With a decent wind behind it, Intel can catch ARM

Intel claims that missing the boat that sailed loaded with all-conquering mobile phones and tablets on board won't leave it scratching about with nothing to flog, as the connected world gives up on the desktop and goes handheld instead.

OK, the boat may already be afloat and being steered by captain ARM, but the chip giant has been left standing on the dock of the bay before, waving a tear-spattered hanky at apparently-departing rivals. And it has a closet stuffed with outboard motors and spare sails and reckons, given a decent enough following wind, it can soon catch up.

At least, that's what we took away from a lively briefing in London yesterday in the company of Bill Leszinske, general manager of Intel's Atom SoC technology development group, who is tasked with convincing sceptics that Intel's late start in the ARM race doesn't mean it's left dead in the water.

It was the admiral of Intel's fleet, CEO Paul Otellini, who initially took responsibility for not realising that smartphones were to become such important computing platforms. Ruing missed opportunities to start work on low-power mobile chips earlier, Otellini said in October last year: “I wish I had been smart enough to start it seven years ago because we’d be in a good position today, but I wasn’t.”

While Otellini - unlike a certain R. Murdoch - may be applauded for saying the buck stops with him, it was the boss of Intel's Ultra Mobility Group, Anand Chandrasekher, who was forced to fall on his cutlass, as the chip giant failed to gain traction in the 'ultra-mobility' space while the ARM destroyer sailed into markets Otellini thought would never amount to much.

"This is not a reaction to ARM," insisted Leszinske yesterday, detailing Intel's System-on-Chip (SoC) plans for the upcoming half-decade. He managed to spit out that statement in the face of a rather large pile of evidence to the contrary.

Still, according to Leszinsk, the chip maker now has all its ducks in a line and with them a strategy that will put Intel Atom-branded chips inside future tablets and in smartphones, where otherwise ARM chips would likely reside - even if this isn't really about ARM - which, of course, it is.

The lined-up ducks include an on-chip interconnect, crucial to the 'system' tacked around the chip, and a growing 'IP' library, along with an accelerated roadmap that promises die shrinkage at twice the rate of Intel's conventional 'tick-tock' desktop chip-making process.

In the realm of handheld gadgetry, "you have to come to market more quickly and add more features," Leszinsk said, outlining how Intel plans to transition from 32 to 14nm within three years, with a stop at 22nm along the way. The code names, for those who love to drop such things into dinner party conversations, are: Saltwell, Silvermont and Airmont.

Intel, quite naturally, still insists that its advantage is its x86 architecture, even if the WinTel axis has recently sprung a leak with Microsoft planning a Windows 8 dalliance with ARM.

We haven't heard the 'Internet runs on x86' argument for a while now, but Leszinske did mutter something about the challenges developers will face knocking up apps for Windows on ARM. He also muttered something about MeeGo, but in a rather half-hearted manner: "Maybe it will be successful, maybe it won't," he said, in a manner suggesting the latter possibility is more likely. For some reason, we suddenly thought of another meaning of the past participle 'Googled.'

Rather clumsily we wanted to know how you go about changing fabs that have been churning out desktop chips by the million to a process delivering phone chips by the billion, and were assured it's not as complicated as one might imagine.

Let's hope so - for Intel's sake. It would be a shame if such an aggressive and rapacious monopolist were to keep on banging out ever faster and more powerful desktop chips that it finds people neither want nor need any more, while all its minnowy rivals mop up with their ARM-based designs, slotting them into phones and tablets and tellies and stringing them together to power data centres.

Cap'n Otellini would then have trouble denying the charge that he just hadn't been paranoid enough.