SAN JOSE, Calif. — There’s no place quite like Taipei for mobile mania, the city where mopeds are the vehicle of choice. So it’s no surprise Intel’s Atom and its rivals from Nvidia, Via and others are gathering there at Computex to pop a few wheelies about new kinds of mobile product concepts.
These days, Intel loves to egg on this mobile mania with its 2W+ Atom. But make no mistake about its agenda.
The world’s biggest (and most narrowly focused) semiconductor maker is hungry for growth. Desktops have peaked, servers are humming along at a moderate pace and only notebooks are really growing at a lively pace.
So the x86 giant wants to generate a little excitement about whole new categories of products its marketing managers dream up in their spare time. These days, Intel is generating names faster than they can come up with rational definitions for them: net-tops, net-books, ultramobile PCs, mobile Internet devices.
What’s next? Nanotops? Subtopbooks?
Taipei has always been high on such visions from the smoke-and-mirrors department in Santa Clara. When I first traveled to the island nation for Computex around 1990, it was the year of the Palmtop PC, little clamshell devices with Chicklet keyboards, black-and-white LCDs and dumbed-down versions of Windows (a term potentially redundant as military intelligence).
We were so excited about Palmtop PCs. Every self-respecting ODM in Taiwan had a prototype palmtop at their booth that year. Every Computex attendee in 1990 wanted to be the first to buy one. Within a year the whole category was dead.
At best the devices slipped into your pocket with all the grace of a grapefruit. They were nearly as useful.
Scroll ahead nearly 20 years and see what little we have learned. The Taiwan industry is still as gullible--or I should say as hungry--for a new system concept that promises something better than a single-digit profit.
Ah, but that is not the mobile Internet device. Nor is it the net-top, net-book or nex-gen mobo-mumbo-jumbo. These are systems, quoth Intel, that will bring the next billion users to the Internet because they will be cheap. Certainly less than $300, probably less than $200.
After all, the only expensive item in the nano-mobo box is the Intel processor. Everything else can be a commodity, right? Such is the vision of mobile computing from Santa Clara.
Now everyone is drinking the Kool-Aid. Seeing its future in this zero billion dollar market, Nvidia has rolled out its Tegra, a smaller, lower power alternative to Atom that is just as potent—and just as expensive. Via has a Nano CPU that was awarded a Best of Computex prize as the CPU from the Taiwanese homeland. Even normally sober Broadcom came to Taipei talking nonsense about media codecs for MIDs and UMPCs.
The problem with many of these devices is not that they have not had a powerful enough CPU or video decoder. The problem is there are no good display and input technologies that can be easily tucked into a pocket then rolled out to let human eyes and fingers do real work or have fun.
Engineers sometimes forget they have no power to redesign pockets, fingers or eyes. But sometimes when they are swept up in gadget lust they can forget these truths.
I’ll make one exception. It’s possible some net-tops and net-books may actually be new versions of entry-level desktop and notebook PCs.
Craig Mathias, one of my favorite wireless analysts, raves over his Asus eePC. Mathias says he can’t wait to get his (admittedly tiny) fingers on an MSI Wind because it is small and inexpensive like the eePC, but it has a bigger hard drive (80 Gbyte!) and display (10 inches!!). Now there’s a back-to-the-future moment for you.
Such systems are not new product concepts but stepwise extensions of old ones. They will not open up new markets but create new niches in existing ones.
The future of the mobile market lies in the smart phone. Apple has shown with the iPhone how to create a useful and just-about pocketable device for Web access and telephony.
The first time I saw Andy Bechtolsheim carrying one at a conference he was so excited about it he nearly jumped out of his signature Silicon Valley sandals. "Finally, someone has found a way to put the Internet experience in your pocket," he told me. And as usual, Andy was right.
The iPhone doesn’t need an x86 chip, much as that must frustrate Paul Otellini. According to the latest trends my colleagues at Portelligent have seen in their teardowns, it may not even need an applications processor in the near future.
A simple cellular baseband with an extra ARM core or two will probably do quite nicely for these systems. Nothing fancy. Last-generation hardware is just fine.
But these mobile systems of the future will need a lot of really creative software to make use of new input technologies like multi-touch displays. Software. That’s something Intel and Taiwan generally put at the end of the product-creation cycle as icing on the hardware cake.
That’s why the mobile future is coming not from Santa Clara but from Cupertino. And even 18-plus months after this future was shown to all the world, Intel and Taiwan Inc. have still not quite figured out how to replicate any piece of it except the mobile mania.
By the way, the Atom chip is a significant technical achievement for Intel. Running the x86 at up to 2 GHz at 2W is a genuine milestone, albeit a milestone the latest 1.6 GHz, 2.5 W chips have not quite hit yet.
But this is a laudable chip for which there are few ready sockets. It has too little performance for mainstream notebooks and sucks batteries way too fast to complete in the smartphone space where average application processors run at about 500 milliW.
So you will not find Atom in the next Dell notebook or RIM Blackberry--let alone the next iPhone, the socket for which the chip is ultimately aimed.
I suspect Intel execs knew that. I think they signed off on a business plan that would commit some of their precious 45nm capacity to Atom in hopes that it would pave the way for a 32nm version that might give ARM a run for its money in smartphones in say 2010 or so.
So in the meantime while Intel in investing in this long road to the cellular pot of gold it wants to make as much hay as it can. Thus it has set its marketing engines on revving up the mania for new kinds of products that don’t exist.
Interestingly, they are coupling these pie-in-the-sky platforms with their passion for WiMax. Talk about impractical. Not only will these new mobo-prototypes be too big and underpowered, they will link to a network which you will not be able to find anywhere outside western Senegal.
The rest of the wireless world—including Qualcomm’s Paul Jacobs—has said LTE will probably be the next big mainstream wireless wide area net. So I will not be buying a WiMax-ready, Atom-based MID, nanonet thingie or subtopbook anytime soon. I’ll wait for an LTE iPhone. The big question for us tech watchers is—will it sport an ARM12 or an Atom3?
