Recommended Articles
Fundraising

Late-Stage Money Still Flows to AI Security: Island, Cyera, Armadin Show Where Dollars Are Defensible

By Angeled · October 8, 2026 · Fundraising
Late-Stage Money Still Flows to AI Security: Island, Cyera, Armadin Show Where Dollars Are Defensible
Late-Stage Money Still Flows to AI Security: Island, Cyera, Armadin Show Where Dollars Are Defensible

Over $1 billion in a week says security is not done

Cybersecurity just had one of its biggest weeks of 2026. Island raised a $400 million Series F at a $6.4 billion valuation led by Evolution Equity Partners — more than double its 2024 mark. Cyera took $400 million in a Series G extension from Goldman Sachs Alternatives, pushing total funding past $2.7 billion at a valuation above $12 billion. And Armadin, the autonomous-security startup from Kevin Mandia, closed a $255.5 million Series B at a valuation above $2.5 billion, co-led by Andreessen Horowitz and Accel, just seven months after launch.

Three rounds, three different security theses, one shared bet: the bill for securing AI — and the data it touches — is still early.

What investors are actually buying

The three raises map to three distinct defensible layers:

The control plane (Island). Island sells an enterprise browser positioned as an "agentic control plane" — the layer where employees and AI agents interact with applications and data. The company reports doubling annual recurring revenue every fiscal year since its 2022 launch and now employs about 1,000 people. The thesis: as agents do more work inside the browser, the browser becomes the enforcement point for visibility, control, and auditing.

The data layer (Cyera). Cyera governs the data that people and agents read and write — data security posture management, data loss prevention, and identity access unified. Its July acquisition of Oasis Security for a reported $1 billion shows the category consolidating around platforms, not point tools. Cumulative $2.7 billion raised says data security for agents is being priced like a platform.

The adversary layer (Armadin). Armadin runs swarms of autonomous AI agents that simulate attackers, probing surfaces and chaining minor findings into validated attack paths. With a mean time to exploit measured in days and access hand-offs in seconds, continuous adversarial testing is becoming infrastructure rather than a periodic service.

Where security dollars are defensible for founders

For founders eyeing the space, the pattern in these rounds is instructive:

1. Sell the platform, not the feature. All three winners are consolidating adjacent functions — browser plus DLP plus SASE; DSPM plus DLP plus identity; scanning plus validation plus kill-chain simulation. Single-feature security tools are getting absorbed. If you are building a feature, build it as the wedge into a platform story.

2. Agent-centricity is the price of admission. Every one of these pitches centers on AI agents — governing them, securing the data they touch, or attacking like them. Security products that treat agents as an afterthought will not clear diligence in 2026.

3. Revenue proof separates the funded from the hopeful. Island points to doubled ARR every year since 2022. Snorkel AI, which raised $350 million alongside these rounds, reports over $375 million in ARR. At these valuations, the question is no longer "is the category real?" but "can you show the revenue curve?" Seed and Series A founders should build toward that bar early.

4. The mid-market gap is still open. The billion-dollar rounds all chase the enterprise. Regulated mid-market companies — regional banks, healthcare systems, manufacturers — face the same AI security pressures with smaller budgets and less patience. That is where the next fundable generation may emerge.

Investor takeaways

Late-stage investors are writing $400 million checks to companies with proven revenue and category-creation credentials. The bar for new security startups is correspondingly higher: agent-native architecture, platform breadth, and a credible path to eight-figure ARR. The opportunity is real — AI has expanded every attack surface — but "AI security" alone is no longer a pitch. The pitch is which layer you own.

Bottom line: Security remains one of the few categories where late-stage money moves freely, because the threat is measurable and the budgets are real. Founders should build for the agent era's control points — and bring revenue evidence early.

Angeled.com connects founders with investors.

Read Next