Our mobile phones are getting smarter, even as our laptops are getting dumber. Instead of packing fast processors into a notebook, PC makers are stripping them down into netbooks and other devices they can sell for less. Meanwhile, our mobile phones are looking more like mini computers with multicore processors, larger screens and improved graphics. The convergence of these devices is shifting the way chips are designed.
The brains of a mobile device need more computing abilities and are beginning to resemble the CPUs found in notebooks, while on notebooks and netbooks, chipmakers are trying to follow the cell phone model of integrating the brains and communications devices onto a single chip. Against this backdrop, Texas Instruments is walking away from its legacy as a wireless chip maker for cell phones, and is turning its attention to making the brains that will power tomorrow’s phones, mobile Internet devices and even netbooks.
Last week, I visited the campus of Texas Instruments in Dallas and spoke with Greg Delagi, head of TI’s wireless terminals business, about its bet on smarter phones. Without a baseband (or radio) focus, TI is placing most of its R&D and marketing muscle might behind its wager on application processors.
TI is betting on the trend of special purpose, smart mobile devices that follow the computer industry’s past inclinations to build a device around a CPU (or application processor). Its OMAP 3 chips do well for this, and its OMAP 4 chips, which TI admits have not yet been taped out, could be even better. However, much of the industry seems to be betting against smarter phones, in favor of dumber PCs.
Competitors such as Qualcomm are headed in the opposite direction, offering radios and processors in a single chip. Intel is even integrating its low-power Atom processor onto a chip with a radio for use in mobile devices. Samsung is also attempting to create a wireless radio — perhaps to go with it’s own application processor, which currently powers the iPhone. But TI said last year that it would sell its baseband chip business , but was unable to find a buyer. The company said it won’t invest anymnore in its radio business (though it still will provide custom-designed baseband chips). Even if it wanted to produce an integrated chip for dumber PCs, TI could soon need someone else’s silicon to do it.
With its competition bringing brains and communications together, why is TI keeping a separation between church and state (as David Carey, CEO of research firm Portelligent calls the division of radios and processors)? Delagi says it’s because in the future, phones (and netbooks and mobile Internet devices) will be more like PCs, and vendors will care more about the computing features they can offer on such devices. That means they’ll pick the best processor for the job rather than an integrated combination.
Delagi also points out that the future for chipmakers in the wireless phone is at the high end, because phones with heavily integrated silicon may sell in the hundred of millions, but they’re using low-margin integrated chips. This is true, but it ignores the fact that as mobile phones and PCs converge, PC makers are calling for integrated platforms and chips to power their smaller, cheaper devices. TI even seems to subscribe to the dumber PC model with its mobile Internet device strategy, which packages a lot of chips and software into one reference design for OEMs.
Carey, whose firm dismantles many of these gadgets, says TI’s view — that the application processor will be the defining element around which a netbook, mobile Internet device or smartphone is built — follows the historical trend in the PC industry; as smartphones become more like PCs, TI may have made a good wager. But he also notes that the progress made in semiconductor manufacturing means that multiple functions can be combined on a piece of monolithic silicon that can cost less and create a smaller device — which means the dumber PC model could win out, even on smartphones.
“It’s not A vs. B at the moment, and OEMs will experiment in the coming years, and the market will figure it out in the end,” says Carey. While the market experiments, TI will just have to wait to see if its bet pays off.