[Palo Alto Networks], [CrowdStrike] and [Zscaler] now have a fresh agent-AI problem — Pluto Security just made AI-agent security its own $42M venture category

Pluto Security came out of stealth on Oct. 13, 2026 with $42M of total funding — $8.5M seed led by TLV Partners (Armis founders participated) and ~$34M Series A led by Jump Capital with Skinos Ventures (Check Point co-founder Shlomo Kramer's firm), Ansa Capital, and AlphaDrive Ventures on the cap table. The Israel- and U.S.-based company, founded in 2025, is led by CEO Shahar Bahat and CPO Tal Ofer Regev, both former Palo Alto Networks executives.

Pluto's product centers on monitoring and controlling employee-built AI tools — connectors, add-ons and agents — and intervening while an agent is executing, rather than blocking at the proxy. The 'Plutonium' platform released in April gives any enterprise a no-cost scanner for publicly available AI skills and MCP servers, mirroring the threat-landscape framing HiddenLayer, Zenity and Reco have pushed since2025.

Pluto is the fifth specialist raise in eight weeks. Between Aug. 1 and Oct. 13, 2026, at least five venture rounds for AI-agent-security vendors were disclosed, totaling roughly $352M in new money on top of earlier rounds. Reco separately added $55M on top of its February $30M Series B, taking its own cumulative to $140M.

The category is fragmenting by attack-surface layer: MCP/identity (Reco, Okta, CyberArk), MCP runtime (Zenity, HiddenLayer), data-exposure (Cymphony), insider AI usage (Pluto), and inside-the-model monitoring (Goodfire). That fragmentation is the structural driver behind the bundle math re-pricing — see Chapter 2.

The Big Three platform vendors are no longer competing with each other in a vacuum — they are competing against the cohort above while their own growth already prices the AI-agent story. [CrowdStrike] delivered Q2 FY27 revenue of $1.47B (+26% YoY) and ARR of $5.84B (+25% YoY) with net new ARR of $332.8M (+51% YoY) — an acceleration that coincided with the Aug. 28–Sep. 1 unveiling of Falcon Guardian. [Palo Alto Networks] posted Q4 FY26 revenue of $3.41B (+34% YoY) and closed the $25B CyberArk deal Feb. 11, 2026, bringing identity-security machinery inside the Prisma AIRS 3.0 umbrella. [Zscaler] reported Q4 FY26 revenue of $898.2M (+25% YoY) and ARR of $3.77B, with AI Broker + Endpoint AI Security live since Jun. 9, 2026.

The cohort math: a customer adopting [CrowdStrike] for endpoint AIDR, [Zscaler] for zero-trust/A2A broker, [Palo Alto Networks] for identity (now via CyberArk) and Prisma AIRS for agent-runtime, plus one of Reco/Zenity/HiddenLayer/Pluto for shadow-agent governance, is paying 4–5 separate vendors to cover what each of the three claims as a 'single pane of glass.' The platform-priced multiples — [CrowdStrike] trades at roughly 49x trailing sales, [Palo Alto Networks] at 28x, [Zscaler] at 11x — are being stress-tested against the question: does the platform own this stack, or do five specialists bolt on top of it?

Each of the three is picking a different layer to plant the flag. The numbers behind each decision — and the gap each leaves for a specialist — are below.

[Palo Alto Networks]'s answer is the most expensive in absolute dollars. The CyberArk deal — $45 cash plus 2.2005 PANW shares per CyberArk share, closed Feb. 11, 2026 — folded in the only publicly listed pure-play in machine-identity at scale, and folded it directly into the cycle Prisma AIRS 3.0 shipped roughly six weeks later (Mar. 23, 2026). [Zscaler]'s answer is the cheapest to ship — the AI Broker and Endpoint AI Security are software, gated by the Zero Trust Exchange — but they require the customer to already have AI traffic routed through ZIA/ZPA, which a meaningful share of large enterprises do not. [CrowdStrike]'s answer inherits the endpoint footprint and is the closest to a swap-in, but the AI gateway is still pre-beta and the cross-app data-lineage layer is the one piece still in specialists' hands.

The supply-chain layer that matters most for valuation is the MCP perimeter. Researchers had documented ~8,000 MCP servers exposed on the public internet without authentication by early 2026 (Cloud Security Alliance, Apr 2026), with 30+ CVEs filed in the first two months of the year. Pluto's free 'Plutonium' MCP scanner — and Zenity's agentic-runtime scanning, and HiddenLayer's red-teaming of MCP — is the upstream defense each platform is trying to absorb.

Short-term (days–quarters): the catalysts are [CrowdStrike]'s Falcon AI Gateway GA in Q4 FY27 (currently pre-beta), [Zscaler]'s first quarter of AI Broker attach in the Q1 FY27 print, [Palo Alto Networks]'s first full quarter of CyberArk consolidation in the Q1 FY27 print, and Reco's next reported ARR triple. Pluto's free Plutonium catalog will be a leading indicator of MCP-exposure volume in enterprise environments. The first quoted dollar contribution from agent-security SKUs to any of the Big Three will likely be a footnote in the FY27 10-K cycle.

Long-term (1–3 years): the structural read is whether agent-security stays a $1B-ritual category (the cohort's current 12-month enterprise funding run-rate implies it) or consolidates back into the platforms via M&A — [Palo Alto Networks] has the precedent (CyberArk $25B; the Prisma AIRS family originally came from the Protect AI acquisition). [Okta] is the most natural listed-company consolidator if standards force a single agent-identity framework. The [Check Point Software] lineage — Kramer's Skinos in Pluto's Series A — keeps the Israeli consolidation channel open.

Pluto's $42M raise is not the cause of the agent-security wave; it is the marker that the wave has its own venture bucket. Over six months, $400M+ of new specialist capital (Zenity, Cymphony, HiddenLayer, Reco, Pluto) plus the documented MCP-exposure perimeter produced by the Cloud Security Alliance have forced the platform vendors to commit roadmap — [CrowdStrike] shipped Falcon Guardian, [Palo Alto Networks] shipped Prisma AIRS 3.0 and bought CyberArk for $25B, [Zscaler] shipped AI Broker + Endpoint AI Security + AI Access Graph.

The investable view is two-sided. The platforms are priced for narrative: [CrowdStrike] trades at roughly 49x trailing sales with ARR still compounding at 25%; [Palo Alto Networks] at 28x with the CyberArk dilution still hitting EPS. The specialists are priced for category-formation: Reco's valuation more than doubled in seven months, HiddenLayer and Zenity both raised at materially higher valuations inside a quarter. The clearest near-term read is that the platforms own the seat today and the specialists own the surface tomorrow — and Pluto's $42M exit, with 70% inbound pipeline and $2M ARR in six months, is the proof point that enterprises are now paying for both.