Cybersecurity AI Startup Corma Reaches $1B Valuation in 3 Months

Abstract illustration of AI-driven cybersecurity defense

Last updated: November 17, 2026, 07:00 IST.

Cybersecurity AI startup Corma has raised a $125 million Series A led by NEA at a $1 billion valuation, about three months after launching its product. The company says it has now raised $185 million in total since coming out of stealth in August 2026.

The round was reported by TechCrunch. Corma describes itself as a frontier AI research lab focused on protection, offering AI models and agents trained specifically for cybersecurity defense rather than general-purpose use.

Beyond NEA as lead investor, the round also drew participation from Sequoia Capital, Khosla Ventures and Coatue, a group of investors with deep histories in both enterprise software and AI infrastructure bets.

Abstract illustration of AI-driven cybersecurity defense

Why investors moved so fast

Reaching a $1 billion valuation within three months of a product launch is an unusually fast climb, even by the standards of the current AI funding environment. TechCrunch’s reporting links the timing directly to a cluster of disclosed vulnerabilities that heightened demand for AI-driven defense tools.

A critical PHP flaw was disclosed on October 21, 2026, the same day a Node.js flaw and a security incident involving LiteLLM were also disclosed. That concentration of disclosures in a short window appears to have driven immediate interest in Corma’s product among enterprises looking to shore up defenses quickly.

What Corma actually offers

Corma’s pitch is narrower than a general-purpose AI assistant adapted for security work. The company positions its models and agents as purpose-built for cybersecurity defense, trained specifically to identify and respond to the kind of vulnerabilities that have been surfacing at a rapid pace across widely used software components this year.

That focus places Corma in a growing category of security-specific AI startups, distinct from AI labs building broad assistants that are later adapted for security use cases. Describing itself as a “frontier AI research lab” rather than a conventional security vendor signals that Corma is positioning its own model development, not just its application layer, as the product’s core value.

Coming out of stealth in August 2026 and reaching a $1 billion valuation roughly three months later also means Corma has spent almost its entire public existence in active fundraising, rather than the multi-year gap between early traction and a unicorn valuation that was typical of enterprise software startups before the current AI funding cycle.

Who else is in the category

TechCrunch’s report also names Armadin, a startup from Mandiant founder Kevin Mandia, as another AI-driven security company operating in the same broad space, though the report did not include funding or valuation figures for that company.

Corma’s valuation jump, reached without an extended fundraising cycle, suggests investors are willing to move quickly when a security-focused AI product arrives at a moment when high-profile vulnerabilities are already dominating enterprise security budgets. Mandia built his earlier company, Mandiant, into one of the best-known incident-response brands in the industry before its sale to Google, so Armadin’s presence in the same conversation signals that experienced security operators, not just AI-native founders, see an opening in AI-driven defense.

What to watch next

Corma has not disclosed its customer base, revenue, or specific product roadmap beyond its general positioning as a defense-focused AI lab. Whether the $1 billion valuation reflects proven enterprise adoption or anticipation of future demand is likely to become clearer as the company discloses more about its customers.

The $185 million raised in total since August also puts Corma well ahead of the typical pace for a Series A stage company, reinforcing how concentrated investor attention has become around AI tools built specifically to respond to the software vulnerabilities disclosed this year, rather than general-purpose AI products retrofitted for security.

Source: TechCrunch.