Nominal sells software to engineers who test physical machines such as drones, aircraft, rockets, vehicles and reactors. A single test can produce thousands of sensor channels, camera footage, operator logs and configuration files, each saved by a different tool. Engineers often spend more time finding and lining up that evidence than deciding what it means.

Core is where the data lands. Teams stream or upload telemetry, meaning time-stamped readings like pressure, temperature and voltage, along with logs, video, audio, specifications and hardware configurations. Core puts all of it on one clock, so a sensor spike lines up with the video frame and operator command from the same moment.

Engineers then overlay runs to compare them, build shared workbooks with plots and derived signals, and watch thousands of channels live during a test. Each conclusion stays attached to the underlying data, so the next engineer can pick up where the last one stopped.

Connect runs on the computer next to the test stand. Engineers write Python apps that command motors, valves and power supplies, read sensors through prebuilt or custom drivers, and run repeatable test sequences from operator control panels. Data captured without a network syncs to Core once the connection returns.

Customers can run the platform in Nominal's cloud, which uses U.S. regions operated by U.S. persons for export-controlled work. They can also run it in their own virtual private cloud, on their own servers, or at a site with no network at all.

Analyst is Nominal's first in-platform AI agent. It answers engineers' questions about data in their Nominal workspace, builds editable workbooks, overlays tests, runs frequency analysis, and reads the specs and test plans stored in Nominal. It can turn a written requirement into a pass/fail check, citing the source document and showing its calculation for an engineer to review.

Here is a hypothetical example. A battery team cycles a new pack in a thermal chamber, and one run trips a temperature alarm. Analyst pulls the thermal requirement from the test plan, overlays cell temperatures and coolant flow across every run, and flags that the failing run exceeded the allowed cell-to-cell spread shortly after coolant flow dropped. The engineer then checks sensor placement and decides whether to retest. Analyst does not certify that the pack is safe.

SDKs for Python, MATLAB, LabVIEW, Rust and Go, along with SQL, APIs and an MCP server, let existing scripts and outside agents work with Nominal data. As a result, Nominal can sit alongside a team's existing tools instead of replacing them.

Nominal sells enterprise software subscriptions. Under its commercial terms, each customer signs an order form that lists the software, subscription term, fees and negotiated limits on seats and capacity. Fees are committed for the full term and are non-refundable. Customers who exceed their limits can be required to buy more seats, more capacity or a higher tier at then-current rates.

Professional services are sold alongside the subscription to handle deployment, configuration and adoption. For on-premises customers, Nominal can buy servers on the customer's behalf at prices set in the order form, while the manufacturer keeps warranty responsibility. The terms also let Nominal sell new modules, deployment models or premium capabilities under separate orders, which gives it a way to charge separately for future AI features instead of bundling them.

Nominal's main costs are the engineers who build the platform, storage and compute for high-rate sensor data, model inference for Analyst, security compliance, and support. Air-gapped and on-premises deployments can require additional implementation and support work. Customer-hosted installations shift hosting costs toward the customer, but do not eliminate Nominal's responsibility for supporting its software.

Profit depends on reuse. If an instrument driver or analysis template built for one customer serves the next, onboarding gets cheaper and each new subscription keeps more of its fees. If every deployment needs bespoke configuration on site, services eat into that benefit. Support for Python, MATLAB and LabVIEW lowers switching friction because teams can adopt Nominal without rewriting test code, and new programs inside existing accounts can expand seats and capacity.

Nominal's main rival is the way test data is already handled: NI's test software in labs and production lines, and tooling that customers' own engineers build. The core contest is whether hardware teams adopt a vendor-neutral system of record and AI layer across programs or extend tools they already own, while general data and AI platforms try to take over analysis from above.

NI's SystemLink is a mature direct comparison. It manages test stations remotely, tracks assets and calibration, takes in TestStand results, normalizes files into tables, runs scheduled Jupyter analyses, maps results to specifications, and holds data locally during network outages before forwarding it. Paired with LabVIEW and TestStand on NI hardware, it is deeply embedded in validation and production test.

Nominal differentiates with a single timeline that includes video, live streaming across instruments from mixed vendors, and agent-driven analysis. It also supports LabVIEW and NI data acquisition hardware, so many deployments start out coexisting with NI before competing over where the data and analysis live.

Many teams rely on Python scripts, MATLAB, Jupyter notebooks and internal data warehouses. Teams can customize these tools to fit exact workflows, combining open-source software with paid products such as MATLAB. In return, the team's own staff must maintain ingestion for every file format, clock alignment, access controls, export compliance and uptime.

Nominal sells relief from that maintenance work, and its SDKs keep existing scripts running. Teams with strong in-house data engineering may still see the subscription as a cost they can avoid by building the same capabilities themselves.

General-purpose data platforms give engineering teams another way to store and analyze their measurements. Nominal explicitly supports external agents such as Claude, Codex and Devin through its MCP server, a connector that gives those agents access to Nominal tools and data. Nominal's MCP server and SQL access turn these tools into partners that query Nominal data. The same openness means analysis could happen outside Nominal, reducing it to a storage layer, so Nominal is betting that its hardware data model, test-stand control and Analyst's grounding in engineering documents justify a separate platform.

Nominal's core buyers have been R&D and flight-test teams at defense and aerospace companies. Its expansion paths lead to different budget holders, including factory quality leaders, automaker field quality teams and energy organizations, who handle similar sensor data later in a machine's life.

Nominal positions Connect for use from prototype validation through production-line testing. Operators can run interfaces that validate circuit boards and servos, log results against part numbers, and pull work instructions from manufacturing execution systems such as Manufacturo. Production test budgets belong to manufacturing and quality organizations and recur with every unit built, while R&D test budgets often taper after a program qualifies.

The test logic that qualified a prototype could carry onto the line, giving traceability from early tests to individual serial numbers. If Nominal lands in programs as they move into production, it could sell subscriptions to manufacturing teams and expand contracted capacity. That path would put it in direct competition with NI's production footprint.

On November 6, 2026, Nominal acquired Pull Systems. Working with Porsche and other Volkswagen Group brands, Pull Systems built Workbench for field monitoring, quality and service teams managing hundreds of thousands of vehicles. Nominal says those teams used the telemetry platform to find root causes in minutes rather than months. These teams work near warranty and service budgets, which are separate from R&D test spending.

Integration is ongoing. One cofounder now works on Nominal's developer platform and the other on Analyst. Whether Workbench's automaker relationships turn into Nominal contracts is unresolved, but fleet-scale data volumes could support large capacity subscriptions if they do.

Nuclear developer Antares uses Nominal to stream test data live to partners across the country, and Nominal markets to both traditional and emerging energy companies. Today that work sits in developers' test programs. The next possible buyer is an operator running reactors, turbines or storage sites that needs the same monitoring once equipment is in service.

This is a qualified possibility, not a demonstrated market. Selling to operators would require utility procurement, regulatory record-keeping and plant-network security. Its ability to run without an internet connection provides a starting point, but test-program adoption does not establish that plant operators will select the same software.

Defense procurement concentration. Nominal's traction is anchored in defense primes and defense startups, so budget delays, program cancellations or shifting contracting rules could slow renewals across many accounts at once. Its export-control and compliance investments also tie up resources that commercial buyers may not pay for.

AI validation burden. Analyst writes pass/fail checks on safety-critical hardware, and one wrong check accepted by an engineer could undermine trust across a program. Customers may demand extensive validation before relying on agent outputs, which would slow adoption and raise support costs.

Incumbent and DIY friction. NI's installed base and mature internal Python stacks mean Nominal often enters alongside existing tools and must earn displacement program by program. Onsite deployment and long enterprise sales cycles could keep customer acquisition and services costs high even as the valuation assumes rapid expansion.