With shares of Advanced Micro Devices (AMD 3.99%) trading near their 52-week low, down nearly 30% over the past month after the company reported abysmal third-quarter earnings, it may seem tempting to pick up shares of AMD on the cheap in hopes of a successful turnaround. But there are a few important things that investors should consider before investing in AMD.
The core business is falling apart
One of the goals of AMD's turnaround effort has been to diversify away from the PC, and getting its custom-designed chips into the major game consoles accomplished just that. But the Computing and Graphics segment, which includes the company's CPUs and GPUs, is still the largest segment by revenue, and its rapid deterioration spells trouble for AMD.
During the third quarter, revenue from this segment declined by nearly 16% year over year, and a small operating profit during the same period last year was turned into an operating loss. AMD has done a good job at keeping costs in check and preventing huge losses in this segment, but the declines don't appear to be over. AMD expects total revenue to decline by 13% sequentially in the fourth quarter, and some of this decline will probably be the result of continuing weakness in its CPU and GPU business.
Why is AMD doing so poorly in these markets? It's not difficult to figure out. AMD has spread itself so thin, with its limited resources going toward PC CPUs, GPUs, server CPUs, both traditional x86 chips and ARM Holdings chips, and its semi-custom business. Intel (INTC 0.67%), which is dominant in the PC CPU and server CPU markets, outspends AMD by nearly a factor of 10 when it comes to R&D. Intel also fabricates its own chips, and the first chips based on its 14nm process will be shipping in volume this year. AMD spun off its manufacturing business a few years ago, and Intel's manufacturing lead over other foundries puts AMD at a severe disadvantage.
In the GPU market, rival NVIDIA (NVDA 1.20%) has managed to win market share from AMD over the past few years. NVIDIA now has a 65% share of the discrete graphics card market, and it hasn't been suffering from the same issues plaguing AMD. Not only did NVIDIA grow its GPU revenue during its most recently reported quarter, but profitability also improved significantly.
Why has NVIDIA been able to sustain such a lead over AMD? Again, it's simply a matter of a lack of resources. NVIDIA now outspends AMD in R&D, and it will be difficult, if not impossible, for AMD to match NVIDIA's new products in terms of both performance and efficiency as long as this R&D gap remains. NVIDIA recently launched some new GPUs, the GTX 970 and 980, and the company has managed to provide the same level of performance as AMD's flagship GPU for about $150 less, forcing AMD to slash prices. With AMD unable to commit the same amount of resources to the GPU business as NVIDIA, profits are going to be hard to come by.
The semi-custom business isn't that exciting
AMD's semi-custom business, where it designs custom products for its customers, was launched when the company won the business of both the PlayStation 4 and the Xbox One. This has been a profitable endeavor for AMD, with the operating margin of the segment containing the semi-custom business currently sitting in the mid-teens. AMD would be posting big losses if not for the semi-custom business.
However, the potential for growth may be limited. The company announced that it had secured two new semi-custom design wins during its recent conference call, but these are far smaller than the game console wins. These two deals will produce $1 billion in revenue over three years starting in 2016, or about $333 million per year. At the current 16% operating margin of the semi-custom segment, that's about $53 million in additional operating profit annually. Compared with the $310 million in operating profit earned over the past 12 months companywide, these new deals provide a nice boost.
But the game consoles will peak at some point within the next few years, and there's no guarantee that AMD will win the next round of consoles. The downside to this push into semi-custom designs is the limited time frame of the deals. AMD will need to continually score new design wins to make up for the end of old ones.
Margins could also prove to be lower for new semi-custom deals compared with the game consoles. The game console deals are unique in that they required an x86-based system on a chip combining a powerful CPU and GPU, and AMD was really the only option available. For something like a 64-bit ARM CPU, which is reportedly one of AMDs new design wins, AMD doesn't have the same advantage. Going forward, the mid-teens operating margin of the semi-custom segment may not be sustainable.
AMD's core CPU and GPU business is doing poorly, and there's really no reason to believe that things are going to get better. The semi-custom business provides the opportunity for growth, but the new deals are smaller than the game console deals, and it's unclear whether the high margin of the segment will prove sustainable. Whether AMD turns out to be a good investment from here depends on how successful the semi-custom business becomes, and the points that I've laid out are things investors should consider before investing in AMD.