Lithoric

Is the photonic known-good-die problem real? You tell us.

We’re testing one hypothesis: that fab-calibrated models, plus packaging and assembly physics, plus the end-link criteria that decide scrap vs ship (BER, FEC, eye opening, power budget, economics), can flag bad co-packaged optical dies early enough to matter — with customer refinements kept isolated so foundry and design IP is not pooled.

Fully anonymous · no name, email, or tracking required · ~2 minutes
1.Which is closest to your role?
2.“Discovering a bad photonic die only after co-packaging is a serious cost problem in CPO.”
Strongly disagreeStrongly agree
3.“Wafer tests alone miss a meaningful share of post-package / end-link failures; you also need assembly and application physics.”
Strongly disagreeStrongly agree
4.What share of final-module failures do you believe are already visible from wafer-level data alone?
30%
None are visible from wafer aloneNearly all are
5.“My organization would trial a software tool that flags likely-bad dies before packaging — against our module and link criteria.”
Strongly disagreeStrongly agree
6.What level of data isolation would your organization require to trial a tool like this (process / design IP stays yours)?
7.What single result would convince you this works — or that it can’t?
8.Strongest reason this would not work — or anything we’re missing?
9.Optional — an email, only if you’re open to a short follow-up. Leave blank to stay fully anonymous.
Responses are used only to validate or reject this hypothesis. No analytics or cookies run on this page.

Thank you — that genuinely moves the needle.

Your answer goes straight into the evidence we weigh this bet against.

← Back to Lithoric