Despite the turbulent times over at Advanced Micro Devices (AMD), the company still manages to push on and release new products for its core CPU/APU segments. Its "Llano" and "Trinity" products, while lacking on the CPU side compared to its competition, still manage to lead very nicely when it comes to the integrated graphics portion.
That being said, it's time to look towards the future as the 2013 client roadmap from AMD has just been leaked. To give investors more color, I aim to do an analysis of the roadmap and give some sense of the competitive position AMD will have during 2013 on the desktop client PC side of things.
Without further ado, here is the roadmap, courtesy of donanimbaber:
Performance CPU Analysis
At the high end, AMD will get through the year on its "Vishera" processor. This is the high end FX refresh that recently launched. According to reviews, the latest FX chips are quite a bit faster and more power efficient than last year's "Bulldozer" products.
The worrisome thing here is that the processors come in at a fairly large die size - 315mm^2 - partly due to the fact that the chips are built on a 32nm process, but also due to the emphasis on very large caches throughout the core (for workstation/server oriented applications). This is twice the area of Intel's (INTC) "Ivy Bridge" processor which weighs in at 160mm^2 (and it comes with integrated graphics). Intel's advantage here is partly due to the 22nm process, but the intended market (client v.s. server/workstation) also helped keep the chip slimmer.
AMD will be selling these chips quite a bit cheaper than Intel's at similar performance levels, so its margins profile will obviously continue to trail its rival's. There is a competitive silver lining here, though.
Note that AMD - before taking the inventory write-down charge - operated with margins in the 44% - 48% range. The prices - and die sizes - of AMD's new FX processors are virtually unchanged from their 2011 models, so I don't expect too much of a hit on the gross margin front from the competitive prices. But the interesting thing is that, relative to its Intel competitors, the product is much more competitively positioned than last year's FX chip according to Anandtech,
Vishera is a step in the right direction for AMD, it manages to deliver tangibly better performance than last year's disappointing FX processor without increasing power consumption.
More importantly though, is the effect on the prices of competing Intel products. Since the release of AMD's new FX chips, the prices on Intel's flagship Core i5 and Core i7 chips have come down since the AMD release:
A quick look at the boxed prices for these CPUs at the leading e-tailer, Amazon.com (AMZN) quickly reveals that Intel and/or the partners have been subtly lowering prices of the Intel chips to below these official prices:
It is unclear whether Intel is lowering its prices to the distributors or if the distributors are just lowering their markups, but one thing is clear: AMD's new products are affecting Intel's pricing, even if in a subtle way.
Mainstream APU Analysis
While AMD competes in the higher end workstation/enthusiast/gamer CPU space, its bread and butter and main focus with the APU (CPU + GPU built in together). The main advantage that AMD has in this space over its rival Intel is its integrated graphics. While Intel's products here are quickly improving, AMD holds an unquestionable lead in the integrated graphics space in every measurable metric: performance and image quality. Anandtech had this to say on the matter,
Ivy Bridge was a good step forward for Intel, the problem is that only the high end Ivy Bridge configuration borders on acceptable. The HD 2500's performance is really bad unfortunately. It's easy to appreciate how far Intel has come when we look at improvements from one generation to the next, but when you start running benchmarks on Trinity it really compresses the progress Intel has made.
The drive for better integrated graphics is attractive to OEMs for the obvious reason that they can enable fundamentally new features (better gaming, multimedia) without needing to buy a more expensive discrete chip from AMD or Nvidia (NVDA). Of course, these graphics solutions are unsuitable for hardcore gamers, but for casual ones, these products add significant embedded value in the system at very little additional cost.
In this space, the leaked roadmap indicates that the first half of 2013 will be dominated by the current "Trinity" APU (32nm "Piledriver"), but then in the second half, there is a new APU that shrinks "Piledriver" to 28nm, seemingly improves the integrated graphics, and stays on the same FM2 platform.
There are a couple of things to note here:
- Socket Compatibility Improves CPU + Chipset Linearity - a big problem that AMD had in the last quarter, which likely led to the inventory write-down, was the fundamental socket incompatibility of its "Llano" chips with its "Trinity" APUs. With a bunch of unsold CPUs (with unsold, out-dated platforms to couple them with), these products became "stale" and had to be sold at a significant discount. With the compatibility of the older "Trinity" APUs with the "Richland" APUs, the platform remains the same and stays fresh - removing any incentive for the OEMs to hold back on purchases that would be caused by fear of stale inventory.
- Reuse Of Existing IP - AMD has a smaller R&D budget and lower sales, so it needs to make the most of its development costs. The 2013 product line fully leverages the "Piledriver" micro-architecture. While the original "Bulldozer" was made obsolete within less than a year, the new "Piledriver" will get a lot more mileage. The graphics cores in the new APUs are generally inherited from the higher end discrete graphics chips, the sales of which help amortize the brunt of the development cost before being handed down into the APU lines.
The move to 28nm should shrink the die sizes a bit, which should help improve the gross margin line slightly. Further, as the next generation "Steamroller" processor should also be built on 28nm, this will give AMD time to become familiar with the new process on a familiar technology base before moving to a significantly updated core design. This mirrors Intel's "Tick-Tock" strategy and should help mitigate risk on the process side going forward.
ASPs are not likely to change.
This is AMD's "sleeper" - the "Jaguar" low power core. While Intel is busy beating AMD up in the high performance segment of the market, AMD has quietly been able to seize the "good enough" category with superior performing parts to Intel's Atom with its "Brazos" and "Brazos 2.0". AMD continues the fight with its low power 28nm APUs code-named "Kabini" based on these "Jaguar" cores.
Here's what we know about this new core:
The bottom line is that in the instructions per clock side (that is, how much actual processing work is done per "GHz") goes up by over 15% compared to the previous generation, the clock speed goes up by over 10% (likely enabled by the move from 40nm to 28nm), the floating point unit (decimal number crunching ability) is significantly widened, and the core count doubles. The performance enhancements here should be quite tangible and will make AMD a very deadly force in the low end of the notebook/thin and light segment.
Here's why I believe that this - and not the high end - will be AMD's "secret weapon:"
- Move Into Tablets - While AMD released its Z-60 APU for tablets, it is not really good enough on the power consumption and platform integration front to really be viable. The new Jaguar - thanks to a new micro-architecture, tighter platform integration, and a process shrink - should be able to go where AMD's previous products have not been able to.
- Good Enough Ultrathins - Intel has been very aggressively pushing its "Ultrabook" line of laptops - essentially very thin and light notebooks with excellent performance. However, a primary criticism of these sleek new devices is that they are priced expensively ($679 and up). A major market opportunity for AMD here would be to partner with the OEMs, build high quality laptops (using good metals, batteries, etc.) but then stick in a "good enough" quad-core "Jaguar" based APU. This would allow for high quality machines to be built on an inexpensive processor - helping AMD gain market share with the masses.
The major fear that AMD shareholders should have is if AMD decides to simply aim these new chips are Netbooks. As is quite obvious now, the Netbook has gone way of the dodo, thanks to cheap tablets.
AMD's 2013 roadmap is sensible. On the mid to high end side, we see a logical evolution of the current product line. At the right prices and with the right OEM deals, AMD should be able to position themselves quite well. The exciting stuff happens with the brand new "essential" APU with the "Jaguar" cores in the "Kabini" APU. These cheap chips - coupled with "good enough" performance - could be just what AMD needs to move into newer, low power segments as well as take advantage of the budget ultra-thin notebook category that is likely to spring up.
AMD isn't dead yet, and it doesn't need to "beat" Intel to remain viable - it just needs to pick and choose battles that it can reasonably win. It seems that on the PC client side at least, AMD is making the right moves. Execution and delivery will be key here, which have traditionally been AMD's weak points.