Once again, Intel (NASDAQ:INTC) has fallen significantly behind ARM Holdings (NASDAQ:ARMH) and its various licensees in terms of mobile processor performance. Today's best Atom CPU, known as Airmont, offers roughly the same performance that ARM's Cortex A15 CPU core -- which first appeared in devices in late 2012 -- does.
ARM's two most recent high-end CPUs -- the Cortex A57 and the A72 -- have essentially lapped Intel's current best Atom core. Although Intel may close the gap in future generations, I have something close to zero faith that the company will successfully be able to do so.
In my view, it is time for Intel to move away from building low-power x86 cores for much of mobile and transition to designs from ARM Holdings. Here's why.
ARM's mobile cores are better and higher performing; why "reinvent the wheel"?
ARM offers a pretty comprehensive set of CPU cores for mobile devices aimed at different performance and power levels, as you can see in this leaked slide:
ARM offers multiple tiers of CPUs, ranging from ultra-low cost designs that sip power to higher performance, higher cost designs. ARM has also made it clear that it plans to pump out new CPU cores across the various tiers each year as mobile devices tend to be refreshed at a yearly clip.
These yearly refreshes also seem to bring some very impressive performance enhancements, to boot.
Intel, on the other hand, offers just one line of Atom CPUs and the company has done a dreadful job of delivering consistent performance improvements year-after-year, and I don't think investors should take it on faith that this will change anytime soon.
For example, the company's Airmont CPU core which launched as part of the company's Cherry Trail and Braswell chips offered little to no performance improvement over the prior generation Silvermont processor.
This is, frankly, a joke. To be clear, though, the fault here doesn't lie with the engineers who designed and implemented the core, but rather the folks within Intel who decided that this level of performance would be acceptable for a core intended to launch in the second half of 2014.
I believe that Intel would be better served by developing system-on-chip designs around ARM's cores. Intel could then focus its efforts on trying to differentiate in other areas such as the modem, imaging, and more.
And, since Intel claims it has a manufacturing lead over the rest of the industry, it should be able to offer higher performance implementations of these ARM cores over its competition at lower cost.
Intel could still keep Atom around for Windows devices
Although I believe that Intel would be better served by using ARM designs in chips aimed at Android and potentially Windows 10 mobile devices, there would still be a place for Atom.
In particular, Atom could still be used as the basis for low-cost Windows PCs and convertible devices as well as in low-cost, high-core count server chips. In these markets, compatibility with the x86 instruction set is valuable and lackluster performance relative to ARM's high-end mobile designs can be forgiven in these market segments.
And, hey, if Intel ever actually gets its act together with Atom, all the better for Intel's positioning in these markets.
Will this happen? Unlikely
That being said, even though it seems to make sense to do so, I do not believe Intel will transition its mobile processors to ARM.
In that case, as an Intel investor, I would hope that the company does whatever it needs to in order to make sure that its Atom processors are consistently competitive with the best core designs that ARM has to offer.
I think that if Intel actually takes Atom seriously, it could put together some impressive designs. Sadly, given that Intel seems to want to push its big "Core" processors down in terms of power consumption, I don't think Atom will ever be given the engineering resources required for it to be a true leader in the market.