AI cybersecurity startup Corma completed a $125 million Series A funding round at a $1 billion valuation, with NEA leading the round and existing backers Sequoia Capital, Khosla Ventures, and Coatue returning to participate. The company has raised $185 million total since launch.

Corma exited stealth in August with AI-native security tools designed to defend against AI-enabled attacks. The timing proved fortuitous: on October 21, critical vulnerabilities surfaced in PHP and Node.js—widely deployed infrastructure components—while LiteLLM, a popular proxy for routing requests across multiple large language model providers, disclosed a separate incident. The clustering of incidents within hours underscored a tactical reality: AI amplifies both attack surface and attack velocity, creating acute demand for defensive solutions.

The company positions itself around frontier AI research rather than off-the-shelf defenses. That differentiation matters in a crowded field. Kevin Mandia's Armadin, Anthropic, OpenAI, and Microsoft all have skin in AI security. But Corma's framing as a research-first player—not just a wrapper around existing models—targets a specific buyer psychology: enterprises willing to pay premium multiples for proprietary moats.

The venture economics here are straightforward. The AI security market is nascent, compressed multiples are still available, and winners typically consolidate within three to five years. NEA and Sequoia's participation signals confidence in Corma's ability to build either a defensible research capability or a sales engine capable of moving from early adopters into mid-market deals. The $1 billion valuation on $185 million raised—an 18-month journey from stealth—reflects investor conviction that the problem will compound faster than solutions can.

What remains to be tested: whether Corma can sustain the gross margins (typically 70-plus percent for AI-native security software) necessary to justify the valuation, and whether its research advantage proves durable against better-capitalized competitors from cloud vendors or legacy security incumbents.