Mandiant founder’s AI hacking startup Armadin raises $255.5M from a16z and Accel at a $2.5B+ valuation

Armadin just turned a flashy demo into a very expensive vote of confidence. The California startup raised $255.5 million in a Series B co-led by Andreessen Horowitz and Accel, putting the company at more than $2.5 billion. That takes its total funding to $445 million, which is a lot of capital for a company Kevin Mandia founded only in September 2025.

Mandia is not exactly a newcomer trying a moonshot from the couch. He founded Mandiant in 2004, sold it to Google in a $5.4 billion deal announced in 2022, and now says the same basic lesson applies: attackers who use AI can move faster than defenders built around periodic checks. Armadin’s pitch is that instead of occasional penetration tests, it runs a swarm of specialised AI agents that reason like an attacker across a company’s systems.

The company says it already put that idea through a live test. On August 3, it said it sent 26,000 AI agents against an institution’s network for three days, with permission. The swarm reportedly chained 38 validated attack paths and produced 238 security findings. On paper, that sounds less like a product launch and more like a stress test for the whole category.

The new money comes from a crowded, serious group. Bain Capital Ventures and Redpoint joined as new investors, while 8VC, Ballistic Ventures, GV, In-Q-Tel, Kleiner Perkins, and Menlo Ventures also backed the round. Armadin says it will use the cash for its platform, research, training, and getting the product to customers. No revenue figure was given, and no customer was named.

And the timing matters. Security vendors are now trying to prove they can keep AI agents inside the lines before those agents wander off into someone else’s network. Armadin says its own agents run in locked-down sandboxes, pass through filters, and are watched in real time. That’s the real sale here: not just offensive AI, but offensive AI that supposedly knows where the wall is.