Nominal raises $240M at a $3B valuation to bring AI agents to hardware testing

Testing is where a machine's design meets physics, and the software behind it has not kept up. Lightspeed describes propulsion engineers copying data off test stands onto USB drives and into folders of Python scripts only one person understands[3]. Nominal's argument is that a general-purpose AI model cannot fix that by itself, because the data is scattered across instruments, files, simulations and teams, and a sensor reading means little without the hardware configuration and past experiments around it[2].

The company describes its platform in four layers: connections to the physical hardware being tested, a data layer that gives engineering data a common structure, an intelligence layer that lets models and agents use that context, and applications built on top[2]. It runs in the cloud, on premises, in air-gapped facilities and on edge devices, and engineers can build their own tools and agents on it through SQL, APIs and MCP[1].

Nominal Analyst, the company's first AI agent, investigates engineering data and runs complex analyses, with its answers traceable to the underlying evidence, while engineers keep control of decisions and approvals[1]. Lightspeed says two more are on the way: Execute, which runs a test, controls instruments and can abort if it detects the hardware going wrong, and Plan, which helps decide which test points will give the most information when every test uses up a prototype[3].

Nominal says more than 80 commercial and government organisations use it, including five of the seven largest defence contractors[1]. At Anduril more than 300 engineers use it, and analysis that took five to six hours now happens in near real time. An HII pilot cut some production test steps by about half, and Hermeus cut the time between uploading and reviewing data by 80% while building its Quarterhorse Mk 1 aircraft in seven months[1].

The US Air Force recently selected Nominal for a $71.2 million Strategic Funding Increase to support Collaborative Combat Aircraft work[1]. CEO and co-founder Cameron McCord is a former US Navy submarine officer and Anduril alum[4].

Much of the attention in AI goes to models and chips. Nominal is a bet that the bigger gains in physical industries come from wiring AI into the data and workflows engineers already depend on. The new funding goes to agents and applications across a machine's life, from requirements verification and planning to scheduling and fleet health, and to new industries and regions, including through acquisitions such as Fid Labs and Pull Systems this year[1]. With robots, drones and autonomous vehicles all needing constant testing and updates, that tooling layer is starting to look like a market of its own.