@bubbleboi May I ask
1. Would inferencing be slow with CXL based memory Vs HBM. Would real time coding still be possible
2. Isn't $MRVL 's CXL is better than alab
3. HBF has write endurance issue, you think it replaces HBM mainly due to cost ?.
@BluthCapital I would argue this is the wrong example. This use case forces price sensitive consumers from going to camel camel camel & increases engagement and adoption.
@aleabitoreddit 1. CPO might be too early for 2028 & there is no revenue pipeline (yields, adoption etc.)
2. Lite lasers are better than Sive
Sive can still double, but there are risks