Armadin raises $255.5M Series B at $2.5B+ valuation

How much did Armadin raise in its Series B?

Armadin said on Oct. 1, 2026, that it raised $255.5 million in Series B funding in a round co-led by Andreessen Horowitz and Accel, giving the AI-native cybersecurity company a valuation of more than $2.5 billion, according to Cooley’s coverage of the financing and a separate report from SecurityWeek. The company, based in Palo Alto, California, describes itself as building “effective autonomous security” with AI-native tools designed to help defenders identify and respond to exploitable risk faster than conventional approaches.

The latest round adds to the company’s earlier financing history. Cooley said it had previously advised Armadin on $189.9 million of Seed and Series A funding in March 2026, while SecurityWeek reported that the company first surfaced quietly in late 2025 with an initial $24 million seed investment and formally launched with $189.9 million in funding in March 2026. SecurityWeek also said Armadin has now raised a total of $445 million.

Who backed Armadin’s round?

According to Cooley, the Series B was co-led by Andreessen Horowitz and Accel. New investors included Bain Capital Ventures and Redpoint. Existing investors participating in the round included 8VC, Ballistic Ventures, Google Ventures, In-Q-Tel, Kleiner Perkins and Menlo Ventures. Those names suggest the round brought in a mix of new and returning backers across venture capital and strategic security-focused firms.

The financing was announced as a valuation event rather than an acquisition or debt transaction. The reported terms describe equity funding, with the valuation set at more than $2.5 billion. No deal-completion caveat was included in the supplied reporting beyond the round announcement itself, so the company’s disclosure should be understood as an announced financing, not a secondary-market pricing update or buyout.

What does Armadin’s security platform do?

Armadin says it uses swarms of specialized AI agents to behave like skilled adversaries, probing an organization’s attack surface, attempting to exploit weaknesses, and chaining issues into validated attack paths. In the company’s framing, those paths can run from an internet-facing vulnerability through lateral movement and eventually into cloud resources or sensitive systems. Armadin says the point is to show customers the attack path and potential blast radius so they can focus on exploitable risk instead of disconnected findings.

The company’s pitch is built around a mismatch it says is growing between AI-powered attackers and traditional defenses. Armadin argues that periodic penetration tests and vulnerability scanners cannot provide a continuously updated picture of risk, and founder Kevin Mandia said AI allows attackers to find and chain weaknesses faster than human teams can respond. Those statements reflect the company’s own view of the market and its product category.

SecurityWeek identified Mandia as Armadin’s founder and CEO. It also said he founded Mandiant in 2004, sold it to FireEye in a $1 billion deal in 2014, and that Mandiant was later acquired by Google for $5.4 billion. Armadin’s co-founders include CTO Travis Lanham, chief offensive security officer Evan Peña, and chief architect David Slater.

How will Armadin use the new capital?

Armadin said it will use the new capital to expand its agentic security platform and scale research, model training, and go-to-market operations. The company also said it is already running agentic attack campaigns in production for Fortune 500 companies and government customers, indicating that it is positioning the platform for enterprise and public-sector deployments rather than a purely experimental use case.

The company highlighted one recent live exercise with TENEX.ai in August. According to Armadin, the three-day event involved 1,300 attacks using 26,000 agents and about 17 million offensive actions against more than 25,000 services. Armadin said the exercise produced 238 security findings, including 98 described as significant, and that it chained them into 38 validated attack paths. The company also said the agents operated without privileged credentials, source code access, or whitelisting of security controls.

Armadin said each action is checked through a control layer overseen by a safety model trained with feedback from human security experts. That detail matters for buyers evaluating whether autonomous security tools can be deployed safely in production: the company is not describing unchecked agent behavior, but a supervised workflow intended to constrain actions while still testing realistic attack sequences. Whether that approach proves durable at scale will depend on how the platform performs in live customer environments as the company grows.