Arm-backed Armadin Secures $189.9M to Build AI-Driven Cyberattack Simulation Platform

Armadin will use the cash to double its engineering staff, adding AI researchers and software developers, according to SiliconANGLE. The company says the new hires will accelerate the rollout of its autonomous threat‑simulation engine across Fortune‑500 networks.

The round, led by Arm’s venture arm, also attracted participation from several undisclosed strategic investors, the report adds. At $189.9 million, the seed‑Series A combo sets a new benchmark for cyber‑security financing, eclipsing prior seed deals in the sector.

CEO Kevin Mandia, former FireEye chief, told the outlet the funds will “speed up the creation of realistic, AI‑generated attack scenarios.” He said the platform will automatically craft novel exploits by pulling recent research from arXiv and feeding it to a DeepSeek‑based director that guides the simulation, a method described in a public GitHub repository linked to the project.

The technology builds on Karpathy’s autoresearch framework but adds a “creative director” binary to inject randomness and avoid the “Blank Page Problem” of static LLMs, the code notes explain. The director summarizes current code, fetches a random ML paper abstract, and generates a new experiment directive, creating a constantly evolving attack model.

Armadin plans to pilot the platform with early‑stage enterprise customers later this quarter, SiliconANGLE reports. The company aims to replace manual red‑team exercises with continuous, AI‑driven testing that can adapt to emerging threats in real time.

Industry watchers note the infusion of Arm capital signals a broader push to embed AI deeper into hardware‑level security solutions. Arm’s involvement also gives Armadin access to the chipmaker’s ecosystem, potentially enabling tighter integration of the simulation engine with next‑generation processors.