Quick Facts
- AI agent security attracted $3.6 billion in venture capital and $96 billion in M&A in a single year, a 270% year-over-year increase in deal value.
- Palo Alto Networks, SentinelOne, Cato Networks, and Check Point have each made acquisitions targeting AI agent security in the past 18 months.
- Gartner projects the AI cybersecurity segment will grow from $10.82 billion in 2024 to $172 billion by 2029, a compound annual growth rate of 73.9%.
AI agent security has become the fastest-forming cybersecurity category since cloud security emerged a decade ago. Venture capital and M&A are pouring in, and established security vendors are buying startups at an accelerating pace.
The numbers show a market moving fast. Momentum Cyber tallied $96 billion across 400 transactions in the most recent full-year period, up 270% in deal value from the prior year. CB Insights recorded 782 AI acquisitions, running 1.5 times the 2024 pace. Eight cybersecurity deals cleared $1 billion in 2025 alone.
What Enterprises Are Deploying
The urgency is driven by rapid adoption. A December 2025 survey found organizations running a mean of roughly 37 agents each. By April 2026, nearly 38% of organizations reported more than 100 agents deployed. Cisco reports 85% of enterprises have agent pilots underway.
CrowdStrike sensors now detect more than 1,800 distinct AI applications running on enterprise endpoints, representing nearly 160 million unique instances. Gartner predicts 40% of enterprise applications will feature embedded task-specific agents by 2026, up from less than 5% in early 2025.
The Deals Reshaping the Space
Palo Alto Networks has moved most aggressively. In April 2025, the company acquired Protect AI for more than $500 million. It also announced plans to buy Israeli startup Koi for approximately $400 million, according to Israeli business outlet Globes, targeting risks created by agentic tools that act with broad system privileges.
SentinelOne completed its acquisition of Prompt Security on Sept. 5, 2025. The total consideration was approximately $133.6 million in cash, 1,555,099 shares of Class A common stock, and 415,109 assumed options. Prompt Security focused on preventing prompt injection, data leakage, and misuse across generative and agentic AI tools.
Other deals have followed. Aim Security was acquired by Cato Networks. Lakera AI was acquired by Check Point. F5 acquired CalypsoAI in January 2026. Alphabet finalized its $32 billion acquisition of Wiz in March 2026, the largest pure-play cybersecurity acquisition on record.
Of 402 AI security vendors tracked, 232, or 58%, were founded between 2023 and 2025. Thirty-five have already been acquired. Acquisitions peaked in 2026 with 19 deals, compared with 12 in 2025 and 4 in 2024.
Where the M&A Map Points
Strategic adviser Itay Sagie, writing for Crunchbase News, argues the real opportunity is not one broad AI security category. Buyers are moving on specific control points: agent identity, data access controls, prompt security, MCP servers, traffic monitoring, and auditability.
Recent deals reflect that logic. Kiteworks acquired Israeli startup Bonfy.AI, which focuses on real-time data classification and policy enforcement. Israeli startup Huskeys raised a $27 million Series A led by Blackstone to secure complex internet traffic generated by autonomous systems.
The underlying problem is non-human identities, which include API keys, service accounts, and authentication tokens created when agents are granted access to enterprise systems. In the rush to deploy agents, many organizations are granting those identities broad permissions without adequate monitoring or governance controls in place.
Gartner named agentic AI the top technology trend of 2025 and predicts 33% of enterprise apps will include agentic AI by 2028, up from less than 1% in 2024. For security vendors and their buyers, the question is no longer whether this market is real. It is which control points they own when the consolidation wave crests.
Read more: The Emerging M&A Map For AI Agent Security
This article was written by an AI agent. Spotted an error? Send a correction and we will fix it.
