Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
My personal opinion: Why would Intel stop developing 10nm? It would be stupid of them to stop it. Part of Intel's success was the full control of its processes and, in the past at least, a leadership in this domain. Without it, Intel wouldn't have enjoyed its success.
There is only one way Intel can go: open up its fabs and become a foundry. The market is large enough in the high end. There is room for TSMC, Samsung and Intel (Glofo has already given up on 7nm).
Intel has to get its cost under control, provide customers with everything they need to use Intel's processes and not treat those as second to inhouse designs. This is exactly where Krzanich has failed big time, and that from a process guy. Still can't believe it. I repeat it over and over again: Intel needs much more volume in the future to keep up with the foundries. Only way to achieve this is to diversify away from the stalling PC business. It needs foundry, mobile and memory business to achieve this. TSMC has a huge foundry business, Samsung has a huge memory business (and growing foundry) and what does Intel have: still only its PC business.
How does Intel react? Well, they just sold IMFT, the memory joint venture with Micron, to Micron. WTF? This includes the precious 3D Xpoint technology, btw. Doesn't seem Intel's management got any better without Krzanich. The board should be fired asap, if you ask me.
AMD in oversold territory. Exactly where I wanted it to be!
Ah: There are leaks about Zen 2 offering 13% higher IPC than Zen (for server type loads). That would basically mean that it could close the gap in single thread performance and extend the lead in multicore applications. We'll see about maximum frequencies on 7nm but the foundries claim that 5 GHz would be possible. That would make it a great package for servers!
Thanks for the clarification. This would mean that Samsung is the first to get EUV right. Doesn't surprise me, to be honest. They are leading semiconductors from now on, like it or not. Going to be a tough fight for TSMC for sure. Intel still not in the foundry game, for whatever reason (probably deceived by drones).
Samsung now ready to produce 7nm chips with EUV in volume. TSMC still uses immersion lithography for 7nm with many exposure steps and masks. TSMC was first with 7nm but Samsung is first with EUV. We'll see what it does to their foundry business and how it competes with TSMC in terms of pricing, mask costs etc. Samsung definitely is head on head with TSMC.
I think Intel also uses EUV for its 10nm process, not sure though. Anyone got information on that? Maybe some of their problems are related to that, since EUV seems really hard to do.
AMD's share price fell back quite significantly in the last days. Looks like a lot of traders jumped off the ship. Glad I sold my leveraged longs at around 30$ base price. If it will fall closer to the lower 20ies range, I will buy leveraged calls again. Could be a good opportunity to jump in for anyone who considers owning the stock.
That is not old news but new news. You said mine was old. You couldn't find anything official before this anouncement because there was none. TSMC doesn't release information about Apple before they did.
Anyway, now it is official and very clear that TSMC 7nm is in volume production.
By the way, AMDs share price is doing good, too good if you ask me. I trimmed my leveraged calls and mainly hold the plain stock for now. Was a very good trade but mid to long term, I can still see significant upside for AMD, but before that, it is clearly due for correction. Guidance for next quarter wasn't strong at all and I can't see how they won't disappoint the currently extremely high expectations. Will buy more when it corrects. That's almost like the Bitcoin craze now.
Well, you could put that as positive, saying Intel has larger than expected demand, or as negative, that Intel struggles with 10nm and therefore has less capacity than planned
What is really going to be interesting: Which TSMC process are they going to use and how will it compare to similar/same designs made on Intel processes. It may be the first possibility to compare an Intel design on an inhouse vs. a foundry process, apples to apples. Let's hope someone is taking the effort to do some etching.
The other question is: Is there a reason why Intel is outsourcing their budget designs? Is it because TSMC actually can make them cheaper?
Hasn't been mentioned here: The upcoming iPhone seems to use a 7nm processor, the A12. It is not official yet but going to be announced soon. That would mean 7nm volume production essentially would need to be in place right now. The day Intel lost its lead (actually: fell behind).
7nm server parts from AMD at beginning of next year? Design already taped out according to them, so - why not?
@Andy
Look: I never did and never will care about what you write. It is of no use for me.
I always post replies into multiple quotes from different persons in one post in order not to clutter this board even more. Readers normally would have read the previous posts so I expect them to know who wrote what.
AMDs numbers weren't great to be honest. Cash flow still negative, only small revenue growth and guidance was light, with revenue lower than this quarter. What I liked was strong growth in mobile CPUs, where AMD made a very good job. 50% (unit?) growth of EPYC vs the previous quarter is also good. AMD also stated in its conf call that Ryzen products have a 50% plus margin, quite above the 37% overall margin they currently have. They also stated some margin pressure in the desktop market, where Intel seems to fight back.
Intel, on the other hand, had a great quarter at first glance. Very good numbers indeed. Even data center, with that pressure from EPYC, was doing really good. AMD stated, they are still in the lower single digits of market share. Still quite a way to go to the double digit market share they are targeting by next year.
Motley Fool article about the AMD-China Server Joint Venture. Not much new here, just another try to understand this thing. It is unknown how much AMD will benefit from this JV and how much Intel may be losing, but this is definitely something to watch closely. The chinese server market is huge.
Since China now also has grips on another x86 license (through VIA), it could well happen that the chinese will try to "learn" something from AMDs core IP. AMD will have to protect its IP well in this constellation. Since Lisa Su has a taiwanese background, I suppose she knows how to deal with it.
AMD Zen based chips for Chinese supercomputers are going to be available soon. Those are using Zen cores as a license model (and AMD also being a shareholder of those companies), while AMD also gets paid royalties. This model makes a lot of sense for the Chinese, since they want to be able to control the security parts of the design (the cores itself are said to be provided only as hard macros to avoid IP theft). Could mean big business for AMD, especially in Chinese supercomputers, where Intel will have a tough time with its closed designs/silicon. Seems like a really smart move from Lisa Su to me.
RISC V is the other story, where Europe wants, certainly in the longer term, base its own supercomputers on. That is definitely going to take a long time, but I think, looking in the far future, RISC V could replace x86 and ARM as the dominating architecture/instruction sets. An open instruction set just makes a lot of sense and you can build good processors around it without a problem. The processor design itself needn't to be open source since RISC V is released under BSD license. Probably in the future, most computers and embedded systems will be based on a license free instruction set like RISC V. This would remove a major obstacle in todays computing world, where a handful of companies basically block progress by not fairly licensing their instruction sets to competing processor designers. Actually, I never understood why you could patent something like an instruction set - I mean, it's a rather trivial thing to define and more of an innovation road block than anything else..
Gartner says PC market growing for the first time in six years. Consumer demand is still weak. I'd expect the tightening competition between AMD and Intel will lead to higher performance increases and thus will make system replacements more attractive, also for the consumer crowd. It was that competitive spiral between AMD and Intel which lead to peak sales of PC processors in the 2000's. That sluggish 10% performance gain every second CPU generation from Intel didn't make many people switch.
No, same sort of clown. All in the "all clowns when they have a different opinion than I do" category. Simple mind...
This is peak Intel we are seeing now, believe me. A bunch of serious missteps, such as those Intel made in the past years, eventuallly do turn against a company. This is the time now. They can turn that ship, but it is going to be slow, painful and costly from here. In my opinion, they have to fire the upper management and change the company's rotten culture.
The only reason I am not shorting Intel is that being long AMD is the much better bet.
I guess you have heard that AMD offically stated that Zen 2 based EPYC samples, made on TSMC's 7nm process, are evaluated in its labs. Looks good for an end 2018 launch. The server chip is said to be reaching to up to 64 cores for a single slot, distributed over four single dies on a carrier.
What you probably haven't heard is that there seem to be supply issues with Ryzen on Globalfoundries 14/12nm process. While AMD is very successful in the retail business, where it is said to have a 70% share vs. 30% from Intel (and mainboard producers seem to confirm that), AMD is still weak in the volume business. Possible reason could be named supply issue of Ryzen CPUs, since most of the market is business related and any supply issues of components are unacceptable for the big PC producers.
In that light, it totally makes sense for AMD to first concentrate on the high margin server business with its shift to 7nm, since that process likely also isn't going to be available in high volume for a while. AMD certainly has another option, which is using other foundries, like TSMC, Samsung and Globalfoundries together. It seems they are doing this now.