Three months ago Corma raised a $60 million seed round led by Sequoia. Now it has a $125 million Series A at a $1 billion valuation, led by NEA, with Sequoia Capital, Khosla Ventures and Coatue all coming back in. That brings total funding to $185 million in the space of one season.
Corma, based in Tel Aviv and San Francisco, calls itself a frontier AI lab for defense. It trains its own foundation models for defensive security work and deploys them as an “AI workforce” that runs inside a company’s existing security tools, next to the human team. The release doesn’t break down how the money will be spent beyond scaling the models and that workforce.
What the models do
The pitch is breadth. Corma says its workforce handles many defensive jobs, not one narrow task. That covers reading audit logs, security events and network flows, and tying together weak signals spread across systems and long stretches of time. That last part is where human analysts tend to lose the thread, because the clues sit in different tools weeks apart.
Corma also takes a swipe at the general models. Coding-focused AI, it argues, isn’t tuned for defensive work, so a lab trained specifically on security data should do better. It’s a reasonable bet, and the same one a number of AI security start-ups are making right now. The proof will be in detection quality, which the release doesn’t put numbers on.
Demand claims
The company says demand is growing across the Fortune 100, Fortune 500 and Fortune Global 500, and that enterprises are handing more of their security work to its models. It increased server capacity 24x to keep up. There’s no customer count and no named customers, so those claims stay Corma’s own for now.
The team mixes researchers from Google DeepMind with security people from Israel’s Unit 8200, covering frontier AI research and both offensive and defensive security. Alon Pluda is co-founder and CEO. He framed cybersecurity as a contest of intelligence against intelligence. Aaron Jacobson, the NEA partner on the deal, said Corma is “building the intelligence layer that defensive security needs.”
A unicorn valuation at Series A, three months after a seed, says more about investor appetite for AI-native security than about anything Corma has shown publicly yet. With the offensive side of AI moving fast, defenders are getting funded to keep pace. Whether a dedicated defense model beats the big general labs at this is the open question for 2027.