Palo Alto Networks agrees to acquire Israeli endpoint security company Koi Security, sources say for ~$400M; Koi raised $48M across two rounds, per PitchBook
Koi has developed a software engine that uses large language models and AI agents to detect malware and vulnerabilities in applications.
Context & Ripple Effects
The reported agreement follows January reports that Palo Alto Networks was in talks for Koi and comes after Koi's $38M Series A for its centralized endpoint-protection layer. It also extends Palo Alto Networks' established use of acquisitions to add security capabilities, including its earlier purchase of cloud-data security company Dig Security.
First-order effects
- Palo Alto Networks would add Koi's endpoint-security engine, which uses LLMs and AI agents to identify malware and application vulnerabilities, to its security portfolio.
- Koi's investors and team would move from building an independent endpoint-security company to integrating its technology within a larger cybersecurity vendor, subject to completion of the reported deal.
Second-order effects
- Endpoint-security rivals face added pressure to demonstrate how their own products detect threats and vulnerabilities in AI-enabled applications, rather than treating AI solely as a feature layer.
- Enterprise customers evaluating endpoint tools may gain another reason to favor broader security platforms if Koi's detection capabilities are integrated with Palo Alto Networks' existing offerings.
Third-order effects
- The deal points to continued consolidation around AI-enabled security functions: large platform vendors can buy specialized detection technology instead of developing every capability internally.
- If this pattern persists, independent security startups may be valued increasingly for whether their AI workflows can be embedded into a major platform, narrowing the path to standalone scale.
The trend: Cybersecurity platforms are using acquisitions to assemble AI-native detection and control capabilities across the enterprise attack surface.