Corma, a three-month-old AI lab building foundation models for defensive cybersecurity, has raised $125 million in a Series A round that values it at $1 billion. NEA led the round, with Sequoia Capital, Khosla Ventures and Coatue joining, taking total funding to $185 million.
A unicorn in three months
The round comes just three months after Corma emerged from stealth in August with $60 million in seed funding. The pace is striking even by the standards of the current AI funding boom: a 25-person company, founded in 2025, now carries a ten-figure valuation.
Corma, based in Tel Aviv and San Francisco, trains foundation models built specifically for cyber defense and deploys them as an autonomous workforce. The system works alongside human security teams, using their existing tools to investigate threats and carry out defensive work from start to finish.
The claims behind the valuation
The company says its workforce already completes 99 percent of security tasks autonomously and successfully, at 100 times the speed and 95 percent lower cost than human teams. Those figures are Corma’s own claims, not independently verified results, and no customer has been named.
Corma says demand from the world’s largest organizations has been overwhelming, and its biggest constraint is securing enough compute to keep up. Its blog points to October’s wave of critical vulnerabilities, which it calls Black October, as proof that defensive capability is falling behind offensive AI.
What the money is for
The new funding will go toward scaling Corma’s foundation models and expanding compute capacity, with selective hiring alongside it. Pluda says the next generation of the models will be shared soon, though no date has been given and no independent benchmark has been published.
Our Read
The speed of this round says more about the market than about Corma’s models. Investors are paying unicorn prices for defensive AI before any independent test exists, betting the best defensive models will own enterprise security budgets. That bet is plausible, but still a bet.
Corma is not the only AI lab to reach a ten-figure valuation on a big early round. Sand.ai did the same this month with a $200 million round for video AI, and Upwind raised $400 million to secure AI cloud systems. Corma’s round is smaller, but just as aggressive.
The valuation also reflects scarcity. Very few teams claim to train foundation models purely for defense, and the investors backing Corma — NEA, Sequoia, Khosla and Coatue — are the same names behind the biggest AI-lab rounds of the past two years. Scarcity, though, is not evidence.
Why It Matters
If defensive models work as Corma claims, the economics of security change: a small team could hold off AI-powered attacks that currently overwhelm human analysts. If they do not, $185 million has bought a fast-growing lab with unproven claims. Either way, defense is now a frontier-lab race.
Background
Corma was founded in 2025 by Alon Pluda, a graduate of Israel’s Psagot program and an alumnus of the military intelligence Unit 8200. The company employs about 25 people across Tel Aviv and San Francisco, and says the next generation of its foundation models will be shared soon.
Source: Corma’s announcement and Business Wire release


0 Comments