Altera (ALTR) recently announced that the upcoming FPGA that Intel (INTC) would be building for them contains an ARM (ARMH) Cortex A53 core. This has led the masses to suspect that the floodgates are now open and that Intel will now be building processors for its competitors such as Qualcomm (QCOM), Broadcom (BRCM), and NVIDIA (NVDA).
Do not believe this. This is incredibly unlikely and certainly not to be extrapolated from the Altera deal. This was 100% entirely expected as Altera's FPGAs do not compete with any product line that Intel offers today and typically have included ARM cores (which are a very small part of what the actual chips do).
ARM Isn't A Dirty Word
First of all, Intel has no real beef with ARM itself; if you want more "shocking" details, Intel employs small ARM cores in its X-Gold cellular basebands and very likely also employs ARM in its solid state drive controllers. ARM is not a dirty word and for very low power embedded applications, its IP is most excellent. So, Intel already builds ARM today.
What Intel doesn't "like" ARM for is anything that competes directly in its markets. That is, anything where the primary thing that you're selling is a compute device. This means servers, networking, PCs, tablets, smartphone apps processors, and higher performance embedded (in-vehicle infotainment, industrial controls, and so on). In short, if Intel designs its own products for a given space, it will not build ARM (or MIPS, or Tilera, or anything else that competes with its own products) chips.
This means that Intel won't be building the following:
- Processors for PC chip competitor AMD (AMD)
- Chips for server/networking competitors Broadcom, Cavium Networks (CAVM), Applied Micro (AMCC), Calxeda
- Smartphone/tablet processors for NVIDIA, Qualcomm, Broadcom, MediaTek, Samsung, Apple (okay, I'm not 100% sure on this one at this point as Apple's chips technically do not compete with Intel's, but I seriously doubt it)
- HPC accelerators for NVIDIA
But Intel Still Wants X86 Everywhere
Intel still wants X86 everywhere. This is a goal that has been clearly laid out by Intel's CEO and COO. At the Pacific Crest Global Technology Conference, President Rene James said that while it would be looking to take more foundry business (i.e. premium high end, high margin jobs), the real goal would be to build strategic partnerships in order to get X86 into places that it doesn't go today. In particular, Renee James claimed that it would be pursuing a "semi-custom" business very similar to what AMD is doing. The goal?
Say somebody has an internally designed block of IP that it really wants somebody to build into a full SoC for them. Intel would like to take that IP, embed it into an X86 SoC for them, and then fab the chip for them. That way, Intel is leveraging both its design and manufacturing prowess while at the same time, even in markets that it wouldn't necessarily play in on its own.
Really, this is completely expected and frankly, I'm shocked that somebody in the public eye hasn't made a bigger deal about countering this FUD. Intel isn't going to be giving up its X86 mobile strategy to build ARM mobile chips (except maybe for Apple, but I still doubt it), and Intel's long term goal still really is "X86 everywhere".