There is a particular kind of founder who only needs a name. Kevin Mandia is one of them. The man who built Mandiant into the firm governments called when things went badly, and who sold it to Google, has a new company. And investors just handed it a quarter of a billion dollars to teach AI to think like a hacker, in defense of the good guys.

Armadin announced on October 1 that it has raised $255.5 million in Series B funding at a valuation of more than $2.5 billion. The round was co-led by Andreessen Horowitz and Accel, with new money from Bain Capital Ventures and Redpoint, plus a deep bench of returning backers: Google Ventures, Kleiner Perkins, Menlo Ventures, In-Q-Tel, 8VC, and Ballistic Ventures. Total funding now sits at $445 million, for a company that only emerged from stealth seven months ago.

The core idea: fight AI with AI

Let’s be honest about the problem first, because it’s the kind of thing that usually gets framed with doom. Frontier AI models have compressed the time between a vulnerability being disclosed and a working exploit appearing. What used to take attackers weeks can now take hours. The old defenses, a penetration test twice a year and a scanner that spits out a list of findings, were built for a slower world.

Armadin’s answer is an inversion that feels obvious once you hear it: deploy an autonomous swarm of specialized AI agents that reason like a skilled adversary. Not a scanner that flags individual issues in isolation, but agents that chain individually low-severity weaknesses into full, validated attack paths. The company’s description of a kill chain is worth quoting in plain terms: it can run from unauthenticated remote code execution at the perimeter, through lateral movement inside the network, to full cloud compromise. The point is the chain, not the links. Scanners score each finding on its own and miss how they connect. Attackers don’t.

This is the part that should make security teams sit up. Armadin says security teams get to see the exact attack paths an adversary would use in production, with the blast radius of each mapped out, and can cut those paths before anyone exploits them. In other words: you get to watch the heist in rehearsal and lock the doors it would have used.

Why this raise matters right now

Two things make the timing notable. First, seven months after emerging from stealth, the company says it’s already running agentic attack campaigns in production for Fortune 500 enterprises and government customers. That is a very fast path from stealth to production, and it suggests the demand side is urgent. Enterprises aren’t buying a vision here; they’re buying capacity.

Second, the money flooding into AI-native defense is becoming one of the defining investment themes of 2026. Investors have been pouring capital into early-stage startups building protections against AI-driven cyberattacks all year. A $255.5 million Series B at a $2.5B-plus valuation is among the largest raises the category has ever seen, and the investor list reads like a vote of confidence in both the founder and the thesis.

Mandia’s pedigree is doing real work here. He has sold a security company to Google before, and he has been on the receiving end of the nastiest incident-response calls in the industry. When he says periodic testing can’t keep pace anymore, it lands differently than when a first-time founder says it. The track record is the pitch.

The “good-guy red team” model

Zoom out and Armadin represents a structural shift in how security gets bought. The traditional model is expertise-as-a-service: hire a red team for a few weeks, get a report, fix what you can, repeat next year. It’s episodic, expensive, and the attackers don’t take semesters off.

The agentic model is expertise-as-software: the adversary simulation never stops. The swarm keeps probing, keeps chaining findings, keeps updating the map of how an attacker would actually get in. For a Fortune 500 company with cloud estates that change daily, continuous is the only honest answer. Your infrastructure doesn’t pause between pen tests. Why should your testing?

There is a nuance worth holding onto. These systems are powerful, and power in security tooling always raises the dual-use question. But the framing here is firmly defensive: the agents find the paths, the security team closes them. The company exists to make the defense faster than the offense. In a year when AI safety has been a constant drumbeat, a well-capitalized defense-first company is good news for everyone who isn’t an attacker.

What to watch next

Three things will determine whether this raise is remembered as a landmark or just a big number.

Proof of production value. The Fortune 500 claim is the one to watch. If Armadin can show that continuous agentic testing measurably shrinks the window of exposure, competitors will have to match the model, and the whole pen-testing industry reorganizes around it.

The talent magnet effect. $445 million in total funding, a Mandia-led company, and a mission that reads like a spy novel: this is a recruiting machine. In a security talent market that has been brutally tight for years, that matters. The best defenders are going where the hardest problems are.

Pricing the defense premium. Right now, agentic security is enterprise-only by economics. The question is how fast the model gets cheap enough for the mid-market companies that are actually the softest targets. The sooner that happens, the bigger the dent in the attack economy.

The takeaway

Strip away the funding theatrics and the story is simple. The same AI advances that made attacks faster are now being aimed at defense, by one of the most credible security founders alive, with a quarter-billion dollars of fresh fuel. The attackers have had the momentum. This is the market voting, loudly, that the defenders are catching up.

If you’re in New York this week and security talk over dinner sounds fun (it is, trust me), there’s a full lineup of things to do across the city to pair with the conversation. And if the funding news has you dreaming of your own security startup, fuel up properly first: NYC’s best breakfast sandwiches are a fine place to sketch a pitch deck.