Armadin Raises $255.5M Series B at $2.5B+ Valuation for AI-Powered Cybersecurity

Armadin, the AI cybersecurity startup founded by Mandiant creator Kevin Mandia, has raised $255.5 million in a Series B funding round co-led by Andreessen Horowitz and Accel at a valuation of more than $2.5 billion.

The California-based company is building an AI-powered offensive security platform designed to continuously test enterprise and government systems against increasingly capable cyber threats.

Bain Capital Ventures and Redpoint joined the round as new investors, while 8VC, Ballistic Ventures, GV, In-Q-Tel, Kleiner Perkins and Menlo Ventures also participated.

The latest financing brings Armadin's total funding to $445 million, including $189.9 million raised through its combined seed and Series A rounds announced in March 2026.

The company plans to use the new capital to expand its platform, cybersecurity research, AI training and customer deployment efforts.

From Mandiant to AI-Powered Offensive Security

Armadin was founded in September 2025 by Kevin Mandia alongside chief technology officer Travis Lanham, chief offensive security officer Evan Peña and chief architect David Slater.

Mandia is best known for founding Mandiant in 2004. The cybersecurity company became one of the industry's best-known incident response and threat intelligence businesses before Google agreed to acquire it for $5.4 billion in 2022.

Mandia later co-founded Ballistic Ventures, which led Armadin's seed round and continued backing the company in its Series B.

The entrepreneur is now applying his cybersecurity experience to a different problem: using AI to continuously attack systems before real attackers can exploit them.

By March 2026, Armadin had already grown to more than 60 employees, according to the source.

An AI Swarm That Attacks Like a Real Adversary

Traditional penetration testing generally happens periodically. Security teams or external specialists test a company's infrastructure at scheduled intervals, identify vulnerabilities and provide recommendations.

Armadin argues that this approach is becoming increasingly inadequate as attackers gain access to AI systems capable of searching for vulnerabilities continuously.

The company's platform uses specialized AI agents that reason across an organization's systems like an attacker.

Rather than simply identifying isolated vulnerabilities, the agents attempt to determine whether several weaknesses can be chained together into a meaningful attack path.

A vulnerability that appears relatively minor on its own could become significantly more dangerous when combined with another weakness elsewhere in an organization's infrastructure.

Armadin's system is designed to identify and validate those chains.

The company's approach was demonstrated in a large controlled test announced in August.

With the permission of a live institution, Armadin deployed 26,000 AI agents against the institution's network for three days.

The company said the exercise resulted in 38 validated attack paths and 238 security findings.

The demonstration was intended to show how autonomous AI systems could conduct offensive security testing at a scale that would be difficult for conventional human security teams to replicate.

Controlling Autonomous Cybersecurity Agents

Giving AI agents the ability to conduct offensive security operations creates an obvious challenge: ensuring the agents do not operate beyond their authorized environment.

Armadin says its agents operate inside locked-down sandboxes.

Each action is passed through a filter designed to prevent activity outside the approved scope, while agent behavior is monitored in real time.

The company is also contributing to NVIDIA's open-source OpenShell software, which is designed to establish boundaries around autonomous AI agents.

The need for such safeguards is becoming increasingly important as more companies deploy AI systems capable of interacting with software, networks and external systems.

The cybersecurity industry is therefore facing a two-sided AI race.

Attackers can use AI to discover and exploit weaknesses faster, while defenders can use AI to identify and remediate those weaknesses before they are exploited.

Kevin Mandia's Vision for AI Security

Mandia argues that offensive security has become particularly important in the age of autonomous AI.

His view is that attackers can increasingly identify and chain vulnerabilities faster than conventional security teams can respond.

Armadin's strategy is therefore to continuously train defensive systems against an AI-powered offensive capability.

The company's pitch is not simply automated vulnerability scanning.

Instead, Armadin is attempting to create an autonomous security testing environment that can reason across an organization's infrastructure and discover attack paths that may otherwise remain hidden.

The company is targeting enterprises and government agencies responsible for protecting critical systems.

Investors Continue to Back the Autonomous Cybersecurity Market

Armadin's latest financing comes amid a broader surge of investment into autonomous cybersecurity companies.

Seattle-based XBOW raised $120 million in a Series C in March 2026 at a valuation above $1 billion. It subsequently added another $35 million from strategic investors including NVIDIA's NVentures, Samsung Ventures, Accenture Ventures and SentinelOne's S Ventures.

That brought XBOW's total funding to more than $270 million.

Horizon3, the San Francisco-based company behind the NodeZero security testing platform, raised $250 million in August at a valuation above $2 billion. Its total funding reached $428.5 million.

Together, these companies demonstrate the growing investor appetite for cybersecurity platforms that use AI to automate offensive security testing.

Armadin's $445 million in total funding now puts it ahead of both companies in terms of capital raised.

A Rapidly Growing Company With a High Valuation

The scale of Armadin's latest financing is particularly notable because the company was founded only around 13 months before the Series B.

Its valuation of more than $2.5 billion reflects significant investor confidence in the potential of autonomous cybersecurity.

However, the company has not disclosed revenue figures or named customers in connection with the latest round.

That leaves an important question around the company's commercial maturity.

Armadin has demonstrated that large-scale autonomous security testing is technically possible. The next challenge is turning those demonstrations into a repeatable enterprise product that customers deploy and renew.

For cybersecurity customers, reliability and safety will be critical. AI agents operating against real networks need to identify vulnerabilities without creating additional disruption or moving beyond their authorized scope.

The Race to Automate Offensive Security

Armadin's rise illustrates how quickly AI is changing the cybersecurity market.

Traditional penetration testing depends heavily on human specialists working through defined assessments. AI agents could potentially make that process continuous, allowing organizations to test their environments far more frequently.

At the same time, autonomous agents introduce new risks that require their own safeguards.

That creates a market where offensive AI and defensive AI are developing simultaneously.

With Kevin Mandia's experience building Mandiant, a $445 million funding base and backing from firms including a16z and Accel, Armadin is positioning itself as one of the major players in this emerging category.

The company's next phase will be about proving that its AI-powered offensive capabilities can translate from controlled demonstrations into reliable cybersecurity infrastructure for enterprises and government agencies.

At a valuation above $2.5 billion, investors are betting that autonomous security testing will become a core part of how organizations defend increasingly complex digital environments.