One of the first things that we talked about was the “launch” of AMD’s 2.5HBM based 380/309 GPUs. As of right now the work expects them to make the official announcement at Computex. What many are failing to see is that a “launch” and general availability are not the same. Right now the word on SK Hynix is that 2.5D HBM is not exactly a common part. It is true that the use of an interposer layer makes building the GPU simpler (hence the 2.5D as 3D HBM is right on top of the processor), but that does not mean that the GPU will magically become more plentiful. AMD might have a small stock of these products just so they can sell a few, but that is not going to help their cash flow.
Moving on to AMD’s plans for 14nm everything that they are releasing (in a early update to investors) shows 14nm parts coming no sooner than Q3 2016. This is in line with what we were told which was mid to late 2H 2016, although we did report on the rumors of parts hitting as early as 1H 2016. The parts do look impressive they might hit the market too late to have much of an impact. Remember Intel is already working on 10nm and those parts will hit the streets around the same time as AMD’s Zen powered 14nm APU/CPUs. It could be a wash for AMD which is as bad as a product failure at this stage.
The fact that AMD released an update to investors early could be an indication that some are getting nervous. In our experience company does not need to send out an update like this unless someone is asking questions or there is a problem. Clearly some investors are doing this and want to know what AMD has planned. This is never a good sign for a company regardless of what they produce.
AMD has some good products coming to the market. The 380/390 with 2.5D HBM will be a game changer for the graphics market. By comparison NVIDIA will probably not have theirs ready until 1H 2016. This gives AMD a decent lead if they can get the products to their partners and then to the consumer. We are just not confident that they can do this with the supply limitations we are hearing about.
By contrast Zen looks like it will be a great CPU core, but it will be late to market. AMD should have been working on getting Zen out this year. Unfortunately they have been stuck using an older process since they did not have the funds to help push their fab partners to a better one. When AMD owned their own Fabs they were in a better position to keep pace with Intel… at least until the ATi purchase. After that they just did not have the money to spend on keeping those Fabs updated and running. This is one of the reasons they sold them all off.
In the end AMD has suffered from more than a few setbacks. Most of these were due to poor leadership. Ruiz and his purchase of ATi hit AMD hard and they have still not fully recovered from it. The lack of real leadership since that time has also had an impact including Rory Reeds tenure as the boss. During that time the shift to a mobile focus made a lot of talented people leave the company while they lost a massive amount of ground to Intel and ARM. Now they are in a position where they need to do something to get things going, but they do not really have the money to do it in the timeline they need.
Like we said a couple of days ago: the next 18 months are going to be very telling for AMD’s future. Right now there are very few bright spots in that time frame. There is optimism from AMD fans and some members of the press, but if you look at things realistically they are not that good. If they were, well there would not be rumors about someone buying AMD flying around and we would have seen a few more product launched this and last year.