Intel’s Promise to Reinvent the PC Falls Flat
Reinventing the PC was the promise that the world’s largest chipmaker made from the stage at its annual Intel Developer’s Forum in San Francisco on Monday. And if ever there was a moment when Intel needed to generate some excitement, this was it. Last week, Intel slashed its sales forecast for the September quarter, as demand for PCs lagged around the world.
In remarks before an audience made up primarily of product engineers and software developers, Intel’s chief product officer, Dadi Perlmutter, went through a batch of almost-interesting new capabilities for the personal computer, and showed off a handful of designs for new mobile PCs, aimed at dislodging that iPad or smartphone in your hand with something small and light that contains an Intel chip.
There were PCs that were small, PCs that were light, PCs with displays that detached from their bodies, and PCs that folded and twisted over the keyboard. All of them look like reheated concepts from the Tablet PC era circa 2002, their sole physical advantage being that they’re thinner and lighter than before, and touch-ready. Like an iPad.
Then came demos of two features around the personal computing experience from Nuance, maker of the Dragon line of voice control applications — one showing how soon you’ll be able to search Google, tweet and play digital media files with voice commands, a la Apple’s Siri, and the other, 3-D gestures a la Microsoft’s Kinect system for the Xbox. If this was Intel’s idea of generating excitement about the future of the PC, it didn’t work.
And what about Moore’s law? Certainly, the trend in computing power — enabled by the ability to shrink and thus double the number of transistors on a chip every two years or so — first observed 47 years ago by the Intel co-founder would enable some super-awesome new feature set that will make the PCs of 2013 seem quaint and silly and less awesome than PCs from before?
Not so much. Here’s a startling-when-you-think about it fact about Haswell, the code name for the new chip Intel plans to release next year. Moore’s law essentially allows you to make one of two fundamental choices as the transistor sizes on individual chips shrink with each successive generation. You can either build a chip that gets twice as much work done using about the same amount of power as the previous generation. Or you can build a chip that gets the same amount of work done using half the amount of power. Which choice do you think Intel made with Haswell? Same work, less power.
Of course, there are a lot more finely grained details about what Haswell will do (Anand Lal Shimpi has a detailed rundown on its many technical details here). Since mobile devices of every flavor and form factor imaginable are in, concerns over battery life have overtaken raw power on the list of priorities for Intel’s customers, the PC makers. For years, the need was always for more power, and Intel was ready to oblige. Now, computing power is so plentiful on a PC that most routine applications don’t even come close to taking advantage of it all.
And I wasn’t the only one who came away with this impression. At an informal gaggle of chip industry analysts after Perlmutter’s speech, the consensus was that the biggest benefit of Haswell is the improved power performance.
Patrick Moorhead, a former executive with Intel’s main rival, AMD, and now an industry analyst, put it to me in starker terms: “Intel has a week to show that the PC has a bright future over the next five to 10 years,” he said. “It’s all about usage models that can be done on a phone platform with wireless display and peripherals versus a PC platform.”
I don’t know that we’re quite that far along; I still rely on my home PC (okay, it’s a Mac), enhanced by a few external hard drives for backup purposes, to be the central storage and retrieval spot for all my digital stuff — my photos, my music, my videos, and many, many documents. And when I want to get some real work done, I turn first to that machine. The iPad I use to catch up on episodes of “Breaking Bad” and to read The New Yorker. I have a hard time imagining a world were PCs and tablets don’t coexist in most households in a long-term digital symbiosis, but it’s clear that those PCs will be replaced less frequently and will become more marginalized as consumer attention shifts to tablets and smartphones, devices where, at least to now, Intel isn’t strong.
PCs as we know them may in time go out of style, but there’s one thing that Intel can count on: It will still be called upon to build the best and most powerful chips it can for the millions of servers that will be needed to power the cloud-based services all those smartphones and tablets will be using. Battery life doesn’t matter in servers, though the need to use less power in data centers, and thus do more while operating at a lower cost, is always in style. The data center is at least one place where the one thing that Intel does better than anyone else — deliver ever more computing muscle for lower cost on an almost annual basis — will never go out of style.