The AI-security M&A map (2023–2026)
Who bought whom in the AI governance/security landscape, verified against primary sources (acquirer press releases, SEC filings) during the Phase 3 research pass. Dates and confidence are explicit. This page is the filed-back answer to “which of these vendors is still independent, and who owns the rest?” — the question that dates half a survey’s answer options.
Convention: closed = completion confirmed by a primary source with a date. announced / pending = agreement public but no completion source found as of 2026-06-28. Deal values are marked reported when press-sourced rather than officially disclosed.
The platform roll-ups (who is consolidating)
Palo Alto Networks — building the AI-security platform by acquisition
| Target | What it brought | Status | Date | Value |
|---|---|---|---|---|
| cyberark | Privileged access + machine/NHI identity | Closed | 2026-02-11 | ~$25B |
| portkey | AI gateway (now the Prisma AIRS gateway component) | Closed | 2026-05-29 | undisclosed |
| Protect AI → prisma-airs | Model scanning, red-teaming, posture, runtime | Closed | 2025-07-22 | reported ~$650–700M |
| Talon → prisma-access-browser | Enterprise browser | Closed | 2023-12-28 | reported ~$625M |
PANW is assembling identity (cyberark) + AI gateway (portkey) + AI runtime/posture (prisma-airs) + browser under one roof. For a CTO this means several formerly-independent “best of breed” tools now arrive as one suite — fewer throats to choke, more lock-in.
Cisco — buying the AI trust & SOC layer
| Target | What it brought | Status | Date | Value |
|---|---|---|---|---|
| splunk | SIEM / observability | Closed | 2024-03-18 | $28B |
| Robust Intelligence → cisco-ai-defense | AI runtime security / validation | Closed | ~2024-09-24 | undisclosed |
| galileo | LLM eval / guardrails | Closed (reported) | ~2026-05-22 | undisclosed |
| astrix-security | Non-human / agent identity | Announced intent — pending | 2026-05-04 | reported ~$400M |
Note: the seed listed Astrix as “acquired”; as of 2026-06-28 it is announced intent, not closed. Treat as pending.
CrowdStrike — adding SaaS posture + AI detection/response
| Target | What it brought | Status | Date | Value |
|---|---|---|---|---|
| pangea | AI detection & response (AIDR) / guardrails | Closed | ~2025-09 (announced 2025-09-16) | reported ~$260M |
| adaptive-shield | SaaS security posture (SSPM) | Closed | ~2025-01 (announced 2024-11-06) | reported ~$300M |
| Charlotte AI | Agentic SOC (built in-house, not acquired) | — | — | — |
SentinelOne
| Target | What it brought | Status | Date | Value |
|---|---|---|---|---|
| prompt-security | Prompt-layer DLP / AI runtime | Closed | ~2025-11 (announced 2025-08-05) | reported ~$250M |
F5
| Target | What it brought | Status | Date | Value |
|---|---|---|---|---|
| calypsoai | AI guardrails + red-team (now F5 AI Guardrails / AI Red Team) | Closed | 2025-10-08 | reported ~$180M |
Other notable AI-security tuck-ins
| Target | Acquirer | What it brought | Status | Date | Value |
|---|---|---|---|---|---|
| aim-security | cato-networks | AI/agent security (Cato’s first acquisition) | Announced | 2025-09-03 | reported ~$350M |
| lakera | Check Point | AI guardrails / prompt-injection defense | Announced (close exp. Q4 2025) | 2025-09-16 | reported ~$300M |
| apex-security | Tenable | Agent/AI security → Tenable One | Announced/integrating | 2025-05-29 | reported >$105M |
| promptfoo | OpenAI | Open-source eval/red-team → OpenAI Frontier | Closed | 2026-03-09 | undisclosed |
| trojai | A10 Networks | AI detect (red-team) + defend (runtime firewall) | Announced | 2026-06-15 | undisclosed |
TrojAI→A10 Networks was NOT in the seed CSV — surfaced during research. A recent (2026-06-15) deal; confirm completion.
Deals surfaced during research — NOT in the seed CSV
These were not flagged in the seed; the research pass found and verified them. They materially date several survey answer options.
| Target | Acquirer | What/why | Status | Date |
|---|---|---|---|---|
| veza | ServiceNow | Identity/data-access governance | Closed | 2026-03-02 (~$1B) |
| securiti | Veeam | DSPM / data+AI governance | Closed | 2025-12-11 ($1.725B) |
| symmetry-systems | zscaler | DSPM / data-access graph for AI agents | Announced | 2026-05-21 |
| splxai | zscaler | AI red-team + runtime | Closed | 2025-11-03 (Q1 FY26) |
| stytch | Twilio | Agent/auth identity | Closed | 2025-11-14 |
| natoma | Snowflake | MCP-native NHI/agent governance | Announced (intent) | 2026-05-27 |
| trulens (TruEra) | Snowflake | LLM eval (TruLens stays OSS) | Closed | 2024-05-22 |
| langfuse | ClickHouse | LLM observability (core stays OSS) | Announced | 2026-01-26 |
| helicone | Mintlify | LLM observability (now maintenance mode) | Announced | 2026-03-03 |
| whylabs | Apple | ML monitoring (acqui-hire; product wound down) | ~2025 (med conf.) | — |
| seraphic | CrowdStrike | Enterprise browser security | Announced | 2026-01-13 ($420M) |
| layerx | Akamai | Browser security extension | Announced | 2026-05-14 (~$205M) |
| styra / OPA team | Apple | Authorization (acqui-hire; OPA stays CNCF) | ~2025-08 (med conf.) | — |
Snowflake and Zscaler are each running two-deal AI roll-ups here (Snowflake: TruEra + Natoma; Zscaler: SplxAI + Symmetry Systems). Apple quietly absorbed two open-source-adjacent teams (WhyLabs, Styra/OPA) as acqui-hires.
Independents that are themselves acquirers (consolidating, not consolidated)
1password → Apono (2026-06, JIT/agent access); cranium → Aiceberg (2026-05); varonis → Cyral (2025-03); menlo-security → Votiro (2025-02); bitsight → Cybersixgill (2024); securityscorecard → HyperComply (2025); workos → Warrant/FGA (2024); snyk → Probely + Invariant Labs (2024–25); okta → Auth0 (2021). These remain independent.
Big-platform / infrastructure M&A (context, not AI-native)
| Target | Acquirer | Status | Date | Value |
|---|---|---|---|---|
| wiz | Google / Alphabet | Closed | 2026-03-11 | $32B (largest Google deal) |
| HashiCorp (hashicorp-vault) | IBM | Closed | 2025-02-27 | $6.4B |
| normalyze | Proofpoint | Closed (reported) | ~2024-11/12 | undisclosed |
| sumo-logic | Francisco Partners (take-private) | Closed | 2023-05-12 | ~$1.7B |
| forgerock | Thoma Bravo → merged into Ping Identity | Closed | 2023-08-23 | ~$2.3B |
| weights-and-biases | CoreWeave | Closed | 2025-05-05 | reported ~$1.7B |
Still independent (verified, as of 2026-06-28)
- hiddenlayer — AI model security; $50M Series A (2023), M12 + Moore. Independent.
- witnessai — AI access governance + runtime; $86.5M total, Series B 2026-01. Independent.
- (more confirmed as research waves complete — see each vendor page)
What it means for a hedge-fund CTO
- Half the AI-runtime/guardrail field has been acquired by platform vendors (PANW, Cisco, CrowdStrike, SentinelOne, F5, Check Point, Cato) in 2025–2026. If you standardize on a platform you already own, you may get an AI-firewall/guardrail capability “for free” — check before buying a standalone.
- The independents that remain (hiddenlayer, witnessai, pillar-security, and others pending research) are the ones to diligence on viability vs acquisition risk.
- Survey implication: several answer options are now product lines inside a suite, not companies. The survey-blueprint and each category’s Survey notes flag these.
Sources
Per-deal primary sources are cached under raw/sources/ (acquirer press releases + SEC filings),
cited on each vendor page’s Sources section. This map aggregates those; see the vendor pages for
the specific citation + confidence per claim.
History
- [2026-06-28] Created from Phase 3 Wave 1 verified M&A research (33 vendors). Pending-deal and reported-value caveats preserved. To extend as later waves surface more deals.