Skip to content

Corma’s $125M Series A highlights a cybersecurity AI divide—defenders are behind, and scal

·StartupReader editorial desk· 3 min readReviewed by our editors

Corma.ai has become the latest AI-powered cybersecurity unicorn, raising $125 million in a Series A round that values the company at $1 billion. NEA led the investment, joined by returning backers Khosla Ventures, Coatue, and Sequoia Capital. The timing is significant, as the company’s research exposes a growing asymmetry in AI’s role in cybersecurity: while large language models are highly effective at launching attacks, they perform poorly at detecting and mitigating them.

Corma’s own tests illustrate the problem starkly. When the company challenged leading LLMs to breach a simulated corporate network, they succeeded in 85% of attempts. Yet when asked to identify those same malicious files, the models detected only 12%. This disparity, Corma argues, gives attackers a clear advantage. The startup’s proprietary models, however, outperform competitors like GPT-6 Astra and Claude Opus 5.5 on DefenseBench V5, a benchmark that measures both accuracy and speed in defensive tasks.

What sets Corma apart is its focus on automation—not just detection, but response. The company’s algorithms analyze data from firewalls, network monitoring tools, and other security applications, then generate tailored remediation steps. Unlike other AI systems, which struggle with false positives, incomplete data, or fragmented logs, Corma claims its models can operate autonomously, escalating to human teams only when necessary. A Slack chatbot interface also allows administrators to interact directly with the system, such as verifying whether a malware removal effort was successful.

The company’s rapid growth has created operational hurdles. While demand from major enterprises has surged since its stealth exit, scaling infrastructure to meet that demand has been a primary challenge. The new funding will accelerate expansion, including improvements to its AI models—a critical move as the gap between offensive and defensive AI capabilities continues to grow.

This round reflects a broader shift: investors are increasingly backing AI-driven cybersecurity, but not all approaches are equally effective. When we first covered Corma’s emergence, its swift transition from stealth to unicorn status signaled how urgently the industry needs better defensive tools. Now, with competitors like Noeri (which raised $110 million for self-improving AI systems) and traditional security vendors integrating AI into their offerings, the race is on to prove that automation can level the playing field—or at least narrow the advantage attackers currently hold.

The question now is whether Corma can maintain its benchmark performance as it scales. The company’s claim that its models can accelerate breach response by a factor of 100 is ambitious, but the real test will be delivering that speed in real-world enterprise environments. As cyberattacks grow more sophisticated—and AI-enabled—the stakes couldn’t be higher. If Corma succeeds, it could redefine defensive cybersecurity. If it fails to scale, the gap between attackers and defenders will only widen.

For now, the $1 billion valuation suggests investors believe the company can bridge that divide. Whether it can do so in practice remains to be seen.

Sources: siliconangle.com

“The funding round reveals a critical imbalance: AI tools excel at automating attacks, but defensive AI remains a work in progress. Corma’s benchmarks suggest progress, but whether it can deliver at scale will determine if the gap narrows—or widens further.”
— StartupReader
ShareLinkedInXWhatsApp

One of the funding rounds reported in the week of 16–22 November 2026. See every round that week.

Read the original reporting

The outlets below did the original reporting.

Related briefs

This brief was drafted automatically from the sources above and published under our editorial policy. Spotted an error? Tell us.