[ The brand mark of a company whose customers literally cannot afford a bad test day. ]
"Accelerate test. Accelerate progress."
The software stack that lets hardware teams validate fighter jets, nuclear reactors, satellites and robots as fast as software teams ship a pull request.
Somewhere this week, a team of engineers will roll a hypersonic vehicle out for a test. They will get one shot. The hardware costs a fortune, the runway is booked, and the people watching represent a customer who does not grade on a curve. When the vehicle comes back, terabytes of telemetry, video, and sensor data come back with it. The old way to make sense of that haul was a weekend of stitching CSVs together and praying the timestamps lined up. The new way is Nominal.
Nominal is a five-year-old software company in Los Angeles that sells something deceptively unglamorous: the ability to look at a test and actually understand what happened. Its customers build the things that cannot fail quietly - fighter jets, nuclear reactors, satellites, rockets, autonomous robots. The company's name is the engineer's word for "everything checks out." It is also, by 2026, worth a billion dollars.
For two decades software ate the world by learning to iterate. Push, test, observe, fix, repeat - thousands of times a day, with dashboards that tell you instantly when something breaks. Hardware never got that gift. A team building a jet engine or a reactor still runs a test, then disappears into a swamp of disconnected files, custom scripts, and tribal knowledge before anyone can say whether the run was good.
The cost of that gap is not abstract. Every slow analysis is a delayed flight, a missed window, a program that slips. In defense and aerospace, where the whole point is learning faster than the threat, a swamp of CSVs is a national-security problem wearing a lab coat. The founders had lived this from three different angles, and they were tired of pretending it was fine.
Nominal was founded in 2022 by three people who had each watched the same problem from a different trench. Their bet was contrarian for the venture world of the time: that the next great software companies would be built for people who work with atoms, not just bits - and that the unglamorous plumbing of hardware test data was a billion-dollar category hiding in plain sight.
Investors found the pitch easy to follow. General Catalyst, Lux Capital, and Founders Fund backed the company out of stealth in 2024. The cap table reads like a defense-tech yearbook - which is convenient, since the customers do too.
Nominal sells the idea that test data should behave like a living system, not a graveyard of files. It ships in two parts that fit together.
The all-in-one data platform for hardware engineering. It captures and monitors telemetry, logs, video, and simulation results, then puts trend analysis and automated reporting in one collaborative workspace - so the whole team sees the same reality.
The edge compute platform - billed as the first Python framework for hardware-in-the-loop test applications. It controls hardware, sequences tests, and processes data at the edge, with the performance and graphics high-precision work demands.
Set up a test stand in hours, not weeks. Watch results in real time instead of next Tuesday. Trust that every run - good or bad - actually taught you something. That is the entire product in one breath.
It is a modest-sounding promise. It is also the kind of promise that, when kept, makes a customer move every program they have onto your software. Which is roughly what happened.
Skepticism is the correct posture for any startup claiming to be essential. So here is the part that is hard to wave away. In the span of roughly ten months, Nominal landed four of the five largest defense contractors in the world, grew revenue sevenfold, and more than tripled its team. More than sixty organizations now trust it with their most sensitive programs.
The customer roster is its own argument: Anduril, Shield AI, the U.S. Air Force, Hermeus, REGENT, General Atomics, Varda Space, Muon Space, Radiant Nuclear. These are not companies that adopt software for the demo. They adopt it because a test went better.
There is a quieter signal underneath the logos. Selling into defense and aerospace is famously slow - procurement cycles measured in fiscal years, security reviews that outlast most startups. A company does not land four of the top five largest contractors in under a year by being merely interesting. It does it by removing a pain those buyers feel every single day, then earning the right to be trusted with the data they guard most carefully. That Nominal cleared that bar this fast is the part worth sitting with.
It would be easy to file Nominal under "developer tools, industrial edition" and move on. The founders see it differently, and the difference matters. Their stated mission is to equip engineers to test and operate critical technology at scale - which in their telling is how a country keeps its technological edge. Investors call it a "dual-use love story." The company calls it building for engineers, by engineers.
Strip away the venture poetry and a simple claim remains: if the people building hard, dangerous, important machines can learn faster from every test, the machines get better and safer sooner. That is the whole bet. It happens to also be a very good business.
The category Nominal is staking out has had occupants before - MATLAB, LabVIEW, and a thousand in-house tools cobbled together by overworked test engineers. What it has not had is a modern, collaborative, cloud-and-edge platform built by people who lived the problem and refused to accept that hardware teams should be second-class citizens of the software era. Whether Nominal becomes the default depends on execution, not on whether the gap is real. The gap is real. Every test engineer who has lost a weekend to mismatched timestamps already knows it.
Return to the hangar. The hypersonic vehicle is back on the ground, still warm, and the terabytes are landing. In the old story, that is where the hard part started - the weekend of stitching files, the meeting next week where someone finally says whether it worked. In Nominal's version, the team is already watching the run resolve in real time, comparing it against simulation, flagging the one channel that drifted, and deciding what to change before the coffee gets cold.
That is the change Nominal is selling, and so far the people building the most demanding machines on earth are buying. The threat keeps learning. Nominal's pitch is that its customers can learn faster - and that, more than any valuation, is the point.