You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
Question: if you believe becoming a foundry is not possible for Intel, and for example GF had a meaningful foundry business and still became a zombie business, is your opinion that the only viable option is for Intel to keep the fabs but stop leading edge node development? Basically go fab-lite with TSMC for leading edge and then internal manufacturing for packaging and trailing node use cases like non-compute chiplets? Because no one will buy the fabs from Intel except some PE firm maybe with 10+ years of wafer contracts with Intel Products…
To be clear -- GF pivoted from doing advanced nodes (7nm) to being a "boutique foundry" (their term) where they specialized on unique fab/circuit types that could be build on derivatives of technologies like 45nm, 32nm, 28nm, 22nm .... and SOI. Basically GF ran out of money during their 7nm development ... and could not afford to develop it. (even though they already had a few good customers -- including AMD (and IBM, Qualcomm) -- lined up to use it)
Intel is in a tougher situation since their big potential customers don't completely trust them to deliver. And those potential customers don't want to bet their company on Intel delivering.
To be clear -- GF pivoted from doing advanced nodes (7nm) to being a "boutique foundry" (their term) where they specialized on unique fab/circuit types that could be build on derivatives of technologies like 45nm, 32nm, 28nm, 22nm .... and SOI. Basically GF ran out of money during their 7nm development ... and could not afford to develop it. (even though they already had a few good customers -- including AMD (and IBM, Qualcomm) -- lined up to use it)
Intel is in a tougher situation since their big potential customers don't completely trust them to deliver. And those potential customers don't want to bet their company on Intel delivering.
GF ran out of money but also they could not develop a decent process technology unaided. Remember, the GF inhouse FinFET processes did not work so they licensed Samsung's 14nm and cancelled 7nm due to lack of customer interest. GF tried to pivot to a TSMC Like 7nm but that failed as well. Samsung 7nm did quite well which is why TSMC is moving 7nm fab space to 5nm and 5nm to 3nm.
GF ran out of money but also they could not develop a decent process technology unaided. Remember, the GF inhouse FinFET processes did not work so they licensed Samsung's 14nm and cancelled 7nm due to lack of customer interest. GF tried to pivot to a TSMC Like 7nm but that failed as well. Samsung 7nm did quite well which is why TSMC is moving 7nm fab space to 5nm and 5nm to 3nm.
I’m talking about how, for many years, Intel hasn’t offered innovative products that truly generate market enthusiasm. Sometimes people blame this on Intel Foundry’s failure to advance quickly enough. But the problem runs deeper that even when Intel products, such as the Gaudi series AI accelerators, are manufactured by TSMC, they still can’t compete. Intel continues to rely on the same old PC platform related products to keep the business afloat.
I’m talking about how, for many years, Intel hasn’t offered innovative products that truly generate market enthusiasm. Sometimes people blame this on Intel Foundry’s failure to advance quickly enough. But the problem runs deeper that even when Intel products, such as the Gaudi series AI accelerators, are manufactured by TSMC, they still can’t compete. Intel continues to rely on the same old PC platform related products to keep the business afloat.
The innovative products have been killed quickly or poorly planned like Optne this was revolutionary I would say in storage media cancellation of Rialto Bridge as a GPGPU.
The last good innovation I can remember is Alder lake and than Lunar Lake
I watched the video and found it to be very underwhelming. Probably I am just not the target audience for the video.
Basically it is a summary of how it got so far, with little to no additional input. It felt as if they were just riding the current wave of Intel bashing to gain some clicks.
I watched the video and found it to be very underwhelming. Probably I am just not the target audience for the video.
Basically it is a summary of how it got so far, with little to no additional input. It felt as if they were just riding the current wave of Intel bashing to gain some clicks.
