Corma raises $125 million at a $1 billion valuation to scale AI cyber defense
NEA led the financing. Corma says demand is outpacing its computing capacity as it builds AI models that carry out security work inside customers’ existing tools.
Loading page…
NEA led the financing. Corma says demand is outpacing its computing capacity as it builds AI models that carry out security work inside customers’ existing tools.
Listen to this story
The financing gives Corma more capital to develop security-specific AI and address a reported bottleneck: insufficient compute to serve demand. Its Corma Workforce product is designed to carry out end-to-end security assignments using customers’ existing tools, with approvals when needed. NEA says the system is already in production at large enterprises and has contained active attacks, claims attributed to the investor rather than independently verified; Corma says it is expanding capacity for organizations on its waitlist.
The November 16, 2026, Series A brings Corma’s total disclosed funding to $185 million, including $60 million announced when it emerged from stealth in August.
Corma says October’s software vulnerabilities and breaches drove a surge in demand from enterprises already using its technology.
NEA says deployments span Fortune 500 and other large enterprises, and that Corma has autonomously detected, investigated and contained attacks; these claims are not independently assessed in the article.
Three months after emerging from stealth, Corma has secured another $125 million to build AI systems for cybersecurity defense. The startup’s November 16, 2026, Series A values it at $1 billion. The new money will fund larger models, more computing capacity and selective hiring as Corma says demand is stretching its resources.
NEA led the round, with Sequoia Capital, Khosla Ventures and Coatue participating again. The financing brings Corma’s total funding to $185 million, following the $60 million it disclosed when it emerged from stealth in August.
Corma introduced itself with a warning: AI’s ability to help attackers was advancing much faster than its ability to help defenders. That remains the company’s argument for building models specifically for defensive security, rather than relying only on general-purpose AI.
In its funding announcement, Corma pointed to October’s wave of vulnerabilities and breaches across widely used software as evidence of that imbalance. It said some large enterprises already had its technology working alongside their security teams when those incidents hit, and demand subsequently surged.
The proposed response is a product called Corma Workforce. Powered by the company’s foundation models—the underlying AI systems driving its agents—it uses customers’ existing security tools and carries out complex assignments from beginning to end, Corma says.
NEA’s investment thesis draws a distinction between finding a software weakness and defending an organization. In its investment post, the firm argues that general models learn heavily from code and human language. Defenders must also interpret logs, events and system settings—information NEA says those models see relatively little of during training.
According to NEA, Corma trains directly on defensive-security data and uses large-scale reinforcement learning in realistic enterprise environments. That approach gives the model practice doing security work against a clear, verifiable outcome, rather than learning about the field only through text.
The investor also describes training designed to pick out faint signals from large volumes of noise and maintain consistency across thousands of decisions in one investigation. Those are important parts of its technical pitch: the task is not simply to produce a plausible answer, but to sustain a chain of security decisions.
NEA describes the deployed product as fitting into a team’s existing routines:
The firm says Corma has multiple production deployments at Fortune 500 and other large enterprises across healthcare, financial services, energy, critical infrastructure and retail. It also says the system has autonomously detected, investigated and contained active attacks. These are claims from the investor leading the financing, not an independent assessment.
Candidly, one of the biggest challenges we are continuously facing is to secure enough compute capacity to keep up with that demand.
Corma, in its November 16 funding announcement
Corma’s spending plan addresses both model development and the infrastructure needed to serve customers. The company says it will accelerate the scaling of its foundation models, increase computing capacity and selectively expand its team. Its announcement also addresses organizations still on the waitlist, saying it is working to expand capacity as quickly as possible.
The next product milestone remains ahead: Corma says it will share the next generation of its foundation models “soon.” For organizations awaiting access, its immediate commitment is to add the capacity needed to bring more customers onto the service.
Loading discussion...
Join the conversation
Explain where speed should outweigh human control—or where it should not.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.