Israeli-founded cybersecurity startup Corma has raised US$125 million in Series A funding at a US$1 billion valuation, just three months after emerging from stealth with US$60 million in seed financing. Led by New Enterprise Associates (NEA), the round includes continued participation from Sequoia Capital, Khosla Ventures and Coatue. Founded by Alon Pluda in 2025, Corma is developing specialised artificial intelligence foundation models designed for defensive cybersecurity. Its ambition is to build an autonomous AI workforce capable of investigating security threats, analysing enterprise systems and executing complex defensive tasks alongside human security teams. The new financing brings the company’s total disclosed funding to US$185 million and makes it one of the fastest-funded cybersecurity AI startups to reach unicorn status.

A 5M Series A Takes Corma to a B Valuation in Just Three Months
Corma announced its US$125 million Series A on November 16, 2026, with New Enterprise Associates leading the investment. Existing investors Sequoia Capital, Khosla Ventures and Coatue also participated, continuing their backing of the company’s cybersecurity research. The transaction values Corma at US$1 billion, bringing its total disclosed funding to US$185 million. The latest investment follows a US$60 million seed round announced in August 2026, when the company first emerged from stealth. Founded in 2025, Corma has attracted an unusually large amount of capital for such a young cybersecurity business. The company plans to use the new funding to accelerate the development of its foundation models, significantly expand computing capacity and selectively recruit additional specialists. Its valuation reflects investor confidence in the possibility that cybersecurity will become one of the most important applications for advanced AI, particularly as attackers gain access to increasingly capable automated tools.
AI Is Increasing the Power of Cyberattacks Faster Than Traditional Defences Can Adapt
The rapid development of generative AI is changing the economics of cybersecurity. AI coding systems can help analyse software, identify weaknesses and automate technical tasks that previously required specialist knowledge and considerable time. These capabilities can benefit legitimate developers and security researchers, but they can also lower barriers for malicious actors. Defensive cybersecurity faces a different challenge. Security teams must continuously monitor enormous volumes of audit logs, network traffic, identity activity, cloud configurations and security events. Most of this information is routine, but a small number of signals may indicate a serious threat. Finding those signals requires connecting evidence across different systems and understanding how seemingly unrelated events fit together. This is why improving AI at writing code does not automatically make it equally effective at protecting large organisations. Corma’s central argument is that defensive security requires models trained specifically for the complexity, consistency and long-running investigative work involved in protecting enterprise environments.
Corma Is Building Its Own Foundation Models Specifically for Cyber Defence
Rather than relying entirely on general-purpose language models, Corma says it is training proprietary foundation models designed specifically for defensive cybersecurity. Its research focuses on the type of information security teams work with every day, including audit logs, network flows, system configurations, identity information and security events. These datasets are different from the code and text used to train many general-purpose models. A defensive model must understand how information from different security products connects, recognise important signals across long periods and maintain consistency while completing multi-step investigations. The company believes specialised training can make AI more effective at this kind of work than simply adapting a general chatbot. Its ambition is to continue improving these models towards what it describes as superintelligence for defensive cybersecurity. That remains a long-term research objective, however, rather than a verified description of the system’s present capabilities.

Corma Workforce Turns Specialised Models Into an Autonomous Security Team
The company’s main product is Corma Workforce, an AI-powered security workforce designed to operate alongside an organisation’s existing cybersecurity employees. Instead of functioning as another dashboard that presents alerts for analysts to investigate, the system is designed to take responsibility for defined security tasks from beginning to completion. Its agents can work across an organisation’s existing security tools, investigate evidence, connect relevant information, coordinate required steps and report the outcome. The platform is designed to support functions including detection engineering, threat intelligence, digital forensics, identity and access management, cloud security and attack-surface management. This approach could help organisations complete more investigations without increasing their security headcount at the same rate as their infrastructure expands. Corma’s product materials also describe collaboration through workplace tools such as Slack and Microsoft Teams, allowing security teams to assign tasks, examine findings and approve actions when required. The goal is not simply to produce AI-generated recommendations, but to complete operational security work within the customer’s existing environment.
Integration With Existing Security Infrastructure Is Central to Its Strategy
Large organisations already operate multiple security products covering identity management, cloud infrastructure, endpoint protection, firewalls and security information and event management systems. Replacing these products would be expensive and difficult, particularly in industries where security operations are closely connected to regulatory requirements and business continuity. Corma is therefore building its workforce to work across existing enterprise systems rather than requiring customers to move their entire security infrastructure to a new platform. Its agents can collect information from different tools, investigate connections between events and coordinate work across the systems already used by security professionals. This creates a different commercial position from a traditional cybersecurity vendor that sells one product for one category of protection. Corma wants to become an intelligence and execution layer sitting across the wider security environment. Its potential value will depend on how reliably it works with different enterprise technologies, respects access permissions and maintains clear oversight for sensitive actions.
Fortune 500 Demand Is Driving a Major Expansion in Computing Capacity
Since emerging from stealth, Corma says it has attracted demand from Fortune 100, Fortune 500 and Fortune Global 500 organisations. The company has identified healthcare, financial services, energy, critical infrastructure, retail and telecommunications as important enterprise markets. According to Calcalist, the company expanded its server capacity 24-fold to accommodate rising demand. This highlights an important characteristic of Corma’s business: training and operating specialised foundation models requires substantial computing infrastructure. Its ability to deliver enterprise security tasks depends not only on model quality, but also on the processing capacity required to analyse large quantities of customer information and complete complex workflows. Corma has not publicly named all of its customers or disclosed audited revenue figures, so the reported enterprise traction should be understood as company and media-reported information rather than independently verified commercial performance. Nevertheless, the reported demand helps explain why the company is prioritising compute expansion alongside model research and engineering recruitment.
Alon Pluda Is Building a Cybersecurity Company Around the Defensive AI Gap
Corma was founded in 2025 by Alon Pluda, who previously co-founded cybersecurity company Miggo Security. Its team combines experience in foundation-model research, software engineering and hands-on cybersecurity. Calcalist reported that the company employs approximately 25 people across Tel Aviv and San Francisco, including specialists with backgrounds at Google, DeepMind and Israel’s technology sector. This combination is important because building defensive cybersecurity AI requires more than developing a powerful language model. Researchers must understand how security products operate, what information is available during an investigation and which actions are appropriate within an enterprise environment. Pluda’s central thesis is that cybersecurity requires a different type of intelligence from the capabilities that have driven the general-purpose AI race. The company is attempting to turn that thesis into a dedicated product category: an AI workforce that works with existing security teams rather than simply providing another tool for them to monitor.
Corma’s Biggest Test Is Making Autonomous Cyber Defence Reliable at Enterprise Scale
The US$125 million investment gives Corma significant resources to pursue its vision, but defensive cybersecurity remains a demanding environment in which mistakes can have serious consequences. An AI agent must distinguish genuine threats from routine activity, avoid disrupting legitimate business operations and respect the permissions assigned to it. It must also explain its findings clearly enough for security professionals to understand what happened and why an action was taken. Even when automation can complete a task independently, organisations need appropriate controls, accountability and escalation procedures for sensitive decisions. Corma must demonstrate that its specialised models can deliver reliable results across different customer environments, not only in selected demonstrations. It also faces competition from established security companies and other AI-native cybersecurity startups developing increasingly autonomous tools. Its larger opportunity is significant: if it can make AI-based defensive operations dependable at scale, it could change how organisations investigate threats and allocate scarce security expertise. The Series A financing is an important step towards that ambition, but the company’s claim to be building superintelligence for cyber defence will ultimately be judged by measurable security outcomes.