To: Vattila who wrote (55012 ) 4/26/2024 3:55:30 PM From: Joe NYC Read Replies (1) | Respond to of 72186 Client is a 10 billion per quarter market, AI GPU market appears bigger, but without astronomical markups, it would likely be less or equal to 10 billion per quarter. As opposed to competing with the world's most pre-eminent company, NVidia, on client, AMD is competing with Intel which is a little bit limping now I think I am the only one who is expecting a breakthrough in client. I think the stage is set for that over the upcoming year, with upcoming Zen 5. Intel kept getting away insane power consumption to stay competitive until it could not. There are some very vocal comments on Interment about Intel's practices (of pushing the chips beyond their limits), and the discontent may spread.What it may result in is during the upcoming Zen 5 reviews, the reviewers may use safe, non-overclocked settings, punishing the highest performing Intel CPUs the most. Which means the gap between the fastest Zen 5 and fastest Intel 149000K may be a gulf like we have never seen before. Intel may be going backwards in performance, especially MT performance by 5% or more, while Zen 5 will have anywhere between 25% to 40%, even 50% uplift on some FP code. I don't see how Intel's 85% dollar market share in client under these conditions. Realistically, it is not going to be instant. Product introductions will be staggered over the next 4 quarters. Zen 5 desktop and Strix Point will start in Q3, should be boosting ASPs and dollar market share. Then Q4/Q1 Strix Halo - providing Halo over the whole notebook line and Kraken to crack the unit market share. I don't think it is far fetched to expect AMD to double its dollar share from 15 to 30% in client, which is 1.5 billion per quarter run rate, or 6 billion annual. For comparison, all the AI hype is so far over $3.5 billion/ Various excuses why OEM should stay with Intel are disappearing one by one. Intel's IDM model is broken, Intel has to go to TSMC like everyone else now, and Intel made a mistake of signing up with TSMC N3B, while AMD will have a higher performing chips from a cheaper N4P / N4X node.