Hey this is what I love to hear! It took a lot of work and frankly it’s a complicated topic that isn’t explained well. This is exactly what I’m trying to do at this substack.
THANK YOU! This is what I’m trying to do. All the information is out there, I heavily relied on Wikipedia and semi wiki, but no one builds any topics so you get confused. Now whenever you hear advanced packaging you should know exactly what’s going on.
I’ve seen many of these images on different sites and sources before but the poor organization and how it’s scattered made it hard to build a mental model of the space that I could be confident in my understanding. Thanks for your hard work in compiling it all in one place. cheers!
This is exactly the point of it all. To build actual mental models so you know what it all means instead of things being repeated to you over and over w/o any comprehension.
Great stuff Mule. Being cynical for a moment, I would say this has been the thesis for the back-end guys for over a decade. I appreciate that we have the likes of TSMC, Intel etc. showing greater interest in advanced packaging. However, unlike the front-end equipment market, where what TSMC and Intel move the needle, what such customers spend on back-end is not representative of the overall market. A good way to perhaps think about this when Apple adopts a new technology, it typically has a materially impact on the TAM. However, in spite of Apple/TSMC first adopting advanced packaging in 2016, the TAM inflection is still debatable. This because a significant % of the back market are the sub-contracts such as ASE, Amkor, JCET, SPIL etc. and IDMs such as STMicro, Infineon, Osram, NXP etc. who have shown very limited appetite to move towards advanced packaging owing to its high cost. So how can we be confident that this time it is different?
Yes it has been - and frankly I don't think the back-end guys are going to benefit as well. The wafer level is kind of subsuming some of the job of the back end. This time it is more real as I can point to multiple leading edge chips that used advanced packaging plus moore's law is slowing really helps.
I think that STM/Infineon/Others will get there eventually but it will have to be because the demands of their end products have to rise significantly. AP is almost useless for a discrete product where there is no need for high bandwidth, but if we really are doing L3/L4 cars, the bandwidth and compute demands will start to demand AP.
I think it isn't different it just took a long time for it to happen. I get the cynicism but I really believe this is the only option forward for many companies, partially due to economics and partially just the the constraints of Moore's.
Brilliant primer, you made it so easy to absorb all that information, so well planned. Well worth the subscription. The internet is amazing!
Hey this is what I love to hear! It took a lot of work and frankly it’s a complicated topic that isn’t explained well. This is exactly what I’m trying to do at this substack.
The best article / primer on this topic anywhere
THANK YOU! This is what I’m trying to do. All the information is out there, I heavily relied on Wikipedia and semi wiki, but no one builds any topics so you get confused. Now whenever you hear advanced packaging you should know exactly what’s going on.
I’ve seen many of these images on different sites and sources before but the poor organization and how it’s scattered made it hard to build a mental model of the space that I could be confident in my understanding. Thanks for your hard work in compiling it all in one place. cheers!
This is exactly the point of it all. To build actual mental models so you know what it all means instead of things being repeated to you over and over w/o any comprehension.
Great stuff Mule. Being cynical for a moment, I would say this has been the thesis for the back-end guys for over a decade. I appreciate that we have the likes of TSMC, Intel etc. showing greater interest in advanced packaging. However, unlike the front-end equipment market, where what TSMC and Intel move the needle, what such customers spend on back-end is not representative of the overall market. A good way to perhaps think about this when Apple adopts a new technology, it typically has a materially impact on the TAM. However, in spite of Apple/TSMC first adopting advanced packaging in 2016, the TAM inflection is still debatable. This because a significant % of the back market are the sub-contracts such as ASE, Amkor, JCET, SPIL etc. and IDMs such as STMicro, Infineon, Osram, NXP etc. who have shown very limited appetite to move towards advanced packaging owing to its high cost. So how can we be confident that this time it is different?
Yes it has been - and frankly I don't think the back-end guys are going to benefit as well. The wafer level is kind of subsuming some of the job of the back end. This time it is more real as I can point to multiple leading edge chips that used advanced packaging plus moore's law is slowing really helps.
I think that STM/Infineon/Others will get there eventually but it will have to be because the demands of their end products have to rise significantly. AP is almost useless for a discrete product where there is no need for high bandwidth, but if we really are doing L3/L4 cars, the bandwidth and compute demands will start to demand AP.
I think it isn't different it just took a long time for it to happen. I get the cynicism but I really believe this is the only option forward for many companies, partially due to economics and partially just the the constraints of Moore's.
Agreed. Thanks for all the great content.