Armadin is an AI native cybersecurity company founded by Kevin Mandia and a team of veteran security operators. Its product uses autonomous agents to continuously test how an attacker might get into an organisation and move through its systems.
On 1 October 2026, Armadin announced a US$255.5 million Series B round that valued the company at more than US$2.5 billion. The financing came only seven months after the company emerged from stealth.
What does Armadin do?
Traditional security testing often happens at intervals. A company hires a penetration testing team, the testers look for ways into the network and the company gets a report showing what needs to be fixed.
Armadin is trying to make that process continuous. Its platform uses fleets of AI agents, combined with human security expertise, to look for combinations of weaknesses that could form a real attack path.
The company calls this approach a Hyperattack.
Why use AI agents for security testing?
Modern organisations can have thousands of devices, cloud systems, identities and applications. Finding one weakness is useful, but attackers often succeed by chaining several smaller weaknesses together.
AI agents can explore many possible paths at the same time and repeat tests more frequently than a human red team could.
The goal is not to generate a longer list of theoretical vulnerabilities. Armadin says it is trying to identify exploitable paths that lead to something important, so security teams know what to fix first.
Who founded Armadin?
Kevin Mandia is Armadin’s founder and chief executive. He previously founded Mandiant, one of the best known incident response and threat intelligence companies in cybersecurity.
Google completed its US$5.4 billion acquisition of Mandiant in 2022. Mandia later became a general partner at Ballistic Ventures before returning to operating a security company with Armadin.
The founding team also includes Travis Lanham, Evan Peña and David Slater in senior technical and security roles.
The funding
Armadin announced about US$190 million in Series A funding when it emerged from stealth in March 2026.
Seven months later, the company raised another US$255.5 million in a Series B co led by Andreessen Horowitz and Accel. Bain Capital Ventures and Redpoint joined the round, while existing investors included Google Ventures, In Q Tel, Kleiner Perkins and others.
Armadin says it has now raised US$445 million in total.
What does the US$2.5 billion valuation mean?
The valuation is the private company value implied by the latest financing round. It does not mean Armadin raised US$2.5 billion or has that amount in revenue.
Reuters reported the company’s value at more than US$2.5 billion after the Series B.
Who is using Armadin?
The company says its agentic attack campaigns are already running in production for Fortune 500 companies and government customers.
Because cybersecurity vendors naturally limit details about customer systems, public information about specific deployments is limited.
What was the 17 million action test?
In August 2026, Armadin and TENEX.ai announced what they described as the largest controlled live AI cyberattack exercise at the time.
According to Armadin, around 26,000 agents carried out roughly 17 million offensive actions over three days and identified 38 validated attack paths.
Those figures came from a controlled defensive exercise, not an attack on an unrelated target.
Why Armadin is getting attention
The cybersecurity industry is preparing for attackers to use AI in the same way defenders are beginning to use it: more automation, more parallel work and faster adaptation.
Mandia’s argument is that companies need machine speed testing if they are going to face machine speed attacks.
Whether this approach becomes standard will depend on accuracy, safety and whether the agents find genuinely important paths without creating overwhelming noise.
Is Armadin a replacement for security teams?
No. Armadin’s own product positioning combines autonomous agents with human security specialists.
Security decisions still involve business context, incident response, access control and judgment about what is acceptable risk. Automated testing is one part of that process.
For more explainers on technology companies and online services, see Holyprofweb’s Apps & Websites section.
What we verified
The funding, valuation, founders and product description above come from Armadin’s current site, its financing announcement and independent reporting. Details about customer environments are intentionally limited because of the nature of security work.