Startups

Corma Raises $125 Million Series A at $1 Billion Valuation Three Months After Stealth Exit

The Israeli AI security firm, founded in 2025, has now raised $185 million in total as investors bet that purpose-built defense models can counter the advantage attackers hold.

T
By TechQuire Daily Staff TechQuire Daily Staff
November 17, 2026 / 7 min read

Corma, an Israeli cybersecurity startup that applies artificial intelligence to defense, said on November 16, 2026 that it raised a $125 million Series A round at a $1 billion valuation. The round was led by NEA (New Enterprise Associates), with existing investors Sequoia Capital, Khosla Ventures and Coatue participating. The announcement came roughly three months after Corma emerged from stealth and introduced its product, and it brings the company's total funding to $185 million.

The speed of the jump is unusual even by the standards of the current AI investment cycle. Corma announced a $60 million seed round in August 2026, according to Calcalist's CTech. The new capital takes it from a young research outfit to a billion-dollar company in about three months, a trajectory that says as much about investor anxiety over cybersecurity as it does about the startup's own technology.

Corma was founded in 2025 by CEO Alon Pluda, a graduate of the IDF's Psagot program and a Unit 8200 alumnus. Globes reported on November 16, 2026 that Pluda started the company after leaving Miggo Security, which he had co-founded, following a dispute with his co-founders. He served in Unit 8200 until 2022. The company now operates in Tel Aviv and San Francisco, and its staff includes researchers from Google and DeepMind as well as cybersecurity specialists from Unit 8200.

Corma's pitch is that general purpose AI models are helping attackers faster than defenders. The company builds AI models and agents trained specifically for cybersecurity defense, delegating security tasks to what it calls an autonomous AI workforce. It works with large enterprises in healthcare, financial services, energy and critical infrastructure, retail and telecommunications, and says demand has surged among Fortune 100, Fortune 500 and Fortune Global 500 companies since it left stealth.

Key Facts

TechCrunch reported on November 16, 2026 that the Series A was led by NEA and valued Corma at $1 billion, three months after the company introduced its product. The outlet put total funding at $185 million, with backing from Sequoia Capital, Khosla Ventures and Coatue. That total is the sum of the new $125 million round and the $60 million seed round announced in August 2026.

Calcalist's CTech reported on November 16, 2026 that NEA led the round with existing investors Sequoia Capital, Khosla Ventures and Coatue participating. The Israeli publication said Corma had expanded server capacity 24-fold since emerging from stealth to meet demand. It also said the company employs about 25 people in Tel Aviv and San Francisco.

Globes reported on November 16, 2026 that the company's annual recurring revenue is estimated in the tens of millions of dollars, a notable figure for a company that only recently launched its product. The outlet put headcount at 20 employees across Tel Aviv and San Francisco. The discrepancy with the roughly 25 cited by Calcalist's CTech is small, but it reflects how quickly the company is changing during its first months as a public-facing business.

Corma says it develops its language models from scratch rather than relying on Anthropic, OpenAI or DeepSeek models, according to Globes. Pluda told the outlet that general purpose models are strong at software development and at identifying bugs and vulnerabilities, which is why attackers use them so efficiently, but weak at defense, which requires finding a needle in a haystack in enterprise telemetry, logs and events. He added that Corma does not sell CISOs a dashboard but AI agents that let them do their jobs orders of magnitude better.

The funding wave around Corma is not isolated. TechCrunch also reported on November 16, 2026 that Kevin Mandia's startup Armadin raised $255.5 million at a $2.5 billion valuation, and it noted that Anthropic, OpenAI, Microsoft and many other startups are applying AI to defense. Corma argues that it differentiates itself by being a frontier AI research lab focused entirely on protection rather than a broader AI company with a security product attached.

Analysis

The most striking number in the round is not the $1 billion valuation but the $185 million total raised before Corma has had time to build a long public track record. For a company that has been publicly operating for only a few months, that is a remarkable amount of capital to absorb. Corma's ability to compress the timeline reflects a specific market condition: the fear that offensive AI is improving faster than defensive AI is strong enough that buyers and investors are willing to back a credible countermeasure.

What this really means is that investors are not simply funding another security tool. They are funding a thesis that purpose-built models can outperform general purpose models on defense, where the work involves sifting through enormous volumes of telemetry, logs and events for weak signals. Pluda's argument is that general models are optimized for building software and finding bugs, which helps attackers, but are poorly suited to the patient, high-volume detection work that defense demands. If that argument holds, Corma's models could become a platform rather than a feature.

The October 21 incidents gave that thesis a timely narrative. TechCrunch reported on November 16, 2026 that a critical flaw in PHP, a flaw in Node.js and a security incident at LiteLLM, a proxy for routing requests to multiple LLM providers, helped drive interest in Corma. Calcalist's CTech described the broader wave as Black October, a run of critical vulnerabilities and breaches across widely used software that prompted responses from security teams and warnings from governments including Washington and Tokyo. For a startup selling defensive AI, a month like that is both a warning and a marketing event.

The bigger picture here is that cybersecurity is becoming a capital-intensive AI race, not just a software market. Building foundation models from scratch is expensive, and Corma's 24-fold server capacity expansion shows where some of the new money will go. The company's headcount of 20 to 25 people is tiny relative to its valuation, which means the bet is on research talent and compute efficiency rather than on a large sales force. That can work if the models deliver, but it also leaves little margin for a product that fails to generalize across enterprise environments.

Why It Matters

Corma's raise matters because it tests whether specialized AI models can command durable pricing power in security. A new vendor has to show that its agents reduce work, catch incidents that others miss or shorten response times. Corma's claim that it sells agents rather than a dashboard is a direct challenge to the way security teams have traditionally bought and used tools. If the company can prove that its agents outperform existing workflows, the pricing question becomes much easier to answer.

It also matters for Israel's startup ecosystem. Corma is a young company with roots in Unit 8200, and a team that includes researchers from Google and DeepMind. The combination of Unit 8200 experience and frontier AI research is a recognizable Israeli founder story, but the scale of this round is not. A $1 billion valuation after three months gives founders and investors a new reference point for how quickly a security company can be financed when the timing and the narrative line up.

For buyers, the practical question is whether Corma's models can be trusted with the most sensitive data in an enterprise. The company says it works with healthcare, financial services, energy, critical infrastructure, retail and telecommunications customers, sectors where false positives and missed detections carry real costs. The 24-fold server expansion suggests demand, but it does not prove efficacy. The next year of customer deployments will matter more than the valuation itself.

Next Up

Corma's immediate task is to turn its early enterprise interest into durable revenue. Globes estimated annual recurring revenue in the tens of millions of dollars, a useful start but a small base against a $1 billion valuation. The company will need to show that its AI agents can be deployed across large environments without requiring extensive customization for each customer. It will also need to justify building models from scratch as larger competitors push into the same market.

The competitive field is already crowded. Armadin's $255.5 million raise at a $2.5 billion valuation, plus the presence of Anthropic, OpenAI, Microsoft and a long list of startups, means Corma cannot rely on novelty alone. The company's next public milestones will likely be customer case studies, model performance claims and hiring in Tel Aviv and San Francisco. If those arrive quickly, the $1 billion valuation will look like a starting point. If they slip, the three-month jump from stealth to unicorn will be remembered as a very expensive sprint.

Tagged

Comments (0)

No comments yet. Be the first to share your thoughts.