NOUS Engineering services · Pilot proposal

NOUS

Plan. Retool.
Improve.

Digital-twin engineering for manufacturing decisions.

On site. One critical cell first.

Our illustrative factory model with two robot arms, a conveyor and a returning mobile carrier
Our browser prototype · synthetic equipment and signals

Sources & assumptions

Three industry applications, with attributed outcomes. Our browser demos use synthetic data; proposed capabilities and prices require validation.

Downtime / Siemens industry benchmarks

Siemens' 2024 report gives $2.3M per idle hour at a large automotive plant and a 113% rise in automotive hourly cost over 2019–2023. Across surveyed sectors, recovery rose from 49 to 81 minutes; the roughly 65% increase is calculated from those rounded figures. These are production-recovery times, not a separate diagnosis-time measure. Fewer incidents and fewer hours lost do not mean faster recovery per incident.

The report primarily concerns large manufacturers worldwide, not Ontario mid-market suppliers. Dollar values are retained as printed in the report; no CAD conversion is implied. A customer-specific loss rate must be measured. The supplied $30k lower bound and 76.9% increase were not retained because they do not match this source.

Siemens · The True Cost of Downtime 2024 · pp. 4, 10–11 ↗

Ontario / Founder market-sizing scenario

Ontario reports over 700 parts firms and over 500 tool, die and mould makers. Adding rounded categories gives a starting pool of 1,200 businesses, not a deduplicated list of qualified plants. Supplier/toolmaker overlap, multiple sites and buying eligibility require validation. The 350-plant serviceable segment is a founder estimate, not a verified industry count.

Assumed blended annual spend is CA$85,000 per account: 1,200 × CA$85,000 = CA$102M TAM; 350 × CA$85,000 = CA$29.75M SAM; 15–25 × CA$85,000 = CA$1.275–2.125M SOM (4.3–7.1% of the assumed SAM). SOM is a target annual revenue level during years 1–3, not cumulative revenue or contracted ARR. Annual account spend can include one-off audits, deployments and support; these totals are not wholly recurring revenue.

The proposed 30% retainer conversion is unvalidated and does not independently establish the CA$85k blended annual spend. Delivery capacity, purchase frequency, deployment uptake and retainer pricing must support that assumption. The supplied 35% CAGR has no specified source, geography or period and is omitted.

Ontario automotive sector · supplier ecosystem ↗

BMW / Planning

NVIDIA describes FactoryExplorer combining source data through OpenUSD and custom Omniverse Kit tools for measurement and collision review. Its 30% planning-saving figure is projected. BMW separately reports collision checks moving from almost four weeks of physical testing to three days of virtual simulation.

NVIDIA BMW FactoryExplorer case ↗BMW report · 11 June 2025 ↗

Schaeffler / Reconfiguration and commissioning

The NVIDIA case describes motor assembly, inverter-line and warehouse pilots using OpenUSD, simulation and robotic-task development. Schaeffler describes virtual commissioning to compare automation configurations. The task-development time example does not describe retooling an entire factory in half a day; no general retooling-saving percentage is claimed here.

NVIDIA Schaeffler case ↗Schaeffler virtual commissioning · July 2025 ↗

Wistron / Operational improvement

Wistron's 19 March 2024 account describes IoT and production data, process simulation, station dashboards and layout revision. It reports a 51% improvement in bottleneck production-line efficiency. That scope is retained here; the number is not plant-wide OEE, an independent audit or a forecast for this service.

Wistron customer account · March 2024 ↗

OpenUSD and NVIDIA / Architecture

OpenUSD composes assets and scene data. NVIDIA's developer paths support modular Omniverse libraries and custom Kit applications. Robot/process models, calibration, data integration and cost logic need additional engineering. A browser can display a streamed Kit application; the simulation runtime is separate from static hosting.

OpenUSD introduction ↗NVIDIA development paths ↗NVIDIA library architecture · April 2026 ↗Virtual-factory reference architecture ↗Omniverse browser delivery ↗

Our current implementation

The site is a Three.js browser prototype, not a connected Omniverse or PLC deployment. Geometry, machine events, camera/LiDAR-style views and rates are synthetic. Retooling uses custom inverse kinematics, sampled column checks and explicit timing/cost assumptions. It does not provide vendor-calibrated robot validation, continuous collision detection, measured OEE or automatic real-plant fault diagnosis.

The next milestone is one measured cell with approved geometry, robot/tool data, observed timing and agreed acceptance criteria. OpenUSD/NVIDIA and operational-data integrations remain proposed. Industry examples are not customers, endorsements or our outcomes.

Ontario / Eligibility-based support

AMIC has project, financing and stacking conditions. Approval and eligible costs must be checked per customer. No government payment plan or majority subsidy is assumed.

Ontario AMIC criteria ↗

Prospective channels

APMA and U of T Engineering are outreach targets. No affiliation, endorsement or active collaboration is claimed.

APMA ↗U of T Engineering ↗

Founder hypotheses

The forward-deployed service model, CA$35k audit delivered in 2–4 weeks, CA$85k–150k+ deployments and separately scoped retainers require customer and delivery validation. Partner-led scans and controls, reusable CAD/LiDAR-to-OpenUSD workflows and an asset library are proposed operating capabilities, not a claim that the current prototype already provides them.

No guaranteed savings, universal legacy-PLM price or projected margin improvement is claimed. The supplied five-day/83% diagnosis statistic, reinstall period and regional downtime totals remain unverified and omitted. TIAM / CMMA names require clarification. Model capacity and cost changes are illustrative, not measured customer savings.

Evidence reviewed 18 November 2026 · Design reference: Apple Design Skill ↗