Corma’s product has been on the market since August. On Nov 16 the company said it had raised $125 million at a valuation of $1 billion, in a Series A round led by NEA.
For most of the software industry, a Series A is the round that funds the search for a business model. It is usually raised on a few million dollars of revenue, a handful of design partners and a slide about a market that does not exist yet. Corma skipped the part where the market is in doubt. Its founders raised a growth round at the price of a company several stages further along.
What the company sells is a way to handle the queue. Enterprise security teams run systems that produce alerts, and the volume of those alerts has outgrown the number of people available to read them. A typical security operations center triages thousands of signals a week, most of them benign, and the analyst’s job is to find the small number that are not. That work is repetitive, it runs around the clock, and it burns through staff.
Corma’s pitch is that a model can do the first pass and much of the second: classify the alert, gather the context that determines whether it matters, and take the routine action, which for most alerts is to close them with a record of why. Customers pay an annual subscription rather than a per-seat license, which is the right structure for a product that removes seats rather than adding them.
The appeal of that pitch to an investor is easy to see. Security budgets are among the most durable lines in enterprise software, because they are set by incidents rather than by sentiment, and the labor shortage in security operations is structural rather than cyclical. A product that converts a staffing problem into a software line has a customer who can justify the purchase without a new budget cycle.
The appeal to a buyer is narrower than it sounds. Automation that closes alerts is automation that occasionally closes the wrong one, and the failure mode is a missed intrusion with an audit trail showing that a system decided the alert was benign. Vendors in this category spend much of their sales cycle on that objection, and the ones that win usually do it by keeping a human approval step in the loop for anything that would have escalated.
Which makes the pricing worth examining. A $1 billion valuation on a three-month-old product with no disclosed revenue and no disclosed customer count is a statement about trajectory rather than about financials. At ten times forward revenue, a common reference point for high-growth software businesses, the mark implies roughly $100 million of annual revenue within a few years. At twenty times, half that. Neither figure is impossible in security. Both are far ahead of anything the company has publicly shown.
NEA’s willingness to lead at that price is the strongest evidence available about what the company’s pipeline looks like. Venture firms at that scale do diligence on customer calls and contract values, and they do not usually price a round this size on a deck. The diligence would have covered the same three questions any buyer asks: how much of the triage can be automated without an analyst, how quickly a customer’s environment can be connected, and what happens at renewal when the novelty has worn off.
The last question is the one that decides the category. Annual subscriptions are contracted revenue, and contracted revenue is easy to book in the first year of a land-and-expand motion. The test arrives twelve months later, when a customer compares the invoice against the incidents that were caught and the analysts who were reassigned. Security software that automates a task has to survive the quarter in which nothing goes wrong, because that is when the buyer asks what they are paying for.
Competition comes from two directions. The platform vendors that already sit inside the customer’s endpoint and identity systems can bundle triage into a product the customer has bought, which turns a budget line into a feature. And the model providers themselves keep improving the underlying capability, which compresses the advantage of any company whose product is a wrapper around a model.
The defense against both is data. A system that has processed a customer’s alerts for two years knows that customer’s environment in a way a general model does not, and that knowledge is what makes the automation safe enough to leave running. It is also what makes the customer reluctant to leave. Whether Corma has that asset is a question only its customers can answer, and none of them have been named.
What the round does establish is that the market for AI-native security tools has not cooled. Nine-figure rounds at billion-dollar marks were routine in 2021 and became rare in 2023, when software multiples compressed and investors started asking about payback periods. Their return in the security category suggests buyers believe the labor math is real, and that the incumbents are slow enough to leave room.
The three months between product launch and a billion-dollar valuation is the number people will remember. It is also the number that says least about the company. What matters now is the renewal rate in the fourth quarter of next year, and the first customer who is willing to say publicly that the system closed the right alerts and opened the ones that mattered.


