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!
The path to to AI follows computing, communications, networks, data bases and control systems to name a few. What do the readers think the next large ecosystem for semis will be? I personally feel it will be medical and sadly military. AI/ML should also be the largest ecosystem as it will come to dominate almost everything it touches. Many will fight the AI/ML ecosystem as it is set to upset almost everything from automating the professions and extending itself to numerous manual functions by greatly increasing the level of automation in everything. There will be numerous efforts to thwart this expansion by established, entrenched interests who it will be shown that only represent themselves and not the greater good of society. Any thoughts, speculations, or radical expansions of an existing would be appreciated. I feel the only limits will be the human imagination. Thanks
I think currently what we are experiencing is the surge in demand in AI training, but we are seeing plenty of momentum from companies that want to bring AI from cloud to the edge from companies like, Intel (Meteor Lake and Sapphire Rapids), Qualcomm (Qualcomm AI stack), and Microsoft (windows 12), Google (TPU on pixel phone). I think if AI/ML training is the steppingstone for Nvidia. Soon Inferencing will likely drive the next refresh cycle of personal hardware for companies like Intel and Qualcomm.
Chiplets will have their own ecosystem, absolutely. It is building as we speak. It will be similar to the IP ecosystem foundries have. Silicon proven chiplets for specific foundry nodes to limit risk of respins.
I think currently what we are experiencing is the surge in demand in AI training, but we are seeing plenty of momentum from companies that want to bring AI from cloud to the edge from companies like, Intel (Meteor Lake and Sapphire Rapids), Qualcomm (Qualcomm AI stack), and Microsoft (windows 12), Google (TPU on pixel phone). I think if AI/ML training is the steppingstone for Nvidia. Soon Inferencing will likely drive the next refresh cycle of personal hardware for companies like Intel and Qualcomm.
On its earnings call Mediatek suggested successful/ popular use cases of AI on smartphone may lead to replacement cycle. Semi content growth per smartphone for MTK is more a 2025 possibility as 2024 chip design work started before GenAI wave became popular.
