Ex-Mandiant CEO Kevin Mandia Raises $255.5 Million at a $2.5 Billion-Plus Valuation for AI Attacker Swarms

Armadin, the AI-native cybersecurity company founded by former Mandiant chief executive Kevin Mandia, said on Oct. 1 that it raised \$255.5 million in a Series B co-led by Andreessen Horowitz and Accel, valuing the business at more than \$2.5 billion. Bain Capital Ventures and Redpoint joined as new investors, alongside returning backers 8VC, Ballistic Ventures, Google Ventures, In-Q-Tel, Kleiner Perkins and Menlo Ventures. The round takes Armadin's disclosed funding to \$445 million, seven months after it came out of stealth.

The pedigree is the pitch. Mandia founded Mandiant in 2004, later led FireEye, and ran the firm that investigated some of the most consequential intrusions of the past two decades until Google closed its acquisition of Mandiant in September 2022. He launched Armadin on March 10, 2026 with \$189.9 million in combined seed and Series A funding led by Accel, which the company described at the time as the largest combined seed and Series A round in cybersecurity history. Co-founder and chief technology officer Travis Lanham was previously a principal engineer and technical lead in Google Cloud Security Operations; co-founder and chief offensive security officer Evan Peña spent 13 years at Mandiant and built and led its red-team organization to more than 200 consultants; co-founder and chief architect David Slater led engineering on Google's SecOps data platform.

The product, which Armadin calls the "Ultimate Attacker," deploys a swarm of specialized AI agents that map an organization's attack surface, probe systems and chain individually low-severity weaknesses into validated kill chains — in the company's description, running from unauthenticated remote code execution at the perimeter through lateral movement to full cloud compromise. Security teams are shown the paths an adversary would use in production along with the blast radius of each, so the paths can be severed before someone else finds them. Armadin's argument is that periodic human-led penetration tests cannot keep pace with a cycle in which frontier models compress the time between disclosure and a working exploit, and that scanner output fails differently by scoring each finding in isolation.

The numbers supporting that argument are the company's own. Armadin said it has run production attack campaigns for Fortune 500 enterprises and government customers, and its platform page lists 443,000 agents launched against real targets, zero false-positive findings and a fastest reported path to domain compromise of 119 seconds. None of those figures has been independently tested, the round announcement names no customers, and the totals describe company-reported platform activity rather than third-party evaluation — a distinction worth keeping in mind when valuations in this category are set on the strength of a demo.

Investors framed the bet in terms of the people as much as the software. "Offense is uniquely advantaged right now," Mandia said in the announcement. "AI lets an attacker find and chain weaknesses faster than any human team can respond." Accel partner Ping Li, who led the earlier round, said the need for AI-powered offense as a defensive posture "has only grown more pressing," while Andreessen Horowitz general partner David George said Mandia "has built a team pairing elite red teamers with world-class AI engineers."

The category is crowded and well funded. Horizon3 announced a \$250 million Series E at a valuation above \$2 billion on Aug. 3, and established security vendors are shipping their own continuous, agentic testing products. Armadin's differentiation rests on the claim that its agents reason and chain weaknesses like a human adversary rather than following a script — the same claim its competitors make. The company says the new capital will go to platform scale, research, training and go-to-market.

The strategic question is whether continuous, machine-speed testing becomes durable software revenue or settles into a high-end services business wrapped around an ambitious AI claim. Agents that probe live production systems also arrive at an awkward moment: regulators have started asking who is accountable when autonomous software causes harm, and the FTC's newly opened inquiry into agent behavior at frontier labs has put the liability question on the record. For Armadin, the asset is Mandia's operating credibility; the liability is that its core metric, the attack path nobody else has found yet, is exactly the kind of result that is hardest to verify from the outside.