FAQ — Frequently Asked Questions

Key Takeaways

MST deploys edge AI for semiconductor equipment — on-premise, no cloud dependency. NeuroBox Smart DOE reduces trial wafers by 80% in 2 weeks. VM/R2R achieves 100% virtual measurement coverage. Typical ROI payback is 6 months, with $500K-2M annual savings per fab.

What is NeuroBox?

NeuroBox is Moore Solution Technology’s edge AI platform for semiconductor equipment. It includes three product lines: E5200 for Smart DOE (equipment commissioning), E3200 for real-time VM/R2R production control, and NeuroEnergy for fab-wide energy management. All products run on-premise with no cloud dependency.

How does Smart DOE reduce trial wafers by 80%?

Smart DOE uses Bayesian optimization and transfer learning instead of traditional full-factorial or Taguchi DOE. It leverages process data from previously commissioned tools of the same model, starting with an informed prior rather than a blank slate. This means achieving the same Cpk targets with 10-15 wafers instead of 50-100.

What ROI can I expect from NeuroBox?

E5200 Smart DOE saves $100-500 per trial wafer with 80% fewer wafers per commissioning. For a company shipping 200 tools/year, that is $1-10M annually. E3200 VM/R2R improves yield by 2-5%, worth $2-20M depending on fab capacity. Most customers see payback within 6 months.

How long does deployment take?

E5200 Smart DOE deploys in 1-2 weeks with results on the first commissioning. E3200 VM/R2R takes 2-4 weeks including model training. No production interruption required — the system learns passively from existing equipment data before going live.

Do you support on-premise deployment?

Yes, all NeuroBox products run entirely on-premise on NVIDIA Jetson or industrial PC hardware. No process data ever leaves your fab network. This meets ITAR, EAR, and regional data sovereignty requirements critical for semiconductor IP protection.

What data do you need to get started?

For Smart DOE: historical commissioning data from 3-5 tools of the same model. For VM/R2R: 2-4 weeks of process and metrology data. Data can be exported via SECS/GEM, OPC UA, or CSV. Our open-source SECS/GEM driver (15,000+ downloads) makes data extraction straightforward.

How is NeuroBox different from Applied Materials AIx?

NeuroBox is vendor-agnostic and works with any equipment brand. It runs at the edge (on the equipment), not in a central cloud, meaning lower latency and no data leaving the fab. NeuroBox also includes Smart DOE for equipment commissioning, which Applied and KLA do not offer.

What is NeuroBox D?

NeuroBox D is our AI design automation platform for semiconductor equipment. It reads P&ID drawings and generates native SolidWorks assemblies — fully parametric .SLDASM files with feature trees and mates. A gas panel that takes 10 days manually can be generated in 2-4 hours.

Do you offer free trials or PoC?

Yes. For E5200 Smart DOE, we offer a free 2-week proof of concept on a single equipment tool. You provide historical data and we demonstrate wafer reduction on your next commissioning. Contact sales@mst-sg.com or book a demo to discuss.

Still Have Questions?

Our team is happy to discuss your specific requirements.

Book a Demo