Ribbix
String and interconnect placement and solder control.
Solar PV Manufacturing
Panevex
Autonomous operations layer for solar PV module manufacturing — closing the loop from perception to line control across stringing, EL defect sensing, lamination and binning.
Overview
What Panevex does.
Panevex closes the loop on solar PV module manufacturing: string and interconnect control, EL and defect sensing, lamination and encapsulation control, binning and power-class matching and the module digital twin.
Every line deployment produces labelled EL and process traces that train the next model, tightening the loop from cell to finished module.
| Detail | Value |
|---|---|
| Stage | Seed |
| Category | Physical AI |
| Year | 2026 |
| Website | panevex.com |
| Headquarters | Not disclosed |
| Raised | $1.5M |
| Investor | 10xC |
| Model | Per-line software subscription with a yield-outcome component. |
| Measured by | Module yield, EL false-call rate, bin accuracy, lines expanded per customer. |
Products
One loop, module by module.
Every deployment on panevex.com writes labelled traces back into the model, so the system generalises across plants, alloys, grades and geometries.
Elumnex
EL and electroluminescence defect sensing.
Lamira
Lamination and encapsulation process control.
Binova
Bin and power-class matching.
Twinvex
Module digital twin for ramp and yield planning.
Why we invested
The thesis behind Panevex.
Solar module manufacturing is scaling into a margin business where a single EL-detectable defect is caught after lamination — too late to fix cheaply. Power-class binning still leans on batch averages.
Panevex moves EL inspection and binning onto the line. The buyer is a module maker paying in hard currency; the wedge is one inspection or binning step, and module yield is the number that opens the next module.
Who is building it
Co-Founder & CEO
Co-Founder & CEOOwns the module-maker relationships, the pilot programme and the commercial model.
Co-Founder & CTO
Co-Founder & CTOOwns the vision models, the edge runtime and the write-back into string and laminate control.
Building something similar?
Building in Physical AI? We want to hear about it.