Beat 01
Process IP is the crown jewel
A recipe leak is worse than an outage. Every architectural decision starts from keeping recipes, panel designs and defect imagery inside the tenant.
Security
Recipes, panel designs and defect images are the most valuable data in the fab. Isolation, encryption, governance and an air-gapped option are the baseline, not a tier.
Reviewed by IT/OT and IP teams before a single tool is connected
Threat model
Beat 01
A recipe leak is worse than an outage. Every architectural decision starts from keeping recipes, panel designs and defect imagery inside the tenant.
Beat 02
OT segmentation is real and non-negotiable. The fab edge runtime is designed to work with no outbound path at all.
Beat 03
Write-back to tools and robotic handling is treated as a safety-adjacent surface with fail-safe stops and explicit authorisation.
Beat 04
Customer audits, field returns and reliability investigations all require reconstructing decisions years later. The audit log is designed for that, not for debugging.
# closed loop, one panel
perceive substrate → tft → emitter stack → sub-pixel
plan deposition · anneal · transfer · test/bin
act recipe write-back (SECS-GEM)
sense mura ΔL* · sub-pixel opens · particles
optimise yield · ramp · scrap
log immutable audit entry ✓ signed
retrain engineer correction → model
Controls
Each layer assumes the one above it may fail.
Step 1 of 6 · identity
SSO from your directory, RBAC mapped to fab roles, least privilege by default.
Step 2 of 6 · isolation
Per-tenant data separation for recipes, designs, imagery and models.
Step 3 of 6 · encryption
In transit and at rest, with key management aligned to your policy.
Step 4 of 6 · network
OT segmentation respected; on-prem and air-gapped deployment supported.
Step 5 of 6 · control
Authorised, idempotent write-back with fail-safe stop on every path.
Step 6 of 6 · audit
Immutable, signed decision records with model version and approver.
Controls
Your recipes, panel designs, defect images and models never mix with another tenant's.
Standard, enforced, and documented for your review.
Identity federated from your directory with fab-role-aware permissions.
Yield- and quality-grade records that cannot be silently rewritten.
Full operation with no outbound connectivity.
Model improvements travel across the fleet; your data does not.
Air-gapped
In air-gapped mode the control plane is local, models are delivered through your own media process, and telemetry stays inside the fab.
# deployment profile
mode air-gapped
outbound none
control-plane local
model-updates operator-delivered, signed bundles
telemetry retained on-site
audit-export on-demand, signed
Safety
Any path that can move a tool or a panel can be stopped. Degradation is always toward less autonomy, never toward acting on a weak signal.
# closed loop, one panel
perceive substrate → tft → emitter stack → sub-pixel
plan deposition · anneal · transfer · test/bin
act recipe write-back (SECS-GEM)
sense mura ΔL* · sub-pixel opens · particles
optimise yield · ramp · scrap
log immutable audit entry ✓ signed
retrain engineer correction → model
Data handling
If a data class is not needed for a decision or an audit, it is not collected.
| Data class | Location | Purpose |
|---|---|---|
| Recipes and setpoints | Your tenant / on-prem | Control and reconstruction |
| Panel designs | Your tenant / on-prem | Perception context |
| Defect and inspection imagery | Your tenant / on-prem | Classification and training |
| Metrology and telemetry | Your tenant / on-prem | Drift attribution |
| Decision audit records | Your tenant / on-prem | Quality and customer audits |
| Model weights and versions | Tenant-scoped | Rollout, rollback, reconstruction |
Security FAQ
Through three gates. Shadow mode observes and predicts with no write-back; advisory mode recommends recipe, transfer and binning moves that a process-integration or yield engineer approves; graduated autonomy releases low-risk deposition, anneal and transfer control once measured accuracy and twin validation clear the bar. High-impact decisions stay human-in-the-loop.
Per-tenant isolation with recipes, panel designs and defect images scoped to your tenant, encryption in transit and at rest, SSO/RBAC, an immutable yield/quality-grade audit log and an on-prem or air-gapped option. Fleet learning shares model improvements, never your recipes.
The fab edge runtime keeps running without the cloud, agent steps are idempotent, and every control path has a fail-safe stop that returns tools and robotic handling to a known-good state. Degradation drops autonomy back to advisory rather than acting on a weak signal.
Voices from the line
“We do not lose panels because nobody is watching. We lose them because the signal that mattered was buried under a thousand nuisance calls.”
“Ramp is the whole game. If a new product takes two quarters of manual tuning, that is two quarters of margin we never get back.”
“Transfer yield is our ceiling on microLED. Every dead emitter is a repair cycle or a scrapped backplane.”
Quotes are illustrative composites drawn from discovery interviews with process-integration, yield and quality engineers. Named references are [PLACEHOLDER] pending design-partner consent.
The economics
Emiteon is priced and evaluated on the numbers a fab already reports. These are design-partner targets for the first twelve months of deployment.
Figures marked as targets are design-partner objectives, not audited results. Company operating status, customers and outcomes are [ASPIRATIONAL] until independently verified.
FAQ
Through three gates. Shadow mode observes and predicts with no write-back; advisory mode recommends recipe, transfer and binning moves that a process-integration or yield engineer approves; graduated autonomy releases low-risk deposition, anneal and transfer control once measured accuracy and twin validation clear the bar. High-impact decisions stay human-in-the-loop.
Reducing false calls is the wedge, not a side effect. The mura-and-defect agent is trained on true-versus-false-call labels from your own inspection history, so it separates genuine large-area non-uniformity, sub-pixel opens and shorts, particles and stains from nuisance signals, and every call is traceable to the evidence behind it.
Per-tenant isolation with recipes, panel designs and defect images scoped to your tenant, encryption in transit and at rest, SSO/RBAC, an immutable yield/quality-grade audit log and an on-prem or air-gapped option. Fleet learning shares model improvements, never your recipes.
Deposition, photolithography, encapsulation and anneal tools, Mura/AOI inspection and metrology, array and cell test, mass-transfer and repair stations, robotic handling and MES — vendor-neutral via SECS-GEM/HSMS where the tool supports it, with REST and OPC-UA bridges elsewhere.
Every pixel, perfectly uniform.
We would rather answer it before the pilot than during it.