The $3 Billion Bet on AI Hardware Engineering

A four-year-old startup that most people outside the defense world have never heard of just tripled its valuation to $3 billion in eight months. Its wager is that AI hardware engineering, not chatbots, is where the next fortune gets built. The largest weapons makers on the planet are already paying to find out whether it is right.

By Blumefield | November 10, 2026

The valuation that tripled while nobody was watching

On Monday, an Austin company called Nominal said it had closed a $240 million Series C round that values the business at $3 billion. The financing was led by Lightspeed Venture Partners and co-led by Insight Partners, with a supporting cast that reads like a crossover between Silicon Valley and the defense-industrial base. Sequoia, Lux Capital, Founders Fund, General Catalyst and Salesforce Ventures all took part, and so, tellingly, did Lockheed Martin Ventures. You can read the company's own account of the raise on the Nominal blog, and the full investor list in the official press release.

What makes the number arresting is the speed behind it. Back in March, Nominal raised $155 million at a $1 billion valuation. Eight months later that figure has tripled. For a company that sells software to the people who build jet engines and nuclear reactors rather than ad-tech or another consumer app, that kind of climb is rare. It is also a clear signal that investors are now hunting for AI exposure in parts of the economy that sit a long way from the language-model arms race.

Nominal's proposition is easy to state and hard to pull off. It wants to become the system of record for AI hardware engineering, the single place where the messy flood of test data from physical machines gets gathered, organized and, more and more, interpreted by software agents. The company has watched demand for that idea move from a niche pitch to a stampede, and this round is among the clearest pieces of evidence yet that the capital is following the thesis. Blumefield has tracked how the physical-AI story has traveled from the research lab to the balance sheet, and this deal is a loud new data point.

Why AI hardware engineering is suddenly worth so much

For decades, the engineers who design aircraft, submarines and launch vehicles have been drowning in their own data. A single test campaign can throw off millions of readings from sensors, benches and flight recorders, scattered across legacy tools that were never built to talk to one another. The hard-won knowledge about what a given anomaly actually means tends to live inside the heads of a handful of veteran engineers, which makes it fragile and almost impossible to scale.

AI hardware engineering is the attempt to fix exactly that problem. Nominal describes itself as a "Hardware Intelligence Platform" that connects raw machine data with the engineering context around it, so that both people and AI systems can reason about a complex physical product without starting from scratch every time. Rather than build another general chatbot, the company is trying to give AI a reliable foundation to contribute inside high-consequence work, where a wrong answer is not an embarrassment but a safety failure.

The customer list is where the thesis stops being abstract. Nominal says more than 80 commercial and government organizations now use its platform, including five of the seven largest defense contractors in the world. Anduril, the drone and autonomy maker, has more than 300 engineers working on the system, and reports that analysis which once took five or six hours now happens close to real time. The shipbuilder HII ran a pilot that cut certain analysis tasks from hours to minutes and reduced some production test steps by roughly half. Hermeus, which is chasing hypersonic flight, said it compressed the gap between uploading test data and reviewing it by 80 percent while designing and building its Quarterhorse Mk 1 aircraft in seven months.

The agents moving into the test lab

The frontier of AI hardware engineering is not dashboards but autonomy. Nominal's first AI agent, called Nominal Analyst, is built to investigate engineering data on its own, run the analyses an engineer would run, and return answers that can be traced back to the underlying evidence. The company is careful to say that engineers keep control of the actual decisions, reviews and approvals, a distinction that matters a great deal when the machine in question is a weapon or a reactor rather than a shopping cart.

Behind the headline agent sits a deliberately open architecture. A product line the company calls Nominal for Developers exposes the platform through SQL, standard APIs and the Model Context Protocol, so customers can wire the data into their own applications and build custom agents on top of it. The company says further agents are already in development for requirements verification, planning, scheduling and fleet health, each one aimed at a different slice of a machine's life from drawing board to retirement.

Nominal is also buying its way into new territory. It has acquired Fid Labs, whose team is now developing the company's agents, and Pull Systems, which extends the platform into automotive. Crucially for its defense customers, the software can run not only in the cloud but on-premises, in air-gapped networks and on edge devices, which is the difference between a tidy demo and something a classified program can actually deploy.

The defense tailwind nobody is hiding

It is no accident that Lockheed Martin Ventures sits on the cap table. The entire pitch for AI hardware engineering is riding a wave of government money aimed at rebuilding the industrial base. In October the US Air Force selected Nominal for a $71.2 million Strategic Funding Increase tied to its Collaborative Combat Aircraft program, the effort to field cheaper autonomous jets that fly alongside crewed fighters. That award, which requires matching funds from outside the Air Force, puts the company squarely inside one of the Pentagon's priority programs.

The founders fit the moment. Chief executive Cameron McCord spent eight years in the Navy before working on test software at Anduril. His co-founder Jason Hoch came out of Palantir, and the third co-founder, Bryce Strauss, was an engineer at Lockheed Martin. They started the business in 2022 after running into the same broken tooling at defense primes and startups alike, a backstory MIT News chronicled last year, and they now run a company headquartered in Austin with offices in Los Angeles, New York, Seattle, Washington and London. "Hardware engineering is getting its AI moment," McCord said of the raise, framing the next five years as a period when the firms that put AI inside their engineering workflows will pull away from those that do not.

That language lines up neatly with the broader political mood around reshoring and reindustrialization. Washington wants more things built in America, built faster, and built with fewer people, and AI hardware engineering is being sold as the connective tissue that makes all three possible at once.

What the $3 billion is really betting on

For all the momentum, the valuation deserves a hard look. Three billion dollars is a bold mark for a company that has disclosed no revenue figures, in a software niche where incumbents and open-source tools are not standing still. It arrives, too, during a stretch when public markets have grown nervous about anything with the letters A and I attached, as capital expenditure forecasts balloon and investors question when the returns actually land. A private round led by sophisticated backers is a vote of confidence, but it is not the same as a profit.

There is a deeper question underneath the funding math. The promise of AI hardware engineering rests on the idea that autonomous agents can be trusted to analyze data for systems where failure costs lives, and that trust has to be earned result by result, not asserted in a press release. Nominal's own insistence that humans stay in the loop is both reassuring and a quiet admission of how far the technology still has to travel before it earns real independence.

Even the skeptics, though, would concede that the underlying pattern is real. The money, the customers and the government contracts are all pointing the same direction, toward a world where the unglamorous work of testing physical machines becomes one of the most valuable places to deploy artificial intelligence. Whether Nominal specifically grows into its $3 billion price tag is an open question. That AI hardware engineering has arrived as a category worth fighting over is not. For more coverage of how the physical-AI boom is reshaping industry, follow Blumefield.