AI · 20 November 2026 · 4 min read
Corma Raises $125 Million Series A at $1 Billion Valuation to Scale Defensive Cybersecurity Foundation Models
Backed by NEA, Sequoia Capital, Khosla Ventures, and Coatue, the ex-DeepMind and Unit 8200 team is deploying autonomous defensive agents directly into Fortune 500 security stacks.

Lede & Nut Graf
Cybersecurity artificial intelligence (AI) startup Corma has raised a $125 million Series A at a $1 billion valuation just three months after its seed round, bringing total capital raised to $185 million (The AI Insider, Nov. 20, 2026). The round was led by New Enterprise Associates (NEA) with participation from Sequoia Capital, Khosla Ventures, and Coatue.
The rapid markup to unicorn status reflects urgent enterprise demand for automated threat response. As offensive cyber campaigns increasingly leverage AI for vulnerability discovery and lateral movement, Fortune 500 security operations centers (SOCs) are turning to domain-specific defensive foundation models to close the response-time gap.
Technical & Financial Breakdown
Staffed by former Google DeepMind researchers and Israeli Unit 8200 intelligence veterans, Corma develops frontier foundation models engineered specifically for autonomous cyber defense (The AI Insider, Nov. 20, 2026). Rather than forcing enterprises to rip out legacy monitoring tools, Corma’s autonomous agents integrate directly into existing Fortune 500 security stacks.
Demand accelerated sharply following October 2026’s wave of enterprise breaches, prompting Corma to expand its underlying server capacity 24x to handle surging production inference loads.
Second-Order Implications
Corma’s 24x infrastructure ramp illustrates how cybersecurity is emerging as a high-urgency, non-discretionary driver of enterprise AI inference demand. Unlike productivity copilots—where return on investment (ROI) can take quarters to prove—autonomous breach containment directly offsets multi-million-dollar downtime and ransomware liabilities.
For legacy cybersecurity incumbents, the arrival of $185 million-funded, foundation-model-native challengers raises pressure to either build proprietary agentic defense models or acquire specialized startups before Fortune 500 budgets shift.
Caveats & Open Questions
Corma did not disclose its annual recurring revenue (ARR), exact Fortune 500 customer count, false-positive containment rates, or the hardware architecture behind its 24x server expansion ``. Granting autonomous agents write-level containment authority inside live Fortune 500 networks also carries high operational risk if a model misclassifies legitimate production traffic during an incident.
More in AI