The difference between [the P4] and the [Athlon] die size is frigging huge - AMD's Jerry Sanders III
CHIPMAKER Intel's recent quarterly financial results show a worrying trend for the firm and its reliance on x86 desktops and servers.
Intel's new CEO Brian Krzanich arguably has come into the job at a worse time than his predecessor Paul Otellini. The fact that Intel highlighted that its Atom chip has ended up in a single tablet highlights how poorly the firm has been doing in the smartphone and tablet markets.
When Otellini came into the CEO job the situation at Intel was far from rosy. The firm was being battered by its decision to continue with the Netburst architecture and financially it was suffering from ballooning expenses.
Otellini brought fiscal order to Intel and while some commentators suggest he missed the boat on low-power chips, the fiscal prudence Otellini brought to the firm means Intel at least has a fighting chance to go up against the many ARM vendors and its traditional rival AMD.
Krzanich however doesn't have the luxury of boosting the bottom line by cutting costs. Instead, the man known for leading Intel's manufacturing to the top of the semiconductor game needs to figure out how the firm is going to do battle with companies that are equally as well equipped as Chipzilla.
Some commentators are suggesting that Intel should license one or two ARM products and be done with it, but the fact is that Intel already has an ARM license and chooses not to use it. The firm's decision to stick with an all X86 product line - it does produce FPGA/Atom hybrid boards - is not due to vanity or pride but the question, if Intel made an ARM chip, why would anyone use it?
Aside from the obvious point that Intel wouldn't be the only ARM option available to device makers, there is little to say that Intel's ARM chips would have anything unique over those made by Samsung, Qualcomm, Texas Instruments, Nvidia or even AMD. The same argument could be applied to AMD, except that AMD has a wealth of graphics technology that consistantly outperforms Intel.
The fact that Intel's GPU in Haswell can just about keep up with the years-old GPU core in AMD's Richland APU says a lot about how much money, technology and effort a company needs to put into building competitive graphics technologies.
However bemoaning Intel's lack of success in the smartphone and tablet markets as a reason for disappointing financials is a lazy analysis of the whole situation. Something that is far more troubling is Intel's server business, which reported no growth from a year ago.
Intel bet big on server growth and it was a perfectly reasonable play. After all, if Intel couldn't be in smartphones and tablets, why not count on the explosive growth of those devices, which rely heavily on internet services, to grow server sales?
The problem for Intel and other semiconductor vendors is that service providers are becoming increasingly concentrated into firms such as Amazon, Facebook, Google, Apple and even Microsoft. Everyone knows that Google and Facebook build their own datacentre kit, but even those outfits that do not build their own bespoke servers can order directly from the chip vendor and demand significant discounts, so while the number of chips being sold grows, revenues are unlikely to grow proportionately as buying power continues to increase.
Next week Intel will reveal more information about its upcoming Ivy Bridge Xeon and low-power Avoton processors. The firm's two socket Ivy Bridge Xeon processor is its bread and butter server product, so we can expect a considerable push in this area.
For Intel, Avoton is just as important as its Baytrail Atom chips because it needs something to counter what is likely to be an onslaught from some ARM licensees in 2014 in the microserver market. Whereas Intel has very little to lose in the smartphone and tablet markets, the firm has close to 90 percent of the server chip market, meaning that there is pretty much only one way the firm can go.
Intel has the benefit of a mature x86 software stack with Avoton, but what it doesn't have is the benefit of the choice that ARM vendors will bring to potential customers. However the microserver market is likely to make significant use of open source software and operating systems, meaning that ARM chip vendors will not need to rely on proprietary software vendors to support their chips.
While Krzanich's predecessors had the ability to exploit software vendors to sell processors, Chipzilla's new CEO is going to have to compete against more rivals than ever before. µ
Sign up for INQbot – a weekly roundup of the best from the INQ