https://seekingalpha.com/article/4176501-amd-ready-server-outperformance
10 replies (most recent on top)
@4jfc Today niche, tomorrow still niche. My original point stands. You said earlier "Intel lost server business earlier on" which is correct because it's NOT IN THE NICHE SPACE............. otherwise still not much lost
@3iqw - Fair question, and the obvious answer TODAY is not much. EPYC was only released about 9 months ago and is going thru its customer qual stretches right now, much as Lisa Su said in the 1Q18 earnings call. Any sizeable ramp would occur 2H18. The other compelling product is the new Cavium Thunder X2. These guys were always a serious threat in the COMMS space, and this new X2 is much better architected for those applications than XEON is. Cavium is also trusted by the COMMS customers.
@2jpk ...and may I ask just exactly how much market share these niche products in the "specific applications category" occupy as a percentage of the overall market?
@2cko - you have a point, and that was surely the case the last five or so years. But before that, Intel lost a ton of server business early on because they had no answer for AMD. That's why KS was a hero to everyone before DB took over DCG and did basically nothing to advance Xeon architecture. Today's situation is very different. Both AMD and now Cavium have compelling offerings that work better and cheaper in specific applications. Hammering x86 into every hole just won't work anymore.
AMD's only really good for OEMs to use as leverage for negotiating prices with Intel.
They'd threaten to buy AMD and then end up buying Intel anyway, just at a slightly less inflated price.
In the end, Intel still makes a killing.
Seeking Alpha has questionable analyses. That said, Epyc will make many inroads, slowly but surely. For the reasons stated earlier, it will be a tentative embracing overall until one of the big vendors decides to push back hard at Intel. SemiAccurate is sure betting on that.
Yup, Intel's the best. All others are just mere distraction. How else you would think Intel MIT can spend the ridiculously crazy amount of $ every quarter.
Yeah, like any compute infrastructure team is going to bet their jobs on AMD for the long haul.
Better hop on the INTC bandwagon while you still got time!
AMD is a paper tiger.😂
"Top alternative" doesn't automatically mean "enterprise customers will actually migrate"
https://www.forbes.com/sites/greatspeculations/2018/05/23/data-center-group-will-likely-drive-intels-near-term-growth/