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.
First of all: Congratulations to all Intel longs! I still don't see a good long term perspective for this company, as I stated before. I made a nice +100% gain on the shares though, when selling them at the beginning of this year, so no regret really.
I had a quick look at the 10Q, especially the segment breakdown. You could see that, QoQ, revenue was slightly down in the client computing group and up in the datacenter. IoT didn't move much and non volatile memory is up significantly (higher memory prices). Programmable solutions didn't move much either.
Looking at the operating income generated by each of those segments, one can see that income generated by the client computing group was up, even though the revenue was down a bit. This looks like better cost position (yield, staff whatever). The data center group, though revenue increased, was actually generating a slightly lower income than in the same quarter last year (more competition from AMD?). Other contributions to the bottom line are lower losses in the non volatile memory segment (yes, it still makes a loss even though memory prices are at a record high) and in the all other segment. In the all other segment, the Mobileye acquisition is included, non consolidated, so Intel paid a high price to reduce the losses in that segment (selling McAfee and paying a lot for Mobileye).
So, in the end you have a minimal increase in revenue with a larger increase in operating income, due to all those factors mentioned. There are a lot of one time effects and I especially see the loss making memory business and the competition from AMD in the datecenter business as big issues. The improved cost position in the client computing group may reverse once Intel switches to the next process node, starting with low yields. AMD won't allow them to stay to long on the higher yielding older nodes.
Was this report about AMD EPYC vs Intel Xeon posted here before? It seems performance between the two is basically a wash but EPYC sells at less than half the price.
I found that part particularly interesting:
HP about to release a Ryzen based Ultrabook. There have also been Geekbench results leaked in September which indicate that this platform (Raven Ridge) is pretty close in terms of performance to a similar platform from Intel (Intel Core i5-8250U). It's going to be interesting to see what power consumption this will have compared to Intel's solution. Graphics could give it an edge since it integrates a Vega-M based GPU which could give it an edge. Pricing will certainly be crucial for this. If this is competitive, Intel will have to reduce prices once more, now for the big volume and margin market - not going to be good for its financials. Raven Ridge and Epyc could really harm Intel's results.
Leaked AMD roadmap. According to rumours, AMD will release Pinnacle Ridge next year, presumably a Zen derivate on 12nm with smaller design improvements.
Raven Ridge, the notebook design from AMD, seems to be a 2018 release. According to the leaked roadmap, that's with integrated Vega GPU. If the power consumption on platform level will be good with decent processor performance, the better performing integrated graphics may give AMD the edge needed to compete in this high volume market. Very important product for AMD. Going to be tough for them, though.
The more interesting thing probably is, that AMD is planning to have a Zen 2 design, named Matisse, in 2019 already. We'll see which process that's going to be on. Could be 12, 10 or even 7nm. AMD has the benefit of choice, depending on their needs, regarding those processes. They are not forced to use leading edge if it doesn't bring the financial or performance benefits they need. Lisa seems to keep the foot on the pedal. Since Zen was a major performance improvement for AMD, we'll see what they can do with Zen+ and Zen2 and what Intel will deliver at the same time. Will it be the Athlon moment for Intel or the Conroe moment for AMD? Could be the tipping point for either company.
AMD said to produce new CPUs on Globalfoundries 12nm node by next year. If AMD takes the chance to also tweak the architecture a little towards higher performance and clock, Intel will have a hard time to keep its thin lead.
The improved (from Samsung/Globalfoundries 14nm) process is said to bring a 15% density and a 10% performance improvement.
Since we know for a while already, against most claims made on this board, that Globalfoundries 14nm in fact is denser than Intel's 14nm process, at least in real world, this additional step will be tough to overcome for Intel. Maybe Intel is becoming the chased one by next year. Having to spend all that money in new fabs is going to be costly with a shrinking main business.
Samsung projected fab equipment spending:
Samsung to invest 18 bln in semiconductor plants in South Korea
Samsung's chip sales surpass that of Intel
Change of times ...
Definitely not a strong signal with AMDs share price movement, I am not going to deny that. The question is: What is priced in, what are the expectations for the coming quarter and those following? Not that AMD had a good track record in the past. It is always impossible to say what exact scenario the "market" is pricing in and what the time scope is. Fact is: AMD had a very strong run up, the quarterly wasn't particularly strong and many traders cashed in. The Outlook at least suggests that AMD will be at the cross to profitability by next quarter, which already is a strong positive for AMD. It gives AMD air to breathe and in case Ryzen, Naples and what follows will flush some money into AMDs pockets, I have no doubt that, with its obviously decent management and good team of core developers, it has a shot at winning significant market share from Intel in the longer term, especially with the burden of running its own fabs gone and ever stronger foundries in the back. I still see Ryzen as a very strong proof for what substance is left at AMD's R&D team and think what they can do with some more money in their pockets. In comparison, everything Intel did in the past with all its manpower in those years was just plain weak.
We'll see about next quarter, when Ryzen 7 and 5 are fully baked in. We likely will also know more about Naples by then.
Regarding costs, as said, I doubt Intel has a better cost position. Its stagnating business isn't going to be sufficient, margins will sink in the future. On the other side, AMD likely has highest wafer costs initially which are sinking with volume, as is typically with foundry contracts. That means increased margins in the future as Ryzen is shipped in volume.
For Intel, on the other hand, margins are going to shrink since they have to pay ever increasing expenses for next gen processes.
Since it is common on this board to brag about one's success stories in his portfolio (and stay muted about the dead ones), I have to mention Samsung again (sorry, guys). The shares are up more than 100% from the time I bought them. Business is going really really well, despite the problems with the Note 7. The reason is simple: more than 60% of earnings come from semiconductors (mainly memory). Samsung also clearly stated that it is watching the supply/demand balance closely and that they don't intend to start a war on prices in the near future. They can easily do this with their around 50% market share. I'd expect memory prices to remain stable at the currently high prices.
What is also interesting is their statement about the LSI (foundry) business:
Some benchmarks of Ryzen 5 on a spanish website. It seems to be doing pretty well, even against some Intel I7 cores, so AMDs middle class seems to do well. We'll see about pricing. I don't expect Intel to have a significant, if any at all, price advantage. Their small process lead at 14nm is likely eaten by higher costs of their in-house fabs (due to absent foundry volume). I don't think Intel will be able to beat AMD by discounting this time.
Maybe AMD is even considering hopping on the 10nm train soon to improve its cost position even further.