Skip to content

Denso

Who they are / moulding stake. Denso is the world's second-largest automotive parts supplier [Synthesis], and its moulded-parts footprint is enormous in the small-and-precise register: connectors, sensor bodies, ECU and HVAC housings, fans. A Google Patents assignee search for Denso mentioning "injection molding" returned ~2,369 results (retrieved 2026-08-01 during vetting — treat the count as an order-of-magnitude indicator, not an audited figure). [Evidence — query capture] 1

Engineering stack: automation culture, not AI. Denso standardized on Siemens NX + Simcenter 3D with CAE templates that cut analysis time by up to 80% — a Siemens-published case study. That number is template automation and CAD/CAE integration, not machine learning; it does show a culture that invests in engineering-workflow efficiency. [Evidence] 2

Where Denso's public AI actually lives. Autonomous-driving perception: the Denso Pittsburgh Innovation Lab presented AV scene-generation research with Carnegie Mellon University5 at CVPR 2026 (PR of May 26, 2026), plus three Denso IT Laboratory papers. Materials/AV-centric — nothing touches moulding. [Evidence] 3

The targeted Japanese-language check (upgrade path from vetting): searched the Denso Technical Review for injection-moulding + AI content. Outcome: found a Vol. 28 (2023) DTR paper on injection moulding of high-thermal- conductivity PPS — in-mold visualization with high-speed imaging, i.e. serious experimental moulding science, with no ML component. No moulding-AI DTR paper surfaced in EN or JP searches. This sharpens the verdict: Denso publishes deep moulding research and simply has not put AI into it publicly. [Evidence — bounded search] 4

Steelman (the strongest case for them). If any Tier-1 could build moulding AI internally, it is Denso: a giant part library, a template-automation culture, real AI research labs, and researchers who instrument moulds. The ingredients coexist; the public record shows they have not been combined. [Synthesis]

What would change our mind: a DTR or conference paper applying ML to moulding/defect/warpage data, or a Denso Wave / Denso Robotics product doing learned moulding-process control. Either upgrades Denso to a full report. [Inference]

References

  1. Google Patents assignee search, Denso + "injection molding" (~2,369 results, retrieved 2026-08-01; count is a search-index figure) — https://patents.google.com/?q=%22injection+molding%22&assignee=denso ↩

  2. Siemens case study — Denso: NX + Simcenter 3D integrated process and CAE templates, "reduced the time spent for CAE analysis by up to 80 percent" — https://resources.sw.siemens.com/en-US/case-study-denso-corporation/ ↩

  3. "DENSO, Carnegie Mellon University to Present Leading AV Research at IEEE's CVPR 2026," Denso PR, May 26 2026 — https://www.denso.com/us-ca/en/news/newsroom/2026/20260526-01/ ↩

  4. DENSO Technical Review Vol. 28 (2023) — in-mold visualization of high-thermal-conductivity PPS injection moulding (Kurita, Yoshimura, Suzuki, Yokoi); experimental, no ML — https://www.denso.com/jp/ja/-/media/global/business/innovation/review/28/paper-11.pdf?la=ja-jp&rev=386cf93b3ac54315a7e72e1d8f636dad&hash=106770B985F371CB550470903A3497F1 ↩

  5. Carnegie Mellon University — Pittsburgh, USA — https://www.cmu.edu ↩