The server chip market is dominated by Intel (NASDAQ:INTC), and its near-monopoly level market share has allowed the company to extract hefty margins from the segment. Intel generated a 46% operating margin from $11.2 million in sales from its data center group in 2013, an even higher margin than PCs, another market where Intel is dominant. With those kinds of margins, it's not surprising that other companies are vying for a piece of the market, and with mobile devices bringing about the rise of chips based on the ARM architecture from ARM Holdings (NASDAQ:ARMH), it was only a matter of time before competition began knocking on the door of Intel's server empire.
While ARM has high hopes for the server market, with the company aiming for a 10% share by 2017, recent news suggests that stealing market share from Intel may not be that simple. NVIDIA and Samsung, both aiming to power servers with their respective ARM-based chips, have now abandoned those efforts. While this may seem like a good thing for companies betting big on ARM-based servers, like Advanced Micro Devices (NASDAQ:AMD), due to less competition, Samsung's and NVIDIA's lack of confidence should act as a warning sign for AMD investors betting on a server renaissance to turn the company around.
Why are NVIDIA and Samsung giving up?
NVIDIA was exploring pushing the upcoming 64-bit version of its mobile Tegra processor, called Project Denver, as a server processor, but the company has changed strategy and will only look to provide GPUs to complement ARM server chips from other companies. This is very similar to NVIDIA's decision to stop trying to be a mainstream mobile processor provider and instead focus on areas where its graphics expertise matter, as an attempt to compete in the server processor would have almost certainly been a disaster.
Samsung had made a major effort to develop ARM-based server chips, hiring people from AMD's server division to lead the charge, but that project is now being scrapped. Samsung is, by far, the largest seller of smartphones, moving over 300 million units in 2013. It designs some of the chips that it uses in-house, so the company has expertise regarding chip design. But, designing chips is an R&D-heavy endeavor, and the company often ends up using chips from other companies for its phones.
Samsung likely realized that the expense of pushing into servers simply wasn't worth the potential rewards, given both Intel's dominance and other companies vying for the same market. If ARM chips win 10% of the server market by 2017, that would only be around $1 billion of revenue split between the major players, based on Intel's current data center revenue. That's nothing compared to the smartphone processor market, and the return on investment is likely far too low, or nonexistent, for Samsung to continue its server push.
AMD investors should be worried
AMD is making a big bet on ARM-based server chips in an effort to reverse its collapsing server market share. AMD had a 15% share of the market in 2007, but this number has since fallen into the low single-digits. The company expects to claim 25% of the market by 2019, but this number isn't as impressive, considering that ARM's market share is unlikely to be very high at that point. The company will still make x86 server chips as well, but its recent record in that area leaves a lot to be desired.
At stake here is revenue in the hundreds of millions of dollars, assuming that ARM makes major inroads against Intel and that AMD rises above the fray of competitors trying to enter the market. Neither of these is guaranteed.
While ARM may manage to win some server market share from Intel, whether any one company can win enough market share to justify the expense of entering the market is a big unanswered question. Samsung has probably given up for exactly this reason, and it's unlikely that ARM server chips will have a positive impact on AMD's profitability.
The bottom line
ARM's dominance in the smartphone market, along with incredible growth over the past few years, has allowed some smartphone chip companies to do very well. But, the server market is very different. Not only does Intel hold a position that could prove to be unassailable, but even if ARM manages to meet its market share goals over the next few years, the market for ARM-based server chips will be small, likely too small for many companies to even justify entering the market at all.
The smartphone processor market was worth about $18 billion in 2013, for the sake of comparison, and even a small slice of that market would be worth more than a big slice of the ARM server market. ARM servers aren't going to be a panacea for AMD, and offering a different architecture won't change the fact that the company hasn't been competitive with Intel for years.