top of page

Cisco Nears $350 Million Deal for AI Agent Security Firm Astrix as Cyber Risk Grows

  • 2 days ago
  • 2 min read


Cisco is in advanced talks to acquire Astrix Security, a Tel Aviv based startup specializing in monitoring and securing AI agents for between $250 million and $350 million, according to people familiar with the matter. Astrix has built technology designed to track what AI agents are doing, what data they are accessing, and whether their behavior falls within expected and authorized parameters, capabilities that are becoming increasingly critical as enterprises deploy autonomous AI systems into production environments where a single misbehaving agent could cause significant business or regulatory damage.


The deal would make Cisco one of the first major enterprise technology companies to make a dedicated AI agent security acquisition, staking out a position in a market that barely existed two years ago but is now being treated as a board-level risk by large organizations deploying agentic AI. A critical vulnerability in Marimo, a widely used Python notebook tool, was weaponized by attackers within approximately nine hours of its public disclosure this week, a reminder of how quickly adversaries have been able to exploit AI-adjacent software infrastructure. The US National Cyber Director is separately leading a new government effort to identify vulnerabilities in critical infrastructure that AI systems could be used to attack or exploit.


Cisco's acquisition interest reflects a broader conviction in enterprise technology circles that AI security is the next major security investment cycle, following the cloud security wave of the previous decade. Just as organizations had to rebuild their security architectures when they moved workloads to the cloud, they will need to do so again as AI agents begin executing tasks, accessing APIs, and making decisions autonomously across their technology stacks. The Astrix deal, if completed, would give Cisco a foothold in this emerging category and add it to a growing stable of AI security capabilities alongside its existing network and endpoint security franchises.

bottom of page