Optical/IP Networks

Intel's CTO Has a Smashing Time

MUNICH – European Intel Developer Forum 2002 –


Intel's CTO Pat Gelsinger triumphantly holds up the battered remains of a cellphone to the camera. Then, from his podium, he exhorts the delegates watching to follow his lead and heedlessly lay waste to theirs. Intel Corp., Gelsinger smoothly assures them, will pick up the tab.

What's going on here? Has one of the industry's über-suits gone native and joined the cult of Phone Bashing, which, er, probably isn't coming to a high street near you soon.

No, Gelsinger has a point to make -- not content with being king of the desktop, Intel wants to get inside your Nokias as well. "We will move all radio functions into silicon," Gelsinger told his audience.

Intel is already a major supplier of flash memory for phones and other handheld devices, and its new generation of StrongARM embedded processors are also winning converts in the wireless industry. But the point that Gelsinger was trying to make with his Who-style antics is that more than half of the components in a mobile phone today -- capacitors, inductors, and antennas -- are analog and are made out of materials such as metals and ceramics. What's more, these bits also make up more than half the cost of a handset and consume half the electrical power, he says.

Intel plans to make all these components out of silicon and shrink them down onto a single chip -- the theory being that phone circuits made using silicon chips rather than analog components would be cheaper, lighter, and more energy efficient. Gelsinger then showed a wafer containing hundreds of miniature antennas fabricated out of MEMS (micro-electro-mechanical system) silicon. The audience was silent as he held up this harbinger of a new wave in communications, as no one dared confess that the components were so small they couldn't actually see them.

There are already startups making single-chip radio systems using the industry standard (and comparatively cheap) complementary metal-oxide semiconductor (CMOS) production techniques. Bluetooth booster Cambridge Silicon Radio (CSR) has had just such a chip out on the market for quite a while now. Intel invested in CSR in 2000.

However, Intel's plans are just a tad more ambitious. The next step, says Gelsinger, would be to enable "intelligent roaming" -- allowing devices to move seamlessly from personal area networks (Bluetooth), to 802.11 hotspots, to a wide area network, whether it be GSM, GPRS, 3G or something else, without dropping the connection. The user would be completely unaware of the handover from one network to the next.

This internetwork roaming capability is something that companies like Hewlett-Packard Co. and Nokia Corp. are also looking at.

Intel's solution is to carry, not one, but three radios inside an appliance. And that, Gelsinger says, makes the problem of shrinking radios three times as challenging. Interesting to note, Intel has also been active in the movement towards software radios (see WLAN = Windows Wireless Networking?), which would do away with separate radios but require a wide-bandwith antenna and soft- or firm-ware to switch between frequencies and support the different modulation schemes that WLAN and WAN systems use.

Eventually, says Gelsinger, we will get to the point where not only has the radio been shrunk down to a single chip, but the radio becomes only a small part of the chipset. Then, every chipset can have a radio on it.

Naturally, it's Gelsinger's job as CTO to come up with waffle like this about the future. The thing about Intel, however, is that it's been right a lot of times in the past, starting with Gordon Moore, who is still an emeritus chairman at the company.

"I was holding in my hand a circuit with 30 transistors," says Moore in a video clip. "And I knew that in the lab we had a circuit with 60 transistors. So I predicted that the number of transistors in a circuit would keep on doubling every year, all the way to 60,000 transistors. And that prediction turned out to be more accurate than I ever would have imagined."

At the time Moore made that prediction, Gelsinger observes, Intel didn't have a clue where it would find the technology for making transistors ever smaller. Today, on the other hand, the company claims to have already invented the basics of the technologies that will carry it into the future. "Our vision today is clearer than it has been at any time during the past two decades," he contends.

Incidently, we didn't see anyone follow Gelsinger and smash up their mobile. Clearly, no one was particulary confident that they'd ever see any cash back from the chipmaker.

Special to Unstrung by Pauline Rigby, Senior Editor, Light Reading

Sign In