Advanced Micro Devices: Did You Get Fooled Again?

| About: Advanced Micro (AMD)

I'm going to be upfront with you, my dear readers. I don't like AMD's (NYSE:AMD) chances or its long-term prospects. I think that this is a company that has over-promised and under-delivered for many, many years. More importantly, I believe that much of the shareholder base (especially given the articles that I've seen published by my fellow authors here and elsewhere) is easily convinced by arguments that seem plausible but often miss some pretty significant details. For instance, how many times have you heard the following?

  • Mantle is going to kick NVIDIA's (NASDAQ:NVDA) butt!
  • AMD's Kaveri with its fast integrated graphics is going to take back share from Intel (NASDAQ:INTC)
  • AMD's ARM servers are going to gain huge adoption because ARM is better and will make all of AMD's problems in the server market go away
  • AMD is going to take back market share with Kabini/Temash in tablets!
  • AMD is going to take back market share with Beema/Mullins
  • Game consoles will bring riches to AMD that analysts aren't modeling in!

The Bull Arguments Aren't Subtle

The problem with these arguments is that they aren't subtle at all. For instance, the Kaveri argument. For years, AMD's chips, beginning with Llano, have had superior integrated graphics to what Intel has had on its parts. During that same time, AMD sold very large dies at very low cost, while Intel has been relentlessly driving down the cost of its chips (smaller die sizes) as well as dramatically improving the power consumption. Oh, and during that time, Intel has managed to improve its graphics to the point where AMD's advantage will essentially be completely gone by the end of 2014.

AMD's strategy of selling huge dies with giant integrated graphics blocks has been a complete disaster simply because the vast majority of mainstream consumer notebooks don't need fast graphics as much as they need zippy general purpose CPU performance and a fast SSD. And, for customers that REALLY need that extra GPU "oomph," a discrete chip paired with a fast Intel CPU is almost always the way to go. In short, AMD can't get paid extra for its larger integrated GPUs, so what happens is that AMD's margins are even worse than they'd be without that extra silicon real-estate.

No Mobile Presence... And No X86 Crutch

It has been my view for quite some time that AMD will be collateral damage in Intel's battle against the various rich and powerful ARM (NASDAQ:ARMH) vendors. Intel is relentlessly trying to drive the cost and power consumption of its processors in tablets and phones down which ultimately leads to Intel recycling these cores and IP to build cheap PC-oriented chips that offer better performance/power/cost than what AMD's parts in this same space offer.

In tablets, AMD's "Hondo" and "Temash" have been abject failures from a performance/watt standpoint and from an adoption standpoint. This is because AMD's parts are - as has been the case across all of its product segments over the last year - not competitive. Against an NVIDIA Tegra 4, Qualcomm Snapdragon 800, or an Intel Bay Trail, Temash is exceptionally underpowered. It also doesn't feature a lot of the power-saving goodness that these smartphone/tablet-oriented cores pack.

While AMD is promising much better things with Beema/Mullins, one should always be very skeptical when a company claims a 100% improvement in performance/watt without a process node change. There's the laws of physics and then there's fairy-tale magic - and I've never been one to buy into fairy tales.

The real problem, though, is that AMD has always had a guaranteed "in" as a second-source supplier to Intel for X86 chips. Thanks to Windows being X86 only and thanks to Windows dominating the client compute space, AMD had a guaranteed slice of the pie. However, with low end PCs (AMD's bread and butter) being cannibalized by tablets (and with Intel turning its attention to this part of the market after leaving it essentially to AMD for years), life is going to get much more difficult for AMD's PC chip revenues.

ARM Servers? Please

There's a lot of hype from a lot of players with respect to ARM servers. I've noticed on the conference calls of each of these players that they've all claimed to have an advantage over the other. But what none of them really acknowledge is that Intel is already hard at work securing that space and is shipping very viable product there today with more very aggressive parts in the pipeline.

While some believe that "ARM" means magic, ultimately the server space is one that Intel has dominated for reasons completely unrelated to instruction set architecture. Intel, thanks to its MASSIVE consumer/client revenues, has been able to gain significant leverage in the server space. And, thanks to its process lead, it has consistently offered best-in-class performance/watt and best-in-class total cost of ownership. If AMD couldn't gain share against Intel in X86, why would AMD do any better in the ARM space?

The only potentially threatening players in the ARM server space to Intel's X86 monopoly are the likes of Qualcomm (NASDAQ:QCOM) or Samsung (OTC:SSNLF), both of whom have similar scale and leverage thanks to their significant mobile presence. But AMD, whose products are barely competitive in PCs and big servers today, really has no competitive edge here except for the party line that they've "been in servers for a decade" (and have been bleeding share for most of that time).

The Results Don't Lie

AMD's computing solutions revenue was down significantly Y/Y during Q4, implying some pretty massive share loss to Intel (which reported flat Y/Y revenue in its PC Client Group). Intel has only begun the roll-out of Bay Trail-M, which should further take share from AMD's Kabini over the next year. Things get even worse as Intel transitions to Cherry Trail and Broxton during 2015 (built on Intel's 14-nanometer process) which will not only have VERY small die sizes (and cost), but should offer some pretty compelling performance.

Sure, AMD has the console business (low margin), and sure AMD claims that it's gaining discrete GPU share against NVIDIA (we'll have to wait on NVIDIA's results to really get a read on this; somehow I think this share gain is likely to only show up within a small sub-niche of the overall dGPU market), but the bottom line is that AMD is struggling to break-even, even after aggressive cost cutting/layoffs.

Did You Get Fooled Again?

Make no mistake, and don't buy the nonsense: AMD's competitive position is dire, its cash flow situation tenuous (it got off easy with GloFo forgiving $190M in wafers), and its position within the industry has never been less important. In a world of X86 dominance, AMD always had a guaranteed role and revenue stream. In a world where ARM is the world's most popular instruction set architecture, and in a world where Intel is so desperate to beat back the ARMy that it will create products that invade AMD's niches within the PC market, it's hard to get excited by AMD's shares.

As I've always said: AMD is a stock that trades on news-flow/hype until the fundamentals come in and reset a lot of the "work" all of the headlines promising wonderful things have done. If you understand that AMD is a TRADE and not an INVESTMENT, then you can make a lot of money long/short. If you're hoping that little ol' AMD is going to take shares against Intel and ride the magical console and ARM server waves to new levels of profitability, then the odds are good that you will come away disappointed. If AMD is lucky, the console revenue and ARM server chips may offset the massive share losses that the company will see in PCs (its largest revenue segment) and in traditional X86 servers (Intel still has 4% or so left to take of this market).

Caveat emptor!

Disclosure: I am long INTC, NVDA. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Additional disclosure: I may go long ARMH at any time.