Intel needs more than just an ARM license

ARM's boss thinks Intel needs to 'stop messing about' and get an ARM license, but it might take more than that.

This industry is full of executives expressing what sounds like pure self-interest, but their comments shouldn't be dismissed so quickly.

Case in point: ARM Holdings' founding CEO Sir Robin Saxby told the GSA Entrepreneurship Conference in Britain that "As an ARM shareholder, I recommend Intel to take an ARM license and stop messing about."

"Well of course you would, Sir Saxby," is the reaction most of us would have.

But his comments have an unfortunate ring of truth to them. Think of where Intel was when Paul Otellini took over in 2005 and where it is now. AMD was clobbering Intel and grabbing market share at an incredible rate. He refocused the company, introduced the tick/tock model, and now AMD is almost a non-entity in semiconductors.

But Intel is getting its tail handed to it in the mobile space. The Atom processor is going nowhere. So Paul's reward has been an elbow to the ribs, pushing him out before he likely wanted to go or had to go (Intel CEOs face mandatory retirement at 65; Otellini is 62). He also has no heir waiting in the wings. Intel has always had a very orderly transition to new CEOs, with the successor picked years in advance. That plan was interrupted when Sean Maloney, the heir-apparent, suffered a severe stroke in 2009. He made a remarkable recovery, but that marked the end of him ever getting the top spot.

Despite Intel's success under his stewardship, all the emphasis is on tablets and smartphones, and that market is utterly dominated by ARM. Intel has been pushing the Atom processor and promising design wins for as long as I can remember, and it still has jack squat to show for it.

Intel got back in the game against AMD by throwing its tremendous resources at engineering. It always has. During every downturn in the economy, Intel's response has been to increase R&D, and it always came out of the downturn in better shape.

Not this time. Atoms are going nowhere, even though the chip has improved greatly since its introduction. The fact is it just can't compete with ARM's ubiquity and support, even though it has x86 compatibility between Core and Atom on its side.

It's hard to make the power efficiency argument. AnandTech has pointed out that Clover Trail is more than competitive with ARM. Of course, it has weaknesses, like the GPU. Qualcomm and especially Nvidia's GPU technology just flatten Intel's.

The real problem with doing ARM now is that Intel would have to reset its efforts. It's gone down the Atom road for years and has a lot of investment in it. Acquiring an ARM license and then doing something special with it would take years, even with Intel's talent. With each passing day, Qualcomm, Samsung, Apple and Nvidia would only pull further away.

Some people point to XScale, Intel's old ARM product, as proof it couldn't make ARM work. Realize, it's not going to be Intel that makes Atom successful, nor will Intel be an automatic success with ARM if it just licenses the technology.

ARM was out there, but it wasn't exactly setting the world on fire until 2007, when Apple introduced the iPhone. Then suddenly, everyone had to have an ARM processor. Within a year, Snapdragon and Tegra were on the market. So was Atom.

What Intel needs is a design win. A big, game-changing one like the iPhone was. That, not an ARM license, will turn things around. Why there is so much reluctance on the part of OEMs is a mystery to me. It could be they simply don't want to let the Intel monster take over smartphones and tablets like it did with PCs.

For whatever reason, Intel has put the engineering into Atom and will continue to do so. It needs to crack the OEMs and get that must-have win. Then you watch how fast Atom licensees line up.

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.
Must read: 10 new UI features coming to Windows 10