The stock works if Fabrinet keeps converting AI-optics demand into
qualified output faster than customers can replace it. Existing telecom,
datacom,
DCI and
HPC ramps already prove demand is real, and new factory space plus slower customer
qualification cycles support a continued premium versus generic manufacturers. But Fabrinet still captures value mainly through services, not end-product IP, so I expect solid compounding with limited multiple expansion rather than an explosive
rerating.