acquisition with only a few team members. Their product was released in November 2025, just a few months before the acquisition was completed. They were developing an 'interactive notebook,' a tool where humans and AI agents collaborate on data analysis and investigation using an interface similar to Jupyter Notebooks.
Co-founder Steve Zhang is the former Chief Scientist at Splunk and the creator of 'Search Processing Language (SPL),' which is widely used as the standard language for security log processing. The Databricks team knew Zhang from his Splunk days, making this acquisition essentially an 'acqui-hire' aimed directly at recruiting the designer of SPL.
3. 'Always Looking for the Next Thing'—Implications of a Continuous Acquisition Strategy
3-1. $5 Billion in Ammunition and the Background of Accelerated M&A
With the $5 billion funding round completed last month, Databricks has sufficient financial leeway. A company spokesperson stated, 'We are always looking for the next thing. Our goal is to stay ahead of the market and fill the gaps in our customers' needs.'
This statement suggests that continuous startup acquisitions will follow. While both of these companies were small, for Databricks, this was an efficient way to achieve two goals: acquiring technical IP and securing talent.
3-2. The Competitive Landscape of the Enterprise AI Security Market
As AI agents become deeply integrated into core enterprise systems, 'security for agents accessing sensitive data' is rapidly gaining importance. As Arthur Mensch of Mistral AI stated in his GTC keynote, 'enterprise data access control and context engines are the core technical challenges,' and companies that can provide answers to this problem hold significant business opportunities.
Databricks is targeting exactly that space, setting the stage for competition with existing players such as Splunk (now acquired by Cisco for $28 billion), Microsoft Sentinel, and Panther Labs.
4. Implications for Investors—Perspectives on Databricks and Related Stocks
4-1. Databricks itself is a private company before IPO
Databricks is currently private, and direct investment is not possible. However, as a related public company context, the adoption of Claude from Anthropic (which is private but accessible via related funds) contributes to Anthropic's revenue and, through data centers, also leads to demand for Nvidia and AWS.
4-2. Trends in the SIEM Security Market
Splunk, a leader in the existing SIEM market, was acquired by Cisco for $28 billion, and the transition to an AI-native approach is progressing rapidly. By entering the security space, players like Databricks that 'possess both data and AI' may increase competitive pressure on traditional security-focused vendors. It is worth monitoring the impact on publicly traded security companies such as Palo Alto Networks and CrowdStrike.
Conclusion
Databricks' approach is not to simply pour the funds obtained from massive financing into advertising and sales, but to quickly acquire small teams capable of solving specific technical challenges. Antimatter's 'integration of agents and data protection' and SiftD.ai's 'notebook experience where humans and AI collaborate'—these two are the core functions of Lakewatch, and they were able to reach the market faster than through independent development. This announcement demonstrates that competitive advantage in the AI era is shifting toward 'who can integrate the best talent and technology the fastest.'