AI-generated illustration. Source: TechCrunch
SAN FRANCISCO, November 16, 2026 — Corma, a start-up that trains artificial intelligence models for defensive cybersecurity, has raised US$125 million in Series A funding at a US$1 billion valuation, just three months after leaving stealth, the company said on Monday.
The round was led by NEA, with existing investors Sequoia Capital, Khosla Ventures and Coatue also taking part, bringing total funding to US$185 million, according to a company blog post signed by chief executive Alon Pluda and to coverage by TechCrunch. In August, Corma had emerged from stealth with a US$60 million seed round. A Series A is typically a company’s first large growth round after its initial seed funding. A valuation of US$1 billion, often called “unicorn” status, is unusually fast for a firm that only introduced its product in the summer.
Corma describes itself as a frontier AI research lab focused on protection, training foundation models — large general-purpose systems that can then be specialised — for defensive security work and deploying them as an “AI workforce” that operates alongside a customer’s existing security team and tools. In plain terms, the company is selling software agents that can take on tasks such as hunting for threats or helping with incident response, rather than only answering questions about them. TechCrunch noted that it is far from alone: established labs including Anthropic, OpenAI and Microsoft, as well as other start-ups, are also building AI for cyber defence.
Pluda wrote that when Corma launched it warned that AI’s offensive uses were advancing faster than its defensive ones, and that this “played out in public” in October. TechCrunch reported that interest in the product rose after a cluster of incidents on October 21, including a critical flaw in PHP, a widely used web programming language, another in Node.js, a popular JavaScript system for running server software, and a security incident at LiteLLM, a tool that routes requests to multiple AI model providers. Corma refers to that period as “Black October”. Those software flaws and the company’s account of how customers used its product should be treated as separate facts: the incidents were real and widely reported, but claims about how much work Corma’s systems carried out for clients come from the company and its investors.
In a statement issued with the funding news, NEA partner Aaron Jacobson said: “Corma is building the intelligence layer that defensive security needs.” He added that, as AI gives attackers an advantage, “Corma’s combination of frontier AI research and deep cybersecurity expertise has the potential to help organisations defend against these threats at scale.” A Business Wire release from the company also said it is headquartered in Tel Aviv and San Francisco, that it had increased server capacity 24 times to meet demand, and that adoption had surged among large enterprises. Those operational figures have not been independently audited in the reporting reviewed for this article.
Pluda said the new money will be used to scale the foundation models, increase computing capacity and “selectively expand” the team. He also said demand has been “overwhelming” and that securing enough computing power remains one of the firm’s biggest challenges, with some customers still on a waitlist.
What to watch next is whether Corma publishes independently verified results from its enterprise deployments, how quickly it can add the computing capacity it says it needs, and whether a crowded field of defensive AI tools can grow into the kind of durable business a US$1 billion valuation implies.
Source : Corma blog post, November 16, 2026; TechCrunch; Business Wire.

